Invasive cardiology · novice to mastery

Cardiac catheterization, explained and simulated.

A simulation-based curriculum that takes you from what a catheter is to performing a full case yourself — foundations, hemodynamics, angiography, and intervention, each built on the one before.

Live · Radial artery

Retrograde from the right radial — the systolic reads higher here than in the aorta.

Your path

Six stages, novice to mastery

Work down the path in order — each stage builds the skills the next one assumes. Every module also stands alone, so you can jump straight to any of them from the sidebar. The path ends with you running a full catheterization yourself.

1

What a catheterization actually is, the room and the equipment, the normal numbers, and the cardiac cycle in motion — the plain-language groundwork before anything interactive.
2

Pressures and waveforms chamber by chamber, then the calculations that turn them into decisions — Fick, Gorlin, shunts, resistances — with the classic artifacts and pitfalls built in.
3

Coronary anatomy, the angiographic projections and how they flatten a three-dimensional tree, stenosis and physiology (FFR/iFR), and reading the artery from the inside with IVUS and OCT.
4

Access, anticoagulation and contrast, and the recognition patterns for the complications that matter — then the structural decisions the modern lab makes: TAVR, TEER, and the adult congenital lesions.
5

Now make the calls. Eighteen branching cases where you acquire the data, catch the artifact, do the math, and live with the consequences. Start with one of these:
6

Stage 6 · The capstone
Perform the catheterization yourself

Everything you have learned, in one place. Choose femoral or radial access, advance the catheter across the valve, and watch a live pressure tracing build under your hands — then measure every number off the waveform you just recorded. The closest thing to standing at the table.

Femoral or radialLive pressure tracingNumbers off the waveform

How to use it

Follow the path, or explore in any order.

The path is the recommended route from novice to mastery, but every module stands on its own — jump to whatever you need from the sidebar. See the full curriculum map → for everything the platform covers.

Anatomy of a case

A diagnostic left-heart case, end to end. Six stages, each with the decision that is actually made there.

The case where the catheter is

The step what happens & the call

Predict, then reveal

An intermediate mid-LAD lesion looks about 50% and the patient has typical exertional angina. How do you decide whether it is the culprit?

The two catheterisations, side by side

Left heartarterial · against the flow

Right heartvenous · with the flow

Door

Radial at the wrist by default. Femoral for large-bore work or a failed radial.

Femoral vein, right internal jugular or basilic vein.

Course

Up the aorta to the coronary ostia and retrograde across the aortic valve into the LV.

RA to RV to PA, then a balloon floated out until it wedges.

Catheter

Preformed curves that are torqued into position — Judkins left and right, pigtail for the ventriculogram.

Balloon-tipped and flow-directed. The circulation does the steering.

What it answers

Epicardial coronary anatomy, LV pressure and any gradient across the aortic valve.

Filling pressures, pulmonary pressures, cardiac output and shunt.

Left-sided filling pressure

Measured directly in the LV.

Read at the wedge without crossing a valve.

Complications of the door

Retroperitoneal bleed from a high femoral stick, radial occlusion, pseudoaneurysm.

Carotid puncture and a neck haematoma. Venous bleeding is easier to control — and under ultrasound pneumothorax is a subclavian problem, not a jugular one.

The door is chosen separately on each side. Collapsing both into one radial-or-femoral habit quietly writes the jugular out of the procedures it usually owns — endomyocardial biopsy, transplant surveillance and any PA catheter that is not coming out at the end of the case.

Checkpoint · the key calls

Three moments from the case above.

Vascular access — radial and femoral

Access is the first irreversible decision of the case. Where the needle enters decides whether a bleed can be compressed, and whether anyone sees it in time.

The access site right wrist · click to place the needle

Result what happens after the stick

Three ways this goes

Too high — above the inguinal ligament

bleeds into the retroperitoneum, where no hand can compress it

Common femoral — over the femoral head

compressible against bone. Wire in, sheath in.

Too low — past the bifurcation

pseudoaneurysm, or an AV fistula if the needle carries on into the vein

The skin crease is not the inguinal ligament. In a large patient it sags well below it — which is how “above the crease” becomes above the ligament.

Click the artery to place a needle. Turn on fluoroscopy to find the femoral head.

Scene Overlay Or place it here

Why the distal radius is the landmark

Reference — the radial case, drug by drug

Before the needle — what is worth doing, and what is not

Step	What the evidence says
Collateral testing to choose the wrist	Not required. In RADAR, thumb capillary lactate, grip strength and symptoms were the same whether the Allen result was normal, intermediate or abnormal, and no patient in any group developed hand ischaemia. SCAI advises against using collateral testing to select an access site.
Sedation and local anaesthetic	The cheapest spasm prophylaxis there is. Spasm is provoked by pain, by repeated punctures and by periosteal contact, so the things that reduce all three come before any drug.
Ultrasound guidance	RAUST, 698 patients randomised: first-pass success 64.8 against 43.9 per cent, forward attempts 1.65 against 3.05, and needing five attempts or more 2.4 against 18.6 per cent. No difference in spasm, pain or bleeding — the gain is in getting in cleanly the first time.

The spasmolytic — through the sheath, once it is in

Agent	Dose	Spasm rate	What it costs
Nothing	—	22 per cent	The placebo arm of SPASM, 1,219 patients randomised across five regimens. This is the rate you accept by giving nothing
Molsidomine	1 mg	13 per cent	A nitric oxide donor. Clearly better than nothing, clearly worse than verapamil — and not marketed in the United States
Verapamil	2.5 mg	8 per cent	Two thirds of the spasm gone. It stings going in and drops the mean pressure a little, which matters in a bradycardic or hypotensive patient
Verapamil	5 mg	8 per cent	No better than 2.5 mg in the same trial — 7.9 against 8.3 per cent. Doubling it buys more burn and more hypotension, not less spasm
Verapamil + molsidomine	2.5 mg + 1 mg	5 per cent	The winning arm. A calcium blocker and a nitric oxide donor work on different parts of the same wall; where molsidomine is unavailable, nitroglycerin is the stand-in

The nitroglycerin cocktail is the practical version of that last row. In a separate randomised trial, heparin alone left spasm at 20.4 per cent; heparin plus 100 µg of nitroglycerin took it to 4.4 per cent.

Adding 1.25 mg of verapamil on top changed nothing (3.8 per cent, p = 0.80) — so if you give nitroglycerin, the verapamil is optional.

Route matters more than most trays admit: pooled across 11 trials, subcutaneous nitroglycerin over the artery cut spasm (RR 0.57) while intra-arterial and topical nitroglycerin did not reach significance on their own.

The American Heart Association accepts verapamil, diltiazem, nicardipine or nitroglycerin.

Beyond the doses above the choice between them is a choice about tolerability, not efficacy: a verapamil-and-nitroglycerin cocktail given before the pull cut painful withdrawals from 34 to 14 per cent and significant pullback resistance from 22 to 8 per cent.

Heparin — what keeps the artery you are leaving behind

Give it before the first catheter, not at the end. Heparin is dosed in international units of anticoagulant activity rather than milligrams, because it is a mixture of chain lengths standardised by bioassay rather than a single molecule; 5,000 units is the usual adult dose for a diagnostic case.

Where the artery matters, dose for weight: randomised against half the dose, 100 units per kilogram cut radial artery occlusion from 8.1 per cent to 3.0 per cent, an odds ratio of 0.35 and about twenty patients treated to spare one artery, with no increase in bleeding and no longer time to hemostasis.

Note what that trial compared: 100 against 50 units per kilogram, not heparin against nothing. The question is how much, never whether.

Haemostasis — the part of the case that decides whether the artery survives

Patent haemostasis is the lowest band pressure that stops the bleeding while blood is still flowing through the artery underneath. In PROPHET it cut occlusion by 59 per cent at 24 hours and 75 per cent at 30 days against conventional occlusive pressure.

Nothing about the procedure predicted which arteries closed; only low body weight did. Occlusion ran at five to eight per cent historically and sits under one per cent in series that do all three of these things.

Reference — choosing the door, and the two ways each one is missed

Which door — decided before the patient is on the table

Route	Reach for it when	Against it
Radial	The default for diagnostic work and most PCI. Randomised evidence establishes lower access-site bleeding and far fewer vascular complications. Mortality is less clean — RIVAL was neutral and SAFARI-STEMI negative — but pooled individual-patient data across 21,600 patients show lower 30-day mortality in acute coronary syndromes, only partly explained by the bleeding difference	Small or heavily calcified vessel; known radial occlusion; a forearm being preserved for dialysis access; large-bore requirements the sheath cannot meet
Distal radial (snuffbox)	Preserves the forearm radial for future grafting or access; haemostasis is quicker	Smaller vessel, two to three times the crossover rate, more spasm — and against best-practice patent haemostasis it did not lower occlusion. A deliberate technique, not a rescue from a failed wrist stick
Femoral	Large-bore work — mechanical circulatory support, structural intervention; failed or unsuitable radial	Higher access-site bleeding, bed rest and the one bleed you cannot see or compress

Femoral — one target, two ways to miss it

Stick	Vessel entered	Why it goes wrong	How it presents
Too high	External iliac, above the inferior epigastric	The puncture sits above the inguinal ligament, so a bleed tracks into the retroperitoneum, where no manual pressure can reach it	Late and quiet — hypotension, flank, back or abdominal pain, haematocrit down by around ten points, often no visible groin haematoma. Watch the rate: bradycardia is commoner than tachycardia
On target	Common femoral, over the middle third of the femoral head — the bifurcation lies at or below the centre of the femoral head in 98.5 per cent of people, and yet 13 per cent of punctures — about one in eight — still land in a vessel other than the common femoral	Below the ligament so a bleed is compressible, above the bifurcation so it is still one large vessel lying on bone	—
Too low	Superficial femoral or profunda	Smaller vessel sitting off the femoral head, with nothing behind it to compress against	Pseudoaneurysm — pulsatile sac, to-and-fro bruit. Or an AV fistula with a continuous murmur if the needle crosses into the vein

Radial artery occlusion is the commonest late complication of the wrist, and it is silent: the dual palmar supply keeps the hand well, and the artery is simply gone for any future case, graft or fistula.

Heparin, a sheath matched to the artery and patent haemostasis are what prevent it — the numbers are in the radial case, drug by drug above.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Barbeau GR, Arsenault F, Dugas L, et al. Evaluation of the ulnopalmar arterial arches with pulse oximetry and plethysmography: comparison with the Allen’s test in 1010 patients	Am Heart J 2004;147(3):489-93	DOI
Valgimigli M, Campo G, Penzo C, et al. Transradial coronary catheterization and intervention across the whole spectrum of Allen test results (RADAR)	J Am Coll Cardiol 2014;63(18):1833-41	DOI
Shroff AR, Gulati R, Drachman DE, et al. SCAI expert consensus statement update on best practices for transradial angiography and intervention	Catheter Cardiovasc Interv 2020;95(2):245-52	DOI
Mason PJ, Shah B, Tamis-Holland JE, et al. An update on radial artery access and best practices for transradial coronary angiography and intervention in acute coronary syndrome (AHA scientific statement)	Circ Cardiovasc Interv 2018;11(9):e000035	DOI
Seto AH, Roberts JS, Abu-Fadel MS, et al. Real-time ultrasound guidance facilitates transradial access (RAUST)	JACC Cardiovasc Interv 2015;8(2):283-91	DOI
Varenne O, Jégou A, Cohen R, et al. Prevention of arterial spasm during percutaneous coronary interventions through radial artery (SPASM)	Catheter Cardiovasc Interv 2006;68(2):231-5	DOI
Chen CW, Lin CL, Lin TK, Lin CD. A simple and effective regimen for prevention of radial artery spasm during coronary catheterization	Cardiology 2006;105(1):43-7	DOI
Kiemeneij F, Vajifdar BU, Eccleshall SC, et al. Evaluation of a spasmolytic cocktail to prevent radial artery spasm during coronary procedures	Catheter Cardiovasc Interv 2003;58(3):281-4	DOI
Abdelazeem B, Abuelazm MT, Swed S, et al. The efficacy of nitroglycerin to prevent radial artery spasm and occlusion during and after transradial catheterization: a systematic review and meta-analysis of randomized controlled trials.	Clin Cardiol 2022;45(12):1171-83	DOI
Hahalis GN, Leopoulou M, Tsigkas G, et al. Multicenter randomized evaluation of high versus standard heparin dose on incident radial arterial occlusion after transradial coronary angiography: the SPIRIT OF ARTEMIS study.	JACC Cardiovasc Interv 2018;11(22):2241-50	DOI
Pancholy S, Coppola J, Patel T, Roke-Thomas M. Prevention of radial artery occlusion — patent haemostasis evaluation trial (PROPHET)	Catheter Cardiovasc Interv 2008;72(3):335-40	DOI
Aminian A, Sgueglia GA, Wiemer M, et al. Distal versus conventional radial access for coronary angiography and intervention (DISCO RADIAL)	JACC Cardiovasc Interv 2022;15(12):1191-201	DOI
Jolly SS, Yusuf S, Cairns J, et al. Radial versus femoral access for coronary angiography and intervention in acute coronary syndromes (RIVAL)	Lancet 2011;377(9775):1409-20	DOI
Le May M, Wells G, So D, et al. Safety and efficacy of femoral access versus radial access in ST-elevation myocardial infarction (SAFARI-STEMI)	JAMA Cardiol 2020;5(2):126-34	DOI
Gargiulo G, Giacoppo D, Jolly SS, et al. Effects on mortality and major bleeding of radial versus femoral artery access for coronary angiography or percutaneous coronary intervention: meta-analysis of individual patient data from 7 multicenter randomized clinical trials.	Circulation 2022;146(18):1329-43	DOI
Schnyder G, Sawhney N, Whisenant B, et al. Common femoral artery anatomy is influenced by demographics and comorbidity	Catheter Cardiovasc Interv 2001;53(3):289-95	DOI

Reverse Barbeau — before they go home, with the ulnar compressed

Same oximeter, same thumb, the other artery. Compress the ulnar and whatever the thumb still shows is arriving up the radial, so the trace now reports the vessel you instrumented. Occlusion is silent and you cannot palpate for it: an absent radial pulse agrees poorly with what the duplex shows.

Against duplex in 503 patients this is a sensitivity of 96 and a specificity of 99.8 per cent for occlusion; thumb and index finger perform the same. Run it before the case with the radial compressed and it is the original Barbeau, which grades collateral supply and should not be used to choose the wrist.

Checkpoint · name the complication

Each needle is placed as described. What follows?

Zeroing, levelling & the phlebostatic axis

A transducer reports pressure relative to its own height. Put it at the wrong level and every chamber pressure is off by the fluid column between them — while the waveform still looks perfect.

The setup transducer level with the phlebostatic axis

Right atrial tracing a · c · v — shape is unchangedy: pressure as the transducer reports it, mmHg · x: time, seconds

Transducer height level with axis

Predict, then reveal

You mount the transducer 10 cm above the phlebostatic axis and record. The right-atrial pressure will read…

What the transducer reports

8

RA mean

true 8

11

PCWP

true 11

17

PA mean

true 17

Level — every pressure is true.

The physics, and why it ruins low pressures first

A transducer is zeroed to atmosphere at whatever height it happens to be clamped, and from then on reports pressure referenced to that height. The fluid column between it and the chamber is the error: sitting high, the column pulls the reading down; sitting low, it pushes the reading up.

1 cmH₂O ≈ 0.74 mmHg — so 10 cm too high costs about 7 mmHg off every number.

The reference is the phlebostatic axis: 4th intercostal space at the mid-anteroposterior chest, roughly atrial level supine.

Zeroing and levelling are two different acts

	Zeroing	Leveling
What you do	Open the transducer to atmosphere and call that pressure zero	Place the already-zeroed transducer at the phlebostatic axis
What it cancels	The offset of the atmosphere and the tubing	The hydrostatic column between transducer and chamber
Doing it alone	Does not accomplish the other. A perfectly zeroed transducer clamped 10 cm high reads 7 mmHg low all case, and the tracing looks entirely normal. Re-level any time the table height changes.

The error is absolute, not proportional — the same 7 mmHg comes off every number equally

Pressure	Normal	True	Reads	% off	What you conclude
Aortic systolic	90–140	120	113	6%	Noise at this scale — 113 is still inside the normal band
RA mean	2–6	9	2	78%	Congestion reads as a normal filling pressure
PCWP	4–12	18	11	39%	Congestion reads as normal — necessary diuresis is withheld

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Kovacs G, Avian A, Olschewski A, Olschewski H. Zero reference level for right heart catheterisation	Eur Respir J 2013;42(6):1586-94	DOI

Checkpoint · correct the reading

The monitor shows a number and the transducer is off level. Give the true right-atrial pressure (0.74 mmHg per cm).

The manifold, the fluid column & damping

Every number you read arrives through a column of fluid, and that column distorts the waveform in a few characteristic ways. Put something in the line, then prove what it is with a fast flush.

The line clean — saline column to the transducer

Under-damped — ζ below 0.45 — rings, and over-reads the systolic peak. Over-damped — ζ above 0.8 — blunts it, and under-reads the peak. The mean survives both.

Pressure radial line · mmHgy: pressure as the system reports it, mmHg · x: time, seconds

What’s in the line

What the tracing reads — against the pressure that is actually in the artery

In the artery

On the monitor

Error

Systolic

—

Diastolic

—

Mean

—

Fast flush

1–2 crisp oscillations

press the button

—

The physics, and the two ways a line lies

The line is a column of saline with a diaphragm at the end of it: a weight on a spring, with friction. Two numbers describe how it behaves, and the panel above shows both live.

Natural frequency, f_n, in hertz — how fast the column can move. A pulse at 72 repeats at only 1.2 Hz, but the sharp corners in the trace are made of much faster components, so the column has to stay honest to about 12 Hz, and to 25 in a tachycardic patient. Under roughly 7 Hz the trace cannot be rescued by anything.

Damping, ζ, a ratio with no units — how quickly a disturbance dies away, expressed as a fraction of the damping that would stop it dead with no overshoot at all. Zero rings forever, one does not overshoot, and 0.65 is the compromise you want.

Neither number is something you measure at the bedside. You infer both from a fast flush, and the table is the whole of it.

	Under-damped (rings)	Optimal	Over-damped (blunted)
Cause	An air bubble, or long compliant tubing — added compliance, a soft spring	A clean, stiff, bubble-free saline column	Contrast, clot, kink, a narrow catheter — and, at the tip, spasm or an occluded ostium
Fast-flush shows	3 or more oscillations after release	1–2 crisp oscillations	None — a slurred, sluggish return
The trace	Overshoots the true peak, then rings	Sharp upstroke, clear dicrotic notch	Slurred upstroke, notch gone, peak clipped, pulse pressure narrow
The numbers	Systolic too high, diastolic too low	True	Systolic too low, diastolic pulled toward the mean
What to do	Find and clear the bubble — stopcocks, connections, flush bag	Trust the trace	Flush it. If it clears, it was contrast. If it stays damped, suspect clot — aspirate, do not power-flush

Two things to carry out of this. The mean survives either error, so a damped arterial line can still guide a pressor when its systolic cannot be trusted.

And a bubble makes the systolic read high, not low — air is compressible, so it softens the spring, f_n collapses and the column resonates with the pulse instead of reporting it.

Drop a bubble into the line above and watch f_n fall from 24 Hz to 11 — straight out of the usable band and into resonance with the pulse.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Gardner RM. Direct blood pressure measurement — dynamic response requirements	Anesthesiology 1981;54(3):227-36	DOI

Checkpoint · read three unlabelled tracings

Each strip is a fast flush on a different line. Classify each one.

Catheters, curves & torque

Diagnostic catheters come preformed, so choosing one is mostly a matter of matching a shape to an aorta. The route and the target pick the family. The width of the ascending aorta picks the number. Rotation only decides where a correctly chosen curve ends up pointing.

The instrument

Two curves and the gap between them

A braided shaft with a soft tip and two set bends. The secondary curve leans on the far aortic wall and is the only backup you have; the primary curve decides the angle at which the tip meets the ostium. The number stamped on the packet is the distance between them in centimetres — so it is a measurement of the aorta, not of the patient.

Two ways in

One engages on advance, one on rotation

A correctly sized Judkins left arrives at the root already aimed at the left main, so it engages as you advance with the hub near neutral. A Judkins right has no far-wall limb and engages on a deliberate rotation instead, which is why a tortuous arch costs you the right coronary before it costs you the left. Its name refers to the right coronary artery, not the right heart.

The shelf the shapes, drawn to one scale — select any catheter

Detail what it is, how it engages, when you reach for it

Aortic root LAO view · engage the coronary

Tip feedback what you would see and feel

—

Torque transmission

hub 0° · tip 0°

1:1

Torque the hub to bring the tip onto the left main. Then try it with a tortuous arch.

Tip pressure 120/78

CatheterJL4 · left

Aortic root width34 mm · normal

Torque the hub (clockwise +)0°

Arch / supportsupported

Three things to prove to yourself — the controls are below

Start here. A JL4, a 34 mm root, the hub at zero and the tip already in the left main with a crisp aortic tracing. That is what a correct choice looks like, and it needs no technique at all.
Widen the aorta to 40 mm. The same catheter now falls short. Fix it by changing the curve rather than the hub.
Set the arch to tortuous and take a JR4 to the RCA. Watch the wind-up figure climb as you turn and get there without a whip.

Reach is set by the curve. Rotation only aims it. The two controls are not interchangeable.

Predict, then reveal

A JL4 sits in the root with a crisp aortic tracing, but the tip is a centimetre short of the left main and rotating the hub does not bring it any closer. Before you change anything, what has that single observation already told you?

Choosing one

Three questions, in this order. Only the last one has a number in it.

Ask	And you have chosen
Which route?	Femoral — JL4, JR4, pigtail covers almost every diagnostic case. Right radial — JL3.5 or a universal curve, JR4, pigtail
Which target?	Left → Judkins left. Right → Judkins right. LV or aorta → pigtail. LIMA → mammary. Graft to the right → Amplatz right or multipurpose
How wide is the aorta?	This is the only number, and it applies to the Judkins left alone. Average adult → 4. Small or a slim young patient → 3.5. Unfolded, dilated, elderly hypertensive → 5

Tortuosity is not on that list. It does not change which curve you need — it changes how much of your rotation reaches the tip. Same catheter, smaller increments.

Wind-up, curve selection and reading the tip

Rotation aims. Push and pull set the height. They are not interchangeable, and most of “it will not go in” is one being used for the other.

A preformed curve pressed against the far wall is a leaf spring, and the far wall is what it pushes off. That is the whole mechanism:

So the classic right-coronary move is a clockwise turn together with two or three centimetres of withdrawal — the turn aims it, the pull seats it.

How faithfully the hub reaches the tip is friction. Straight and supported, transmission is near 1:1 and you are steering. In a tortuous arch the tip sticks while wind-up climbs and then slips all at once and overshoots — torque whip. Small increments, and let it settle before adding more.

When it will not seat — four problems that all look the same

What you see	What it actually is	What to do
The tip floats in mid-aorta and never touches the far wall	The curve is shorter than the aorta is wide	Up a size. JL4 → JL5
The tip rides up the far wall, or the catheter buckles as you advance	The curve is longer than the aorta is wide	Down a size. JL4 → JL3.5
Right height, but pointing past the ostium	A rotation problem, not a sizing one	Small increments. A deep inspiration drops the root and often does it for you
It engages, then falls out on every breath	Not enough backup — too little wall to lean on	Amplatz, deliberately

How you know the width. In descending order: a prior CT or echo has already measured it, and from the right radial a wide ascending aorta on that study is what moves you from a JL3.5 to a JL4 before you open a single catheter.

The screen is next — on an AP or shallow LAO frame the ascending aorta silhouettes against the spine, and the answer you need is coarse: small, average or unfolded. Then an aortogram, if you were shooting one anyway. Last, the catheter itself: a JL4 floating in mid-aorta has just told you the root is wider than 4 cm.

The patient it happens to is usually elderly and hypertensive with an unfolded arch — and the same unfolding that widens the root lengthens the arch, so the sizing problem and the torque problem arrive together.

Curve selection

Catheter	How it finds the ostium	When you reach for it
Judkins Left	Advanced up the ascending aorta it already points at the left main — little or no torque	Routine left system; the number matches aortic width
Judkins Right	Clockwise 45–90° with 2–3 cm of withdrawal in the right cusp sweeps the tip anteriorly onto the RCA	Routine right system
Amplatz	Sits on the opposite wall for more backup	Wide root, anomalous or high origin. It also accounts for a disproportionate share of catheter-induced dissections

Reading the tip. Coaxial engagement keeps a crisp aortic waveform. Damping means the catheter is too big for the ostium or the tip is on the wall; ventricularisation means it is in or across an ostial stenosis.

Either way back out, and never inject into it: the waveform cannot tell you which of the two you are looking at.

Dissection does not come from the whip. In the largest adjudicated series — 96 dissections across 76,104 procedures, an incidence of 0.13 per cent — the commonest mechanism was a wedged contrast injection.

Deep intubation for support, catheter malalignment and prodding are the usual precipitants. Hydraulic and axial, not rotational.

Nor is it a novice complication: 94 per cent were inflicted by high-volume operators, and 84 per cent by operators with five or more years of experience. Once it starts it spreads in about 30 per cent of cases, and the usual trigger for that is a repeat injection through an unchanged catheter.

The view above is LAO, camera over the patient's left, which separates the two coronary origins instead of stacking them. 1.6 builds the whole model.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Klaudel J, Glaza M, Klaudel B, et al. Catheter-induced coronary artery and aortic dissections. A study of mechanisms, risk factors and propagation causes.	Cardiol J 2024;31(3):398-408	DOI

Checkpoint · pick the fix

Each is a real moment at the table.

The C-arm & the angiographic views

A coronary is a three-dimensional tube and every angiogram is a flat shadow of it. You angulate to lay a segment out and pull it clear of its neighbours.

The projection what appears on the monitor

Gantry where the C-arm sits

AP

Straight frontal overview. The left main and circumflex origin are foreshortened here — angulate to open them.

Following the LAD — teal is laid out, amber part-way, salmon nearly end-on. Rings mark where another branch crosses it.

Steeper angles cost dose — steep LAO most of all. 1.7 puts a number on it.

LAO / RAO (detector side)AP

Cranial / Caudal (detector tilt)0°

Follow a vessel

Standard views — jump to one

Reading the label — a view is named for where the detector is

LAO puts the detector over the patient's left, so you are looking from their left and the tube is on your side. RAO is the mirror of it. Cranial tilts the detector toward the head, caudal toward the feet. “RAO 30 / cranial 20” fully specifies the beam.

Steep costs dose, and lopsidedly: in RAO the patient stands between you and the tube, in LAO the tube is on your side of the table. 1.7 Radiation & the C-arm measures that trade in this same gantry, and shows what each piece of shielding takes off it.

Predict, then reveal

The same mid-LAD lesion measures 40 per cent in one projection and 70 per cent in another. Both runs are technically good. Which number goes in the report?

The two ways a view lies

Which view lays out which vessel

Four to carry. Cranial for the LAD. Spider for the left main. RAO caudal for the circumflex. LAO for the RCA, and add cranial for its far end.

The name is the hook: in the spider the left main is the body and the LAD and circumflex are the legs. If they are not spread apart, you are not in it yet.

System	View	What it lays out
Left	AP / RAO cranial	LAD and diagonals
Left	LAO cranial	LAD and diagonals, from the other side
Left	Spider (LAO caudal)	Left main and the LAD / circumflex bifurcation. The bifurcation wants more caudal angulation than the gantry can usually deliver — spider is the obtainable approximation, so take a second caudal view before committing
Left	RAO caudal	Circumflex and obtuse marginals
Right	LAO and RAO	The course of the RCA
Right	add cranial	Distal RCA, crux and PDA

Checkpoint · pick the view

Laid out and clear of its neighbours. Which view?

Radiation & the C-arm

1.6 chose the view. This one prices it. The same two sliders, but the readout is the dose arriving at your eyes — and what each piece of shielding takes off it.

The units, kept simple. µSv/min is how fast dose is reaching you right now — a microsievert is a millionth of a sievert, the unit for biological risk. mSv is the same thing scaled up (1 mSv = 1,000 µSv), and it is what dose badges and the yearly limits are counted in. Gy·cm² is the running total the machine shows — how much X-ray it has put into the patient so far — and plain Gy is that same total expressed as dose — a close stand-in for what lands on the patient's skin, which is the dose that can burn.

The room from the foot of the table

At your eyes per minute, as protected

AP

13.8 µSv/min

Tube under the table, scatter spread evenly. This is the cheapest place the gantry can be.

24 h of this to reach the 20 mSv annual eye-lens limit.

LAO / RAO (detector side)AP

Cranial / Caudal (detector tilt)0°

What the pedal is doing

Protection — switch each on

Standard views — jump to one

Four views, measured — scatter at eye height, per unit of what the machine reports

The primary beam is on the patient. What reaches you is scatter, and diagnostic-energy photons come off the patient preferentially back toward the tube. So the question is never how much beam there is — it is which side of the table the tube is on.

Measured at an operator's eye height in a working lab, per Gy·cm² of the dose–area product your machine already displays:

Projection	µSv per Gy·cm² at the eyes	Against AP
AP	2.3	—
Left lateral	12.0	5.2×
Spider (LAO caudal)	12.2	5.3×
LAO cranial	17.6	7.7×

The cheapest place the gantry goes is not AP but RAO 20 / 0, and the largest saving available to you is a substitution: trading a caudal LAO 60/20 for a caudal AP 0/30 — a similar look at the same vessel — cuts your own dose about eightfold. Move the slider between those two and watch the blob.

Predict, then reveal

The ceiling-suspended shield is the biggest thing you position at the table. In which projection does it help you least?

Why LAO costs and RAO does not — three rooms, seen from the foot of the table

Nothing about the patient's dose changes between these three. What changes is where the scatter is densest, and whether you are standing in it.

The protection ladder — measured factors, not claims

One step back is free. Sixty centimetres to ninety is 2.25× by inverse square alone — more than a second pair of glasses, and it costs nothing. Even ten centimetres, from 60 to 70, is about a quarter off — but the same step buys less and less the further back you already are.

Switch the four toggles on and off above and read the ladder. The two large numbers are not competing. 5.7× is what a ceiling shield does for your eyes, measured on radiologists at work. 19.6× is what a blanket on the patient takes out of what is still left once the skirt and that shield are already up — so it multiplies with the shield rather than beating it. And the two are complementary in the place it matters: the ceiling shield is worth least in left and cranial views, which is exactly where the blanket is worth most.

What it adds up to — over a career, not a case

A single case is trivial. The reason to care is that you will do several thousand of them, standing in the same place each time. The limits meant to keep you safe are 20 mSv a year to the eye lens and 20 mSv a year to the body under the apron — the table below is what turns up when they are ignored.

Finding	What was measured
Cumulative eye-lens dose in interventional cardiologists	129 operators, accumulated over a career averaging 22 years: 25 to over 1600 mSv, mean 423 ± 359 mSv — against a cumulative lens threshold of about 500 mSv, and an annual limit of 20 mSv.
Cataract threshold	Revised down to about 0.5 Gy — lens opacity is no longer treated as having a comfortably high threshold.
Brain tumours in interventionalists	31 physicians reported: 17 glioblastoma; of the 26 with a side recorded, 22 were left-sided — the side that faces the tube. A case series, not a cohort: it cannot give you a rate, only a direction.
Frame rate	7.5 → 4 fps cut median air kerma from 4 to 1.3 mGy in one series; 15 → 3.75 fps significantly lowered DAP in another. The cheapest dose reduction in the room is the one you set before you scrub.

The patient's dose — watch the skin

Your own dose builds up slowly over thousands of cases, and the harm is a small added chance of a cataract or a cancer years later. The patient's dose is a different problem: it all arrives in one sitting, on one patch of skin, and above a threshold it burns.

As you work, the machine shows a running total — the air kerma, in gray (Gy). It is a close stand-in for the dose on the patient's skin: peak skin dose is about 0.78× that number.

The thresholds worth knowing:

2 Gy to skin — where reddening (transient erythema) can begin.
5 Gy of air kerma on the machine — the substantial dose level. Not a limit and not a reason to stop: record it, tell the patient and the referring physician, and arrange a skin review at 2–4 weeks.
Same skin, within 6 months — doses to one field are added together, so a repeat procedure counts on top of the first.

How to keep it down, biggest savings first: run fewer frames (lower the frame rate and use low-dose fluoro), keep your foot off the pedal, collimate tightly, and use less magnification. Keep the detector low over the patient, and change the beam angle through the case so no single patch of skin takes the whole dose.

The full skin-injury picture — monitoring, pregnancy, and who is at highest risk — sits in Procedure & Complications.

Where these numbers come from

Every figure in this module is measured. The model behind the sliders is built from two of the papers below rather than invented, and where a scaling assumption is ours it is marked as such.

Measured — taken from the papers, not invented

Part of the model	What was measured
The angular surface	Four projections at operator eye height — AP 2.3, lateral 12.0, spider 12.2, LAO cranial 17.6 µSv per Gy·cm² — with everything between them interpolated. The dose line labels every value measured projection or interpolated, so no interpolated number is ever shown as if it had been measured.
The RAO limb	Shaped so its minimum falls at RAO 20, which is where the angulation survey found it.
The three pedal modes	The ratios of measured entrance dose rate — low fluoro 13, fluoro 39, cine 282 mGy/min.

Ours — two scaling assumptions, and what each one costs

Assumption	What it costs
Those pedal ratios applied to a nominal 6 Gy·cm²/min for a collimated adult coronary field	The mode ratios are measured, but the absolute rate depends on your field size and your patient.
Four protection factors, taken from four different studies, multiplied together	They are not all measurements of the same thing. 5.7× is eye-lens dose on working radiologists; 19.6× is whole-operator dose with the skirt and ceiling shield already in place. Multiplying them is defensible as a sequence — shield first, then blanket on top — but real layers overlap, so the product is an optimistic ceiling, not a prediction.

Neither affects the comparisons the module is teaching, which are all ratios.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Leyton F, Nogueira MS, Gubolino LA, et al. Correlation between scatter radiation dose at height of operator's eye and dose to patient for different angiographic projections.Scatter at eye height, by projection and pedal mode	Appl Radiat Isot 2016;117:100-105	DOI
Kuon E, Dahm JB, Empen K, et al. Identification of less-irradiating tube angulations in invasive cardiology.Which angulations irradiate least, and by how much	J Am Coll Cardiol 2004;44(7):1420-8	DOI
Morrish OWE, Goldstone KE. An investigation into patient and staff doses from X-ray angiography during coronary interventional procedures.Cine against fluoroscopy, and where backscatter goes	Br J Radiol 2008;81(961):35-45	DOI
van Rooijen BD, de Haan MW, Das M, et al. Efficacy of radiation safety glasses in interventional radiology.The 7.9-10x bench figure, the 2.1x clinical one, and the ceiling shield’s 5.7x	Cardiovasc Intervent Radiol 2014;37(5):1149-55	DOI
Petrucci C. Review of experimental estimates for the protection afforded by eyewear for interventional x-ray staff.Why eyewear factors vary so widely between studies	J Radiol Prot 2020;40(2):R46-R70	DOI
Davidsen C, Bolstad K, Ytre-Hauge K, et al. Effect of an optimized X-ray blanket design on operator radiation dose in cardiac catheterization based on real-world angiography.The blanket’s 94.9% (19.6×), across 7681 real procedures	PLoS One 2022;17(11):e0277436	DOI
Jacob S, Donadille L, Maccia C, et al. Eye lens radiation exposure to interventional cardiologists: a retrospective assessment of cumulative doses.O’CLOC — what a career actually delivers to the lens	Radiat Prot Dosimetry 2013;153(3):282-93	DOI
Shore RE. Radiation and cataract risk: Impact of recent epidemiologic studies on ICRP judgments.Why the cataract threshold was revised downwards	Mutat Res Rev Mutat Res 2016;770(Pt B):231-237	DOI
Roguin A, Goldstein J, Bar O, Goldstein JA. Brain and neck tumors among physicians performing interventional procedures.The reported left-sided tumour cluster	Am J Cardiol 2013;111(9):1368-72	DOI
Kirkwood ML, Arbique GM, Guild JB, et al. Radiation-induced skin injury after complex endovascular procedures.Deterministic skin dose, and where the 5 Gy notification level comes from	J Vasc Surg 2014;60(3):742-8	DOI
Boudjemline Y. Effects of reducing frame rate from 7.5 to 4 frames per second on radiation exposure in transcatheter atrial septal defect closure.What one step down in frame rate buys	Cardiol Young 2018;28(11):1323-8	DOI
Tanıdır İC, Gökalp S, Özturk E, et al. Is it possible to reduce radiation exposure during transcatheter atrial septal defect closure in children?Four steps down in frame rate, and the dose-area product	Turk Kardiyol Dern Ars 2020;48(8):760-5	DOI

Checkpoint · price the room

Three decisions you make with your foot on the pedal.

Floating the balloon — where am I?

You cannot see the catheter. You read its position off the waveform. Inflate the balloon and the bloodstream carries the tip forward: RA → RV → PA → wedge.

The tip flow-directed, from the right IJ

The trace normal haemodynamicsy: pressure at the tip, mmHg · x: time, a 6-second window

Right atrium

where the tip is

—

pressure

—

mean

—

depth

Insertion depth21.0 cm

Balloon

Patient

The four signatures — four traces, one pressure scale

Why the systolic does not change at the pulmonic valve. In systole the valve is open, so the ventricle and the artery are one space at one pressure. Only diastole differs: the RV falls toward zero, the artery holds a residual.

Three patients, one catheter — and the number that separates them

mean PA pressure ≈ (cardiac output × pulmonary vascular resistance) + left atrial pressure Every term in mmHg. PVR in Wood units is mmHg per L/min, so flow × resistance comes out as a pressure.

So a pulmonary artery pressure is high for one of three reasons — too much flow, too much resistance, or too high a pressure downstream. The wedge pressure separates them: it reads through the capillary bed into the left atrium, so it is that downstream pressure.

Subtract it from the PA diastolic and what is left — the gap — is the pressure drop across the lung’s own vessels.

What “capillary” means here. The two names say where the pressure comes from relative to the lung capillaries — the bed the wedge reads through. Pre-capillary: the disease is in the pulmonary arteries feeding those capillaries, so the wedge beyond them stays normal and the gap is wide.

Post-capillary: the pressure is generated in the left heart and pushed backwards through the pulmonary veins, so the wedge rises with the PA and the gap stays narrow.

Why that decides the case. Wide gap → pulmonary vasodilators. Narrow gap under a high wedge → treat the left heart; vasodilate that and you flood the lung.

Formally the split is the wedge at 15 mmHg plus a PVR over 2 Wood units, and PVR predicts outcome better than the gap does — the gap is the read you do at the table.

Then look for: RA pressure and RVEDP climbing, which is the RV failing against the load — and a tall v wave on a high wedge, which is a stiff left atrium or mitral regurgitation.

Why the balloon floats, why the wedge reads the left atrium, and the rule that never bends

Why the balloon floats — the catheter is not steered to the PA, it is carried there

	Balloon inflated	Balloon deflated
The tip presents	A large cross-section for the blood stream to push on	A small stiff point in a moving chamber
Against the wall	Rides over trabeculae	Catches on trabeculae, burrows into myocardium
How it advances	Carried by flow — so it goes wherever the flow goes	Will not reliably cross the pulmonic valve
In the RV	Quiet	Irritates myocardium → ectopy

Why the wedge reads the left atrium. Occlude a branch and flow beyond it stops. No flow means no resistive pressure drop, so tip, capillary bed and pulmonary veins become one static column ending in the left atrium — which is also why wedge a and v waves arrive late against the ECG.

A true wedge against an over-wedge

	True wedge	Over-wedge
The column	Open all the way to the left atrium	A sealed pocket the balloon is slowly compressing
Waveform	Venous shape — a and v waves	a and v waves gone
Over time	Stable	Drifts steadily upward
What to do	Record it	Deflate, withdraw a centimetre or two, try again

Never pull back inflated. The balloon is wider than the orifices and chordae it sits among, so dragging it backwards can tear a leaflet or avulse chordae. Forward inflated, backward deflated — without exception.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Swan HJC, Ganz W, Forrester J, et al. Catheterization of the heart in man with use of a flow-directed balloon-tipped catheter	N Engl J Med 1970;283(9):447-51	DOI
Humbert M, Kovacs G, Hoeper MM, et al. 2022 ESC/ERS guidelines for the diagnosis and treatment of pulmonary hypertension	Eur Respir J 2023;61(1):2200879	DOI

Checkpoint · name the trace

Three traces, no labels.

Reading the wedge — a, v, and when the number lies

A wedge is a left atrial pressure that has travelled backwards through a lung to reach you. Three things spoil it on the way: the breath, the shape of the wave, and a balloon that is not doing what you assume.

What is generating it left atrium, mitral valve, left ventricle

The breath y: pleural pressure, mmHg · x: time, two breaths

The wedge normal · true wedge · breathing spontaneouslyy: pressure at the tip, mmHg · x: time, a 6-second window

Live. Freeze the strip, then drag the cursor to the point you would read.

—

wedge mean, end expiration

averaged over one beat, at the plateau

—

a wave / v wave

the two peaks on this trace

—

PA diastolic

from 2.1 — a wedge must sit at or below it

—

true LVEDP

the answer the wedge is standing in for

The patient

What the left atrium is actually doing.

The catheter

Two of these produce a number that is not a wedge pressure at all.

The breathing

Laboured = active expiration, as in obesity or COPD.

Telling a from v — two different events, not two shapes to memorise

a is a squeeze, v is a fill. That is the whole difference, and every abnormal version follows from it: no atrial contraction in atrial fibrillation means no a wave; a mitral valve that leaks adds the ventricle’s systolic pressure to a chamber that is already filling, which is where a giant v wave comes from.

Timing places them if the shape does not. The a follows the P wave; the v lands after the T. On a wedge both arrive late against the ECG.

Predict, then reveal

A ventilated patient. The wedge trace rides up and down with every breath: about 20 at one end of the swing and about 8 at the other. The monitor prints its own mean of 12. Which of the three is the filling pressure?

Three ways the number lies

The mean the monitor prints averages three things that should never have been averaged together. Each has its own fix.

What distorts it	What it does to the trace	What to read instead
The breath	Your transducer is referenced to the room, so every number carries whatever the pleural space is doing. Spontaneous inspiration drags the trace down; a ventilator pushes it up.	End expiration. The high plateau if they are breathing on their own, the low point if the ventilator is doing it. Never the machine's average.
The v wave	An atrium that is small, stiff, or being filled backwards through a leaking mitral valve throws a v wave that towers over the trace and drags the mean up with it.	The trough of the x descent, just before the v wave climbs. In mitral regurgitation with large v waves that is the point that matched LVEDP — and unlike the a wave it survives atrial fibrillation. Only when the v wave is genuinely large: with a small one it under-reads.
The valve	A wedge stands in for LV filling pressure only if nothing obstructs between atrium and ventricle. In mitral stenosis the wedge reads 22 while the ventricle sits at 8.	Nothing on this trace. The difference is stranded across the valve, and the wedge is an accurate measure of the wrong chamber.

Timing, transmural pressure, v waves and the two free confirmations

Why the wedge arrives late. The pressure travels from the left atrium backwards through the pulmonary veins and capillary bed and then up a long fluid-filled catheter — tens of milliseconds with a stiff end-hole catheter, longer with a balloon-tipped one and longer still as the heart rate rises.

Every wave lands later against the ECG than the same wave on a right atrial trace.

Where the waves land — morphology alone tells a wedge from a damped PA trace

Trace	a wave lands	v wave lands
Right atrium	Just after the P wave	Late systole, into the T wave
Wedge	After the QRS, in the ST segment	After the T wave has finished

The delay matters for a mitral gradient: overlay a delayed wedge on a simultaneous LV trace and planimeter between them and you will exaggerate it, and under-read the valve area with it. Slide the wedge back by the transit time first.

The mean wedge tracks mean left atrial pressure closely; it is the phasic gradient that is corrupted. Where the answer really matters, measure left atrial pressure directly by transseptal puncture.

Transmural pressure is what the heart feels. A chamber is distended by the difference between the pressure inside it and outside it. Your transducer measures only the inside, referenced to the room.

At end expiration in a relaxed patient the outside is close to atmospheric, so the reading is a fair stand-in. At any other moment in the breath it is not.

When end expiration itself lies

At end expiration	Pleural pressure	What the wedge does	What to read
Relaxed patient	≈ 0	Reports transmural pressure faithfully	End expiration
Actively pushing air out — large abdomen, obstructive disease, distress	Positive	Reads high by that amount	Average over three or four cycles — but only an oesophageal balloon gives a true transmural pressure

Not a rare curiosity. Among 329 consecutive spontaneously breathing patients, 29 per cent of those clinically classified as pre-capillary pulmonary hypertension had an end-expiratory wedge above 15.

Why a big v wave is not a diagnosis. The v wave is a compliance phenomenon: volume arriving in the atrium during ventricular systole, divided by how stiff that atrium is. Both halves of the fraction move, which is why the wave tracks acuity rather than severity.

Situation	What the atrium is doing	v wave
Acute severe MR	A large volume arrives suddenly in an atrium that has had no time to dilate	Giant
Chronic severe MR	The atrium has stretched; the same volume barely raises the pressure	Often small or absent
No MR at all	Ventricular septal defect, mitral stenosis, plain heart failure	Can still be large

Across 1,021 consecutive catheterisations with wedge traces available in 208, 36 per cent of the 50 patients with a large v wave (10 mmHg or more above the mean wedge) had no or only trace mitral regurgitation, and 32 per cent of the 37 with severe regurgitation had only trivial v waves.

Only 43 per cent of that severe group produced a large v wave at all. Mitral regurgitation is the commonest cause of a large v wave and the large v wave is still a poor test for it.

Two opposite failures — and the one comparison that separates all three

State	What the balloon did	What the trace looks like	Mean vs PA diastolic
True wedge	Occluded the segment cleanly	Venous shape — a and v waves, arriving late	Below it, or level
Over-wedged	Sealed a pocket and is now compressing it	a and v waves gone; drifts steadily upward	Climbs above it
Partly wedged	Never occluded at all — still connected to the PA through a narrowed channel	Rounded, damped arterial trace	Sits at it, never below

A mean above the PA diastolic should make you suspicious — but it is not proof. A tall v wave drags the mean above it, and in heart failure that combination is common enough that the finding on its own proves nothing.

What condemns a tracing is the shape: no a and v waves, or a damped arterial contour that never falls below the PA diastolic.

Over-wedging is not only a wrong number. The balloon that pressurises a segment can also tear it, and pulmonary artery rupture is the one complication of a wedge that is immediately life-threatening. An overinflated balloon, a tip that has migrated distally, and a long inflation are the setting for it. So the response to a trace that starts drifting upward is mechanical, not interpretive: deflate at once, pull back, then re-inflate slowly with the smallest volume that damps the trace.

Two confirmations that cost nothing

Do this	Because	Confirmed if
Aspirate from the wedged tip	Occluding the segment makes it high-ventilation, low-perfusion, so blood drawn back past the balloon is pulmonary capillary blood — but you must clear the PA column first, so take 7 to 10 mL	Saturation within about 5 points of the patient's own arterial saturation — a relative rule, not a fixed 90 or 95, which a hypoxaemic patient's true wedge never reaches. Expect to fail on the first pass — deflate, re-wedge, repeat
Check where the tip sits	The column only reaches the left atrium if the pulmonary veins in that segment are open	Tip lies below the level of the left atrium — West zone 3

High PEEP and low filling pressures both shrink zone 3. A wedge recorded outside it is reading alveolar pressure, not left atrial pressure. And PEEP is transmitted to the wedge even from inside zone 3, so a PCWP that rises after the PEEP is turned up is not necessarily a rising filling pressure.

What you actually report — the convention changes with the chamber

Pressure	Reported as	Why
Atrial and wedge	A mean	There is no systole and diastole to quote — the trace is a train of a and v waves around one filling pressure
Ventricular and arterial	Systolic / diastolic	The two extremes are the physiology; a mean throws the pulse pressure away
LVEDP	At the Z point	The end of the a wave, at the foot of the systolic upstroke — not the nadir of the trough, which is read before atrial contraction has finished filling the ventricle. Reading the trough is the classic misread

Sources — verified on PubMed; each row links to its DOI

Study	Citation
LeVarge BL, Pomerantsev E, Channick RN. Reliance on end-expiratory wedge pressure leads to misclassification of pulmonary hypertension	Eur Respir J 2014;44(2):425-34	DOI
Fuchs RM, Heuser RR, Yin FC, Brinker JA. Limitations of pulmonary wedge V waves in diagnosing mitral regurgitation	Am J Cardiol 1982;49(4):849-54	DOI
West JB, Dollery CT, Naimark A. Distribution of blood flow in isolated lung; relation to vascular and alveolar pressures	J Appl Physiol 1964;19:713-24	DOI

Checkpoint · three numbers about to be acted on

Three tracings, three decisions.

Cardiac output — two methods, one patient and they disagree

Pulmonary and systemic vascular resistance, cardiac index, valve area and the decision to start inotropes are all built on a flow that is never actually measured. It is inferred, by two methods that fail in opposite directions.

What you are measuring, and the only two ways to get it

Cardiac output is the litres of blood the heart moves each minute. It is the number that decides what every pressure in this study means — a pulmonary artery pressure of 40 is a different disease at an output of 3 than at an output of 10. It cannot be seen or weighed, so it is inferred, two ways:

Method 1 · ThermodilutionInject cold, watch how fast it is washed away. Fast flow sweeps it past — a small tight curve. Slow flow lets it linger — a wide one.

Method 2 · FickDivide the oxygen the body consumes each minute by how much oxygen each litre of blood gave up crossing the lung.

They break for unrelated reasons, so you run both. What you are after here is the output, the cardiac index that goes with it, and — when the two disagree — which of them this patient has broken.

Method 1 Thermodilution — where the number comes from, and why the injectate matters

Why three injections, not one. A single curve can be spoiled by a slow push, a breath or a wandering baseline, and you cannot tell a spoiled curve from a real low output by looking at it.

Three lets you read the spread: tight agreement means the number is real, a wide spread means at least one injection was not clean. The panel reports the mean of three and the spread beside it — the spread is the quality control.

The catheter does the whole measurement by itself. Cold goes in at the proximal port, sitting in the right atrium; the thermistor 30 cm downstream in the pulmonary artery watches the blood temperature dip and come back. The computer integrates that dip.

The injectate sets how much signal you get. A bigger, colder bolus makes a deeper curve, and a deep curve survives noise, drift and a slightly imperfect injection.

10 mL iced gives the most signal; 10 mL at room temperature is the usual compromise; 5 mL at room temperature is the one that produces an unreliable number at the extremes of output, because there is barely a curve left to integrate.

Method 2 Fick — what it needs, and where it bends

No catheter manoeuvre at all: two blood samples and one assumption. Draw the PA sample (mixed venous — blood on its way into the lung) and an arterial sample (blood on its way out). The difference between them is how much oxygen each litre gave up.

cardiac output = oxygen taken up each minute ÷ (haemoglobin × 13.4 × the saturation gap) The oxygen uptake is normally assumed from body surface area rather than measured — that assumption is the weak link, and the control below lets you swap it for a measured value and watch what moves.

Method 1 Thermodilution 10 mL room temperature · injection 0 of 3y: blood cooling, °C below baseline · x: time since injection, seconds

Inject cold saline into the right atrium and watch the blood cool as it passes the thermistor.

—

thermodilution, mean of 3

—

injection 1

—

injection 2

—

injection 3

—

spread

No injections yet.

Method 2 Fick oxygen uptake divided by what the blood gave up

Two saturations and an oxygen uptake. Nothing else.

The arithmetic

—

Fick cardiac output

—

a–v O₂ difference

—

oxygen uptake

How fragile is this answer

Draw both samples to find out.

The patient both methods

The injectate method 1 only

How much cold you put in is how much signal you get back.

Oxygen uptake method 2 only

The convenient number, or the one from the metabolic cart.

Two methods, and only two — there is no third way to get a cardiac output here

	Thermodilution	Fick
What it measures	Right heart flow — how fast a known amount of cold is washed away	Whole-body oxygen flux — how much oxygen the blood picked up crossing the lung
What you must supply	A syringe of saline at a known volume and temperature	An oxygen uptake, plus a PA and an arterial saturation
What goes wrong	Cold is lost on the way, or the same cold is counted twice	The inputs. Oxygen uptake is usually assumed from body surface area
Which way it errs	Slow transit lets heat leak back in — reads falsely high. A leaking tricuspid re-counts the cold — reads falsely low.	An assumed uptake pulls every patient toward the population mean. The saturation gap narrows as output rises, so a sampling error that was trivial at an output of three becomes a litre at ten.
Distrust it when	Output is low, or the tricuspid valve leaks	Output is high, or the oxygen uptake was guessed
So prefer it in	A normal or high output, a competent tricuspid	Low output, severe tricuspid regurgitation, any shunt

Take both, and read the disagreement. Within about twenty per cent use either — ten is inside thermodilution’s own reproducibility and would call almost every patient discordant. Past that, the disagreement is the finding: prefer the method this patient has not broken, and near a decision threshold trust neither alone.

Predict, then reveal

A man in the unit after an inferior infarct. The PA catheter reports a cardiac output of 2.6 L/min and the team is calling for a balloon pump. He is warm, his lactate is normal, his urine output is fine and his mixed venous saturation is 66 per cent. On the right atrial trace there is a large systolic wave. What is going on?

Where the cold goes, why Fick's inputs are the weak link, and what the evidence shows

Stewart and Hamilton, in one line. Put a known amount of indicator into a stream and watch all of it pass a detector downstream: amount in = flow × area under the concentration curve. Rearranged, flow = amount ÷ area. Everything that goes wrong with thermodilution goes wrong in one of those two terms.

Term	What it is	How it breaks
Amount	Injectate volume × how far below blood temperature it is	You delivered less cold than you think — small volume, warm injectate or cold lost in transit
Area	The integral of the temperature deflection over time	You recorded the wrong area — drifting baseline, truncated curve or indicator crossing the thermistor more than once

Why the curve widens as flow falls. Mean transit time is the volume between injection port and thermistor divided by flow. Halve the flow and the same bolus takes twice as long, so the curve is later, broader and — because the area must double — taller.

That width is genuine information: a curve taking fifteen seconds to return to baseline belongs to a patient with low output, whatever the computer prints.

Where the cold goes — the two ends fail for opposite reasons

At	What happens to the bolus	Effect on area	Error	Remedy
Low output	Transit is long, so heat leaks in from the vessel wall, surrounding blood and the catheter itself. A larger fraction never reaches the thermistor.	Lost cold is missing area	Reads falsely high	More cold. 10 mL iced carries roughly four times the indicator of 5 mL at room temperature.
High output	Nothing is lost, but there is barely anything to see. The peak may be a fifth of a degree, and the computer must decide where baseline was before it can integrate.	Baseline drift becomes spurious area	Reads unreproducibly	10 mL iced again, and repeat until the three agree.

Blood temperature is not steady — it drifts with respiration and with any fluid running into the vein. A hundredth of a degree of baseline error across a fifteen-second integration is trivial against a fat curve, catastrophic against a thin one.

What the animal work showed

Model	Condition	Finding
Sheep, electromagnetic flow probe as reference	Below 4.7 L/min	Overestimation grew as the indicator was reduced, reaching 21 per cent for a 5 mL room-temperature injectate. At this end it is accuracy that goes, not reproducibility
Same preparation	Above 7.7 L/min	Reproducibility of 5 mL room-temperature collapsed to r² 0.08; 10 mL iced held at 0.81

The scatter across your three injections is the warning light. More than about ten per cent apart and the answer is not a number — it is a repeat. Even then, a real change between two averaged sets of three has to clear roughly 13 per cent before you can call it a change at all.

Why regurgitation counts the cold twice. Thermodilution assumes every particle of indicator passes the thermistor exactly once. A regurgitant tricuspid valve breaks that: cold is thrown back into the right atrium during systole and re-ejected on the next beat, so some crosses two or three times.

The curve becomes low, broad and often visibly double-humped, and the recorded area is larger than the injected amount justifies — which the computer can only read as low flow.

27 ventilated patients, thermodilution against Doppler — spreads are limits of agreement, not SD

Tricuspid regurgitation	Disagreement with Doppler	Correlation
None or trivial	0.5 ± 1.1 L/min	0.96
Third-degree	1.9 ± 2.3 L/min	0.69

Severe tricuspid regurgitation makes thermodilution read low, and a patient can be declared to be in cardiogenic shock on the strength of it.

Fick's numerator. Oxygen uptake is measurable with a Douglas bag or metabolic cart, and almost never is. The substitute is a formula from body surface area, usually 125 mL/min/m² — a real figure, and an average. An average is not a measurement.

Study	Patients	What it found
Dehmer 1982	108 adults at catheterisation	Uptake 126 mL/min/m² — but SD 26 and a range of 65 to 250. Even this was back-calculated, not collected in a bag
Kendrick 1988	80 catheterised patients	More than half of the assumed values differed from the measured value by more than 10 per cent
Wolf 1998	57 adults	The formulae systematically underestimated high uptakes and overestimated low ones — the error is not random

So the assumption pulls every patient toward the population mean and flattens exactly the extremes you catheterised them to find.

The paper everyone cites for it did not say it. The flat 125 is routinely attributed to LaFarge and Miettinen, and that attribution is wrong. What they published in 1970 was not an index at all but a pair of regression equations, one for each sex, predicting oxygen consumption from age and heart rate. In Kendrick's head-to-head comparison those equations came closest to the measured value — which is the useful half of the point. If you have to assume a number, assume one that at least moves with the patient in front of you. It is still an assumption.

Fick's denominator. The a–v difference is proportional to the gap between two saturations, and that gap shrinks as output rises — so the same sampling error is worth more.

Cardiac output	Saturation gap	A 1-point co-oximeter error	A 2-point sampling error
2.5 L/min	~40 points	2.5 per cent	Barely visible
10 L/min	~18 points	5.6 per cent	More than a litre a minute

An incompletely mixed sample, a tip that has drifted, a slow draw that entrained wedged blood — any of these is a two-point error. The fragility is a property of the fraction, not of your technique.

Agreement in practice — 253 consecutive patients, thermodilution against Fick with measured oxygen uptake

Median absolute difference	Differed by more than a quarter	Correlation	Direction of bias
17.5 per cent	More than 30 per cent of patients	0.64	None — neither reliably high nor low

The absent bias is the uncomfortable part: a consistent bias can be corrected for, a random one cannot. Which is why the number you write in the report should carry the method that produced it.

Five percentages have gone past, and no two of them mean the same thing. They are easy to blur together, and blurring them is how a repeat injection gets skipped or a real change gets called noise.

The number	What it is a percentage of	What it asks you to do
about 10 per cent	The spread across your own three injections — thermodilution’s reproducibility in this patient, right now	Wider than this and the answer is not a number, it is a repeat
about 13 per cent	The least change worth believing between two averaged sets of three, at two moments in time	A smaller “improvement” after an intervention is noise, not a response
17.5 per cent	The median disagreement between the two methods in the same patient, with the oxygen uptake actually measured	Expect it. Two valid methods routinely differ by this much
about 20 per cent	The band inside which the choice of method does not change management	Use either number. This is a tolerance you accept, not one you achieved
±25 per cent	The error carried by an assumed oxygen uptake — before either method is compared to anything	It is inherited by the Fick number, and it is directional, not random

Only the first is about your technique. The second is about time, the third and fourth about the two methods, and the last about the assumption underneath one of them.

Index it, then decide. Four litres a minute is generous in a small elderly woman and inadequate in a large man. Divide by body surface area before you decide anything.

Sources — verified on PubMed; each row links to its DOI

Study	Citation
Norris SL, King EG, Grace M, Weir B. Thermodilution cardiac output — an in vitro model of low flow states	Crit Care Med 1986;14(1):57-9	DOI
Renner LE, Morton MJ, Sakuma GY. Indicator amount, temperature, and intrinsic cardiac output affect thermodilution cardiac output accuracy and reproducibility.	Crit Care Med 1993;21(4):586-97	DOI
Balik M, Pachl J, Hendl J. Effect of the degree of tricuspid regurgitation on cardiac output measurements by thermodilution	Intensive Care Med 2002;28(8):1117-21	DOI
LaFarge CG, Miettinen OS. The estimation of oxygen consumption	Cardiovasc Res 1970;4(1):23-30	DOI
Dehmer GJ, Firth BG, Hillis LD. Oxygen consumption in adult patients during cardiac catheterization	Clin Cardiol 1982;5(8):436-40	DOI
Kendrick AH, West J, Papouchado M, Rozkovec A. Direct Fick cardiac output: are assumed values of oxygen consumption acceptable?	Eur Heart J 1988;9(3):337-42	DOI
Wolf A, Pollman MJ, Trindade PT, Fowler MB, Alderman EL. Use of assumed versus measured oxygen consumption for the determination of cardiac output using the Fick principle	Cathet Cardiovasc Diagn 1998;43(4):372-80	DOI
Narang N, Thibodeau JT, Parker WF, et al. Comparison of accuracy of estimation of cardiac output by thermodilution versus the Fick method using measured oxygen uptake	Am J Cardiol 2022;176:58-65	DOI

Checkpoint · three sets of curves

In each one somebody is about to act on a number.

Oximetry — finding and sizing a shunt

A pressure tells you what the chambers are doing. A saturation tells you where the blood has been. Whether there is a shunt, where it is, how big it is and whether it should be closed all come out of samples drawn on the way in — and nothing else.

What you are looking for, and how a saturation finds it

A shunt is blood taking a short cut. A defect between the left and right sides lets oxygenated blood cross back into the right heart, so the lungs handle more blood than the body does. You cannot see the defect from here — but the blood that crossed is bright, and it stays bright all the way downstream.

Step 1 · Is there one, and where?Sample every level on the way in. In a normal heart the saturation is flat. A step-up between two adjacent levels means oxygenated blood is joining there.

Step 2 · How big is it?Four saturations give Qp:Qs, the ratio of lung flow to body flow. 1.0 is normal; 1.5 or more is in the closure conversation.

The comparison Normal against a shunt — the whole module in one picture

Read the ladder left to right and look for the one place it jumps. Nothing is added to blood on its way through a normal right heart, so the line is flat within a few points of sampling scatter.

A step bigger than the threshold at that level — 7 at the atrium, 5 at the ventricle or great vessel — is a shunt, and the level it happens at is where the defect is.

Trap 1 — a big step is not a big shunt. The size of the jump is the shunt fraction multiplied by how desaturated the venous blood was. A low-output patient with a wide arteriovenous difference throws a big step from a modest shunt; a high-output patient can hide a large one. Only the ratio sizes a shunt.

Trap 2 — take the mixed venous from before the shunt, not after it. Mixed venous is meant to be the body’s returning blood. Sample it in the PA and you are sampling returning blood plus the shunt, so the shunt cancels itself out of your arithmetic. Take it from the great veins instead: (3×SVC + IVC) ÷ 4.

The run no samples drawn

Sample in from the great veins to the PA, then take an arterial sample.

Saturation ladder every sample, in the order the catheter meets themy: oxygen saturation, % · x: sampling site, in catheter order

—

atrial step, %

RA − MV · need 7

—

ventricular step, %

RV − RA · need 5

—

great-vessel step, %

PA − RV · need 5

—

mixed venous, %

(3×SVC + IVC) ÷ 4

No samples yet.

The patient normal first, then four unknowns

Sampling

A co-oximeter reports whole percentages and no two draws agree exactly.

Pulmonary vein saturation the PV in the ratio

Blood leaving the lung is normally 95–100% saturated, so 98 is assumed rather than measured — crossing to the left side to sample it is rarely worth the risk. It is the PV term in Qp:Qs.

The arithmetic what your samples give, and what was actually there

Where this answer can break

Draw the samples and find out.

Predict, then reveal

A fellow presents a study on a 44-year-old woman with a dilated right ventricle on echo. His run reads: SVC 70, IVC 76, RA 86, RV 87, PA 87, femoral artery 97 per cent. He divides ninety-seven minus 86 by ninety-eight minus eighty-seven, reports a Qp:Qs of 1.0, and concludes the defect is trivial. Where did the shunt go?

How to run one

A full run is twelve blood samples, drawn on the way in, in the order below. The middle column is how many samples to take at that level — more than one wherever a single sample could lie to you.

Where to draw, in order	How many samples	Why that many
SVCone high, one low	2	If the two disagree, a pulmonary vein is draining into the vena cava between them. Use the lower one — the upper sample is the contaminated one.
IVCabove and below the liver veins	2	Blood coming back from the liver is not the same as blood coming back from the legs and kidneys.
RAhigh, middle, low	3	A shunt jet does not mix evenly across a chamber, so a single sample can sit in the wrong stream and miss it.
RVinflow, body, outflow	3	The same reason — and where in the ventricle the saturation is highest points to where in the septum the defect sits.
PAthe main trunk	1	Blood here is fully mixed, so one is enough. Never use it as your mixed venous: this is the blood the shunt has already joined.
Arteryfrom any sheath	1	One sample, for the body side of the ratio — the Ao term in the arithmetic above.

All twelve in one sitting — sedation and ventilation drift between runs

The thirteenth port is the pulmonary vein — assumed at 98%, not drawn

Where these numbers come from

1 · Why the ratio is only saturations Apply Fick twice and divide. Almost everything cancels.

Qp = VO₂ ÷ ( Hb × 13.4 × (PV − PA) )

Qs = VO₂ ÷ ( Hb × 13.4 × (Ao − MV) )

Qp:Qs = Ao − MVPV − PA VO₂ and Hb sit in both and cancel. Four saturations survive.

Which is why	A guessed VO₂	A haemoglobin from yesterday
Qp:Qs — the number decisions hang on	no effect	no effect
Qp and Qs in L/min — printed beside it	wrong in proportion	wrong in proportion

2 · What counts as a step-up Right-sided saturations are not uniform even in a normal heart.

Difference, in 980 adults with no shunt	Mean ± SD	Mean + 2 SD
Superior vena cava to right atrium	3.9 ± 2.4	8.7
Right atrium to pulmonary artery	2.3 ± 1.7	5.7
Superior vena cava to pulmonary artery	4.0 ± 2.5	9.0

Applied to the 51 patients in that series who did have a shunt, those limits found 46 — sensitivity above 90 per cent, specificity 94 to 95. All five misses had a Qp:Qs of 1.9 or less.

Level	Mean of the samples	One sample only	A shunt here is usually
Great veins → atrium	≥ 7	≥ 11	Secundum or sinus venosus defect, anomalous pulmonary vein, ruptured sinus of Valsalva into the atrium
Atrium → ventricle	≥ 5	≥ 10	Ventricular septal defect, post-infarction septal rupture
Ventricle → great vessel	≥ 5	≥ 5	Patent ductus, aortopulmonary window, ruptured sinus of Valsalva into the pulmonary artery

Two numbers, one question. The mean-of-samples figure is the stricter and the more reproducible, and it is the one to use when each level was drawn in duplicate or triplicate. The single-sample figure is for the run you did not repeat. Say which one you used.

3 · The quick look and the full run are different tests Most of the information sits in one difference. Most of the diagnosis does not.

Can it…	Quick look SVC, PA, artery	Full run 12 drawn
Catch a large left-to-right shunt	yes	yes
Say where the shunt is	no	yes
Expose a contaminated great-vein sample	no	shows as a split
Recognise shunting in both directions	reads near 1.0	arterial sat, effective flow
Find a small shunt — Qp:Qs under 1.9	usually; the few it missed were all this small	yes, and it says where
Be finished in two minutes	yes	ten, in one sitting

4 · How precise is the ratio, really Against phase-contrast MR flow: close on average, wide in the individual case.

Compared with cardiac MR	Population	Mean difference	Limits of agreement
Beerbaum 2001	Children, atrial or ventricular shunt	~2%	−20 to +26%
Beerbaum 2008	Children, atypical atrial defects	~3%	±21 to 25%

Read a reported ratio as a band, not a point. A run that prints 1.5 — the ratio in the closure criteria — is compatible with a true ratio anywhere here. And 1.5 is never the whole case: the guideline also wants chamber enlargement, symptoms, no cyanosis and a pulmonary vascular resistance under a third of systemic.

1.0

1.2

reported 1.5

1.9

2.1

5 · When the numbers are not what they look like

What you see	What it usually is	What to do
High SVC several points above the low sample	A pulmonary vein draining into the cava — partial anomalous return, which travels with the superior sinus venosus defect and is often invisible on transthoracic echo	Use the lower sample in the formula. The split is the finding, not an artefact
Saturation climbing across inflow, body and outflow of the same ventricle	Blood entering through a mid-septal defect has not mixed yet	Sample every ventricle at more than one level
Mixed venous taken from the pulmonary artery	That is downstream of the shunt; shunted blood is counted on both sides of the fraction	Reconstruct it: three parts superior to one part inferior
Qp:Qs shifting between two runs on the same patient	You are reading the second decimal place of a wide band	Report a range, and repeat the samples that matter

The three-to-one weighting is not arbitrary: the head and arms extract more oxygen than the abdomen and legs, and Flamm measured systemic flow directly, at rest and on exercise, to find the mixture that reproduced it.

6 · Oximetry after an infarct A new murmur and collapse on day three to five. Two diagnoses, opposite operations, one minute to separate them.

Finding	Septal rupture	Acute mitral regurgitation
Step-up, right atrium to pulmonary artery	present	absent
Large V wave on the wedge tracing	5 of 6, with no mitral regurgitation at all	present
Harsh new murmur	yes	yes

In profound shock, surgical series have gone to theatre on the step-up alone rather than complete the study. The urgency is in the mortality: 83 per cent of those in severe cardiogenic shock died, against none of those with mild to moderate symptoms.

7 · When the ratio comes back near one Three possibilities. The arterial saturation separates them.

Qp:Qs near 1.0 because…	Arterial saturation	Next step
There is no shunt	Normal, 95 per cent or more	Stop. The echo was wrong, or the lesion is elsewhere
The samples are wrong	Normal	Redraw the run in one sitting, check the great-vein split
Blood is crossing both ways in equal amounts	Low — eighties on room air with clear lungs	Compute effective pulmonary flow and split the streams

Qep = VO₂ ÷ ( Hb × 13.4 × (PV − MV) ) the flow needed to carry the oxygen if nothing were shunting

L→R = Qp − Qep

R→L = Qs − Qep

Both streams can run near a litre a minute while their ratio sits at 0.9.

That is the physiology that makes closure harmful rather than helpful: close a defect with a meaningful left-to-right shunt and preserved pulmonary vascular resistance, and do not close one where the resistance has risen and the shunt has reversed.

Sources — verified on PubMed; each row links to its DOI

Reference	What it supplies	Link
Antman EM, Marsh JD, Green LH, Grossman W. Blood oxygen measurements in the assessment of intracardiac left to right shunts: a critical appraisal of methodology. Am J Cardiol 1980;46(2):265–71.	The step-up thresholds, by mean of samples and by single value	DOI
Flamm MD, Cohn KE, Hancock EW. Measurement of systemic cardiac output at rest and exercise in patients with atrial septal defect. Am J Cardiol 1969;23(2):258–65.	Mixed venous as three parts superior to one part inferior	DOI
Hillis LD, Firth BG, Winniford MD. Variability of right-sided cardiac oxygen saturations in adults with and without left-to-right intracardiac shunting. Am J Cardiol 1986;58(1):129–32.	How far saturations wander with no shunt; sensitivity and specificity	DOI
Beerbaum P, Körperich H, Barth P, et al. Noninvasive quantification of left-to-right shunt in pediatric patients: phase-contrast cine magnetic resonance imaging compared with invasive oximetry. Circulation 2001;103(20):2476–82.	Limits of agreement against MR flow	DOI
Beerbaum P, Parish V, Bell A, et al. Atypical atrial septal defects in children: noninvasive evaluation by cardiac MRI. Pediatr Radiol 2008;38(11):1188–94.	Ten of eighty-two had a sinus venosus defect with anomalous drainage; also the limits of agreement against MR flow	DOI
Yamasaki Y, Kawanami S, Kamitani T, et al. Noninvasive quantification of left-to-right shunt by phase contrast magnetic resonance imaging in secundum atrial septal defect. Int J Cardiovasc Imaging 2018;34(6):931–7.	The same comparison in adults, and the effect of breath-holding on it	DOI
Shimajiri H, Harada Y, Kinoshita M, Mikami S. Sinus venosus atrial septal defect and partial anomalous pulmonary venous connection in a patient with dextrocardia. BMJ Case Rep 2022;15(2):e245523.	An atrial step-up found on a high-versus-low right atrial sample	DOI
Bethea CF, Peter RH, Behar VS, et al. The hemodynamic simulation of mitral regurgitation in ventricular septal defect after myocardial infarction. Cathet Cardiovasc Diagn 1976;2(1):97–104.	Five of six septal ruptures had a large V wave without mitral regurgitation	DOI
Kopf GS, Meshkov A, Laks H, Hammond GL, Geha AS. Changing patterns in the surgical management of ventricular septal rupture after myocardial infarction. Am J Surg 1982;143(4):465–72.	Mortality by severity of shock	DOI
Stout KK, Daniels CJ, Aboulhosn JA, et al. 2018 AHA/ACC guideline for the management of adults with congenital heart disease: executive summary. J Am Coll Cardiol 2019;73(12):1494–1563.	When a defect should and should not be closed	DOI
Humbert M, Kovacs G, Hoeper MM, et al. 2022 ESC/ERS guidelines for the diagnosis and treatment of pulmonary hypertension. Eur Respir J 2023;61(1):2200879.	Pulmonary vascular disease and shunt reversal	DOI

Checkpoint · four runs

Each one is a ladder and a decision.

PCI Technique — Doing the Case

Everything up to this point decides whether to intervene. This is the part where you do, and almost all of it is one idea: get a rail across the lesion, make the vessel ready to take a stent, and then prove the stent is where it should be at the size it should be.

By the end of this section you should be able to

Run the sequence in order, and say what the wire is actually for
Pick a guide for backup, and recognise when the guide itself is the problem
Decide when calcium has to be modified, and which tool the arc calls for
Size a stent off the reference rather than off the lesion
Optimise a result against numbers instead of against an impression
Take a bifurcation provisionally, and name the case where you would not

The sequence

Five steps, always in this order. The variation between operators is in how much preparation a lesion needs, not in the order.

The wire is the rail, not the treatment. Everything after step 2 — balloon, stent, imaging catheter, atherectomy burr — runs over it. Losing wire position mid-case does not put you back at step 2; it puts you back at step 1 in a vessel that is now dissected and may not let you cross again. This is the same sentence as keep the wire in the dissection card, and it is the single highest-consequence habit on this page.

Guide, wire, support

A PCI fails at the delivery stage far more often than at the crossing stage. Support is what you are buying when you choose a guide.

The guide's job

Coaxial engagement, a stable platform, and enough backup that the balloon or stent advances instead of pushing the guide out of the ostium. Passive support comes from the curve resting on the opposite aortic wall — an EBU / XB on the left, an AL on the right when you need more than a Judkins gives you.

Damping and ventricularization

A damped or ventricularized waveform on engagement means the ostium is occluded, by disease or by the catheter itself. Pull back and do not inject. A side-hole guide relieves the pressure but hides the warning, so it is a considered choice, not a default.

Wires

A workhorse wire for almost everything. Polymer-jacketed for tortuosity or a difficult side branch, at the price of a much higher chance of going subintimal without telling you. Stiff and tapered wires are chronic-total-occlusion tools and cause perforations in ordinary hands. Wire the side branch you may want later before you stent the main vessel — rewiring through struts is harder than wiring an open vessel.

When support still is not enough

A guide extension, a buddy wire, or deep-seating the guide. Each buys delivery and each costs safety: a guide extension is a common cause of proximal vessel dissection, and deep-seating is how you damp an ostium you were not worried about.

Preparing the lesion — and what the calcium arc costs you

Under-expansion is the strongest predictor of stent failure, and calcium is the commonest reason a stent under-expands. Preparation is not a flourish; it is the step that decides whether the stent can reach its nominal diameter at all.

What decides the tool

The arc and the thickness of the calcium, and whether the balloon will cross at all. Angiography sees only dense, obvious calcium; the arc and thickness are measurements that only intravascular imaging gives you — which is why calcium is the strongest single indication to image before you treat.

A non-compliant balloon first

Take it to high pressure and watch for a waist. A balloon that will not fully expand is telling you the stent will not either. Believe it.

Scoring and cutting balloons

Create focal fracture lines in a moderate arc, and stop the balloon slipping — useful in ostial and in-stent lesions for that reason alone.

Atherectomy

Rotational or orbital, for calcium too heavy to crack or a lesion a balloon cannot cross. It removes tissue. Slow flow, burr entrapment and perforation are the specific costs.

Intravascular lithotripsy

Acoustic pressure waves fracture deep as well as superficial calcium without removing tissue, delivered on a balloon at low pressure. Easier to use than atherectomy; it does not help with a lesion the balloon cannot cross, because it is a balloon.

The decision you are actually making is when to stop preparing. The test is not the angiogram, it is a fully expanded non-compliant balloon at nominal pressure with no waist. A stent deployed into a lesion that would not let a balloon open is an under-expanded stent, and no amount of post-dilatation reliably fixes it afterwards.

Sizing — off the reference, not off the lesion

The commonest sizing error is measuring the diseased segment, which is by definition the smallest part of the vessel, and then implanting a stent that matches it.

Diameter

Size to the distal reference so the distal edge is not oversized, then post-dilate the proximal segment to the proximal reference with a shorter non-compliant balloon. A single stent sized to the proximal reference will tear the distal edge.

Length

Cover the whole plaque and land in normal vessel at both ends. A stent that stops in disease is a geographic miss — and the edge left behind is where the restenosis and the edge dissection appear.

Why imaging changes the number

Angiography systematically undersizes, because it shows the lumen and not the vessel. Sizing to the external elastic membrane on IVUS routinely produces a stent half a millimetre larger, and half a millimetre is a large fraction of a coronary.

Optimisation — four numbers, not an impression

A stent that looks good on the angiogram is the normal appearance of a stent that will fail. The angiogram cannot see expansion, apposition or an edge dissection, which are the three things that predict failure.

Expansion

The stented lumen against the reference. This is the one that matters most, and under-expansion is the dominant mechanism of both early thrombosis and late restenosis.

Apposition

Struts against the wall. Large acute malapposition, particularly at the edges, is the substrate for early thrombosis.

Edges

Residual disease and dissection at each edge — the geographic miss, made visible.

Inflow and outflow

The stent can be perfect and the result still poor if the segment either side is not.

The numbers, and where they come from. Minimum stent area, expansion against the reference, apposition and edge dissection are worked through with a live pullback in IVUS / OCT, which is where the thresholds and the trial evidence for imaging-guided PCI sit. This card is the reminder that optimisation is a step, not an afterthought.

Bifurcations — provisional until it is not

Most bifurcations are best treated with one stent in the main vessel and nothing in the side branch. The trials have repeatedly found that a two-stent strategy up front does not beat provisional, and costs more procedure, more contrast and more metal.

Medina

Three digits — proximal main vessel, distal main vessel, side branch — each 1 if diseased. It is a description, not a strategy: 1,1,1 and 1,0,0 are both usually treated provisionally.

POT

Proximal optimisation with a short balloon sized to the proximal reference, after stenting the main vessel. It corrects the malapposition that a single-diameter stent always leaves in the larger proximal segment, and it opens the cell in front of the side branch.

When the side branch is compromised

Rewire it through the distal cell, kissing balloon, then re-POT. Rewiring proximally leaves the struts across the ostium.

When to plan two stents

A true bifurcation with a large side branch, significant disease well into it, and an angle that will make rewiring difficult. That combination — not any one of them alone.

What goes wrong that is specific to doing this

Side-branch occlusion

Plaque shift as the main vessel is stented. Predicted by the size of the branch and how much disease sits at its ostium; prevented by wiring it first, which at minimum gives you a marker and a route back.

Geographic miss

Untreated disease at a stent edge, from a stent too short or a balloon that injured beyond it. It is a sizing and planning failure, and it shows up months later as edge restenosis.

Edge dissection

Frequently invisible on angiography. A large one with residual disease is treated; a small one in a normal segment is usually left. That distinction is an imaging call.

Under-expansion

The one that matters. It is a preparation failure much more often than a post-dilatation failure, which is why the preparation card sits above this one.

No-reflow, dissection, perforation

Recognition and management are worked through in Procedure & Complications, with the Ellis and NHLBI grades and the drugs.

Antiplatelet loading, the P2Y₁₂ agents and their hard limits are on the Drugs tab; the physiology gate that decides whether a lesion should be treated at all is in Angiography.

Educational summary. Technique varies between operators and centres; device selection and strategy follow current guidelines and are individualized.

The Procedure & Its Complications

The half of the cath lab that isn't a waveform or a calculation: how you get in, what you give, what the contrast and the radiation cost, and what to do when the room goes wrong. This is high-yield board territory and it is where most real-world harm happens.

▸ Concept primer · How this section is ordered new to this? start here

The six tabs run in the order these things actually happen in the room.

When	Tab	The question it answers
Getting in and out	Access & closure	Radial or femoral, where the needle goes, and how the hole is closed.
When that goes wrong	Access complications	Hematoma, pseudoaneurysm, retroperitoneal bleed, radial occlusion — how each one declares itself.
Before and during	Anticoagulation & drugs	What is given, when, and what to do when the ACT is wrong.
The price of the pictures	Contrast, kidney & radiation	Doses, thresholds, and who actually gets hurt.
When the room goes wrong	Trouble in the lab	Perforation, no-reflow, arrhythmia, allergy, air.
Off the coronary tree	Beyond the coronaries	Renal, peripheral and structural work done in the same lab.

Educational summary for teaching. Doses and thresholds are typical teaching values — always follow local protocol and current guidelines in practice.

Curriculum Map

The path through CathSim, and where you are on it. Thirteen Foundations modules build one case from the first decision to the last number; fourteen instruments go deep on what those modules opened. Every check you pass is recorded below. This is a work in progress and all content is under review.

Where you are

Foundations — the path

Thirteen modules in one order. Each hands the next a specific capability, and opens the instruments that go deeper on it. Work them in order; the instruments will make more sense for it.

The instruments

Fourteen deep-dive modules. Each rests on work done earlier — the chips say which, and fill in as you pass them. Nothing is locked; the chips are advice, not a gate.

Topics covered

The major cardiac-catheterization topic areas and how far the platform covers each one today.

Design principles

Interactive-first — concepts are learned by doing, through cases, simulation, and live tracings rather than passive reading.

Guideline-based — clinical content follows current society guidelines (ACC/AHA, ESC/ERS, SCAI) and standard references.

Integrated — calculations, waveforms, and angiography are taught in the clinical context that gives them meaning.

Extensible — a data-driven architecture: cases, questions, and waveform models are all straightforward to expand.

Guideline & reference sources

The clinical content is built from current society guidelines and standard references. Each case also cites its own primary sources on completion and in its generated report, so faculty can verify any teaching point.

Domain	Primary source (vintage)
Valvular disease	2020 ACC/AHA Valvular Heart Disease Guideline
Coronary revascularization / FFR	2021 ACC/AHA/SCAI Coronary Artery Revascularization Guideline
Acute coronary syndromes	2025 ACC/AHA/ACEP/NAEMSP/SCAI ACS Guideline
Pulmonary hypertension	2022 ESC/ERS Pulmonary Hypertension Guidelines
Congenital / shunts / anomalies	2025 ACC/AHA/HRS/ISACHD/SCAI Adult Congenital Heart Disease Guideline
Pericardial disease	2015 ESC Pericardial Diseases Guideline
Cardiogenic shock	SCAI SHOCK Stage Classification Expert Consensus Update (2022)
Hemodynamics & technique	Grossman & Baim’s Cardiac Catheterization, Angiography, and Intervention

Guideline vintages verified current as of the last editorial review. Where a domain has a newer focused update, the case citation names the specific document.

Version & review status

Content version 1.7 — 18 branching cases, cited to primary sources.

Last editorial self-review: July 2026. Guidelines current through 2025.

Faculty validation: pending — independent interventional-cardiology review is the next step before curricular adoption. Every clinical claim is source-cited to make that review straightforward.

Educational-outcome evidence: none yet. No claim of improved learning outcomes should be made for CathSim.

Not studied against knowledge gain, in-training-examination or board performance, or procedural readiness.
Defensible today: content coverage and instructional design — both auditable from this page.
Intended first evaluation: a pre/post knowledge assessment in a single fellowship.

Data handling: everything runs locally in the browser. Nothing is stored on a server, nothing is transmitted, and no progress survives closing the tab — which also means there is no longitudinal learner record and no way for a program to track completion. That is a deliberate privacy trade-off today and a roadmap item to revisit.

Change log

Version	Change
1.7	The Foundations course merged in, and one progress record across the whole platform. Full detail — five changes Twelve Foundations modules folded in. The standalone catheter-course and right-heart prototypes are now Unit 1 and Unit 2 of the app itself: one diagnostic left-heart case taken apart into the five stages it actually happens in, then a complete right-heart study. They sit above the instruments in the sidebar because that is the order they are meant to be worked in, and a new fellow can now start at module 0 and go straight through. One progress record. The course scored its ten checks into the address bar and showed the result nowhere; the app scored its nine into local storage and knew nothing about the course. The two are joined. Every Foundations check now counts toward the progress meter, the progress panel, the sidebar tick and this page, and a check passed in an earlier session is restored when the course opens. Redundancy removed rather than tolerated. Teaching the Foundations modules now do better was cut from the instruments instead of being left to contradict them: the old standalone Foundations page, the angiographic-views tab, the right-heart primer and transitions block, the waveform-trainer preamble, the VO₂ evidence table, and the access, landmark and radial-cocktail material in Procedure & Complications. One glossary. The app's terms and the course's were merged and de-duplicated, 67 entries to 86, so a term means the same thing wherever it is met. This page rewritten. It used to describe a curriculum running in parallel with the one the app taught. It is now the map of the actual path: the twelve modules in order, what each one leaves you holding, which instruments it opens, and the live state of every check.
1.6	Visual and orientational rebuild of the teaching graphics, plus a new Cath Simulator (beta) module folded in from the standalone catheter-course prototype. Full detail — seven changes Waveform trainer. The in-the-heart animation was redrawn and slowed so that the mechanism can be read while it plays. 3D projections, rebuilt around a room view. The C-arm, table and patient are drawn in one consistent body frame (x to the patient's left, y superior, z anterior), so LAO/RAO and cranial/caudal angulation are shown rather than asserted. Added live foreshortening and overlap meters, per-view hover explanations, independent zoom on the lab and detector panels, and a close anatomic view. The patient silhouette. Rebuilt to remove a bistable depth reading in which the table could be seen either beneath the patient or in front of them. The body is now a single binned silhouette with pelvis and thighs rather than stacked hulls; the table is occluded by the body when it lies behind and drawn as pale glass when it lies in front. The orientation compass marks each axis with a filled tip when it points toward the viewer and an open ring when it points away — the only cue that separates AP from PA when the axis lies along the line of sight. Label placement. Labels are now placed by searching an ink map of the finished frame, are barred from parking on the patient when they name something outside the body, and tie themselves back with a leader whenever they move far enough that the association would otherwise be ambiguous. Label memory. A station once chosen is re-scored each frame and kept unless something beats it by a clear margin; what is remembered is the offset from the anchor rather than a position on screen; and the drawn position is eased rather than teleported. Labels therefore no longer flicker between near-equal stations as the heart beats and the camera turns. Measured on a slow orbit: 98 station jumps down to 1. The PATIENT'S LEFT / RIGHT tags. They are no longer switched by the sign of the azimuth, which is correct but changes in a single frame and so read as the labels breaking. Each tag is now anchored to the patient's own flank and projected like everything else, so the pair crosses over continuously as the camera swings behind the patient. Each carries a leader to the side it names, neither may be placed across the midline, and they fade only through the few degrees where the two flanks genuinely superimpose. Cath Simulator (beta). Runs the case end to end: the catheter travels the real arterial course in the same body coordinate frame the projections module uses, and the gantry swings to the angle the operator would actually be in for that step. The pressure tracing is generated rather than drawn, so every number underneath is measured off that waveform — peak-to-peak, mean gradient, systolic ejection period, Gorlin and Hakki areas, resting Pd/Pa, FFR, CFR, and the site of the pressure loss on pullback. Four cases ship with it: a normal study and a severe aortic stenosis that cross the valve, and an intermediate mid-LAD and a severe proximal LAD that engage the left main and are interrogated with a pressure wire. The module parks its animation loop whenever another view is on screen.
1.5	Responses to an external content critique, plus a second coverage pass. Full detail — seven changes Pressure recovery was being tested without ever being taught. It is now defined in the glossary and taught in the Gorlin primer, the aortic-stenosis lesion text, and the aortic-stenosis case. Invasive coronary function added as a module — CFR, IMR in both the wedge-corrected and simplified forms, HMR, and the four CorMicA endotypes — so that INOCA is taught rather than assumed. Two multi-beat waveform sequences. HOCM with a premature beat showing Brockenbrough–Braunwald–Morrow, and atrial fibrillation with aortic stenosis showing why gradients must be averaged over ten beats. The library reached nine models. New static teaching. Systemic disease in the cath lab (carcinoid, sarcoidosis, systemic sclerosis, amyloidosis, high-output states, restriction versus constriction) and aortic regurgitation (chronic versus acute, premature mitral valve closure, the balloon-pump contraindication, the AR index). Clarifications and corrections. Clarified that three different numbers are all called “the gradient”; corrected the derivation of the mitral Gorlin constant; disclosed the parallel-projection limitation of the 3D module inside the module itself; and labelled assumed-Fick output as an estimate wherever it is used. Question bank grew from 55 to 81 items across 15 topics, weighted toward the higher-yield areas, with a new systemic disorders topic. Every new item tests something the platform now teaches. Every item re-audited for construction as well as content. Option-length cues were trimmed until the correct answer is the longest choice no more often than chance, and answer-leaking and duplicate-proposition distractors were rewritten. Two clinical errors were corrected in that pass: the pulmonary artery pulsatility index thresholds, which differ between right ventricular infarction and pre-LVAD assessment and had been blended into one number; and the murmur in the acute mitral regurgitation stem, which is characteristically soft rather than harsh.
1.4	Rebalancing of the question bank, with the supporting teaching written first. Full detail — three changes New question topics. Added items in arrhythmia and conduction (previously untested), heart failure and cardiomyopathy, and hypertension and vascular disease. Teaching first. Because no concept should be tested before it is taught, the supporting teaching was written first: a rhythm problems in the room card and post-TAVR conduction teaching in Trouble in the lab, and a new sixth tab, Beyond the coronaries, covering renal artery stenosis thresholds, atherosclerotic disease versus fibromuscular dysplasia, and coarctation, iliac and subclavian gradients. Navigation. Remediation links now open the specific tab that teaches the missed item rather than the module's front page. Fixed a defect in which the Procedure & Complications entry sat outside its sidebar group, so selecting it did not highlight the sidebar or update the page title.
1.3	Full-platform clinical audit, plus a rebuilt Board Room. Full detail — three changes Corrections. The Wiggers mitral-crossover timing and volume normalization, the HOCM aortic/LV relationship, the PCWP a/v relationship (v ≥ a), the RV dip-and-plateau, the RCA projection mnemonics, and the Gorlin flow-dependence explanation. The DPG threshold was retired in line with the 2022 ESC/ERS guideline and reclassified on PVR. Additions. The Procedure & Complications module, a congenital-lesion table, transducer-zeroing and end-expiration reading conditions, and pulmonary-hypertension group / CTEPH teaching. Board Room rebuilt with per-run option shuffling, missed-item tracking, and topic-level remediation links into the teaching modules.
1.2	Added arrhythmia (complete heart block & temporary pacing) and heart-failure (HFpEF with invasive exercise hemodynamics) cases — 18 total — plus an exportable session summary. Updated topic coverage.
1.1	Added per-case primary-source citations, guideline-source map, version stamp, and change log for auditability and faculty review.
1.0	Expanded to 16 branching cases across the major topics with data acquisition, artifact recognition, hand calipers, and generated procedure reports.

Angiography

Coronary anatomy, the angiographic views, and the physiology of a lesion — among the hardest concepts to hold in mind. Click around; everything is interactive.

3D Angiographic Projections BETA

A working catheterisation lab. On the left is the patient on the table with the C-arm at the angles you have dialled in; on the right is the picture that gantry actually produces.

The coronary tree is generated from the anatomy rather than drawn, so every segment foreshortens and overlaps for a reason you can see — and the meters underneath measure it live, in real time, for the exact angle you are on.

The lab — patient, table and C-arm drag to orbit

Red = coronary arteries. The patient lies face-up on the table, the tube under it and the detector above the chest — both swing with the angles you dial in. Click the heart for a close anatomic view.

What the detector seesAP

Simulated angiogram from the current gantry angles.

Views

RAO ← → LAO AP

CAUDAL ← → CRANIAL 0°

System

Show Patient & table C-arm & beam Vessel labels

Zoom room

Where the gantry is — the two diagrams every fellow draws

The naming rule — it is always about the detector, never the patient.

LAO · the detector swings to the patient's left. You end up looking at the heart from the patient's left side, so the spine drifts to the right of the picture and the heart looks larger.

RAO · the detector swings to the patient's right. You look from the right, the spine clears to the left, and the diaphragm rides up into the field.

CRANIAL · the detector tips toward the head. The beam runs feet-to-head, so you are looking down on the heart from above the head. It lifts the LAD off its branches.

CAUDAL · the detector tips toward the feet. You look up at the heart from below. It opens the left main and the circumflex and drops the diaphragm into the field.

The picture is always displayed as if you were standing where the detector is.

The detector rides above the patient; the X-ray tube sits under the table.
The patient is supine — so in AP you are above a face-up patient, looking down onto the front of the chest.
That is why the patient's left falls on the right of your screen.
The compass in the corner of each panel shows it for whatever angle you are on.

What this model does and does not represent

Generated, not drawn. The coronary tree is built in centimetres from the cardiac long axis, the atrioventricular ring, the anterior and posterior interventricular grooves, and the aortic sinuses. The way each segment foreshortens is a consequence of the anatomy rather than an artistic choice, and every percentage on this page is measured live from that geometry.

A model, not patient data. Coronary anatomy is highly variable; this is a right-dominant tree of average proportions.

Parallel projection, not a cone beam. It projects orthographically rather than as a beam diverging from a point source — a deliberate simplification with a specific consequence.

Reproduced faithfully	Not modelled
Foreshortening and vessel overlap, which depend on the direction of view. View selection is therefore taught correctly.	Beam divergence — so no source-to-image-distance magnification. Differential magnification between a near and a far vessel. Image-intensifier pincushion distortion. The geometry underlying scatter and skin dose. Real angulation limits, table height, patient habitus and breath-hold.

Treat it as a view-angle trainer, not a fluoroscopy simulator. Beta feature under development.

Structural Interventions — Decision Pathways

Structural work is selection work. The device usually behaves; the question is which patient it belongs in, and the answer is nearly always a measurement or a piece of anatomy.

By the end of this section you should be able to

Say why the same mitral clip saved lives in one trial and not the other
Localise a shunt from the level at which the oxygen saturation steps up
Site a transseptal puncture for the procedure that follows it
Read “adequate rims” as specific measurements, and name the one you cannot do without
Decide who gets alcohol septal ablation, and name the step that makes it safe
Quote a TAVR outcome knowing which dictionary defined it, and predict who needs a pacemaker

Landmark evidence — and the one idea that reconciles it

COAPT and MITRA-FR tested the same clip in secondary mitral regurgitation and disagreed completely. The reconciliation is proportionality: COAPT’s patients had roughly 30% more leak in ventricles roughly 30% smaller.

Reading the figure

In MITRA-FR the leak is about what that much dilatation would produce. The regurgitation is a symptom of the ventricle, and closing it changes nothing.

In COAPT the leak is larger than the ventricle explains. Now the regurgitation is its own problem, and clipping it is worth doing.

Clip the leak that is out of proportion to the ventricle. Otherwise the problem is the ventricle.

PARTNER · Evolut

Established TAVR across high-, intermediate- and low-surgical-risk severe aortic stenosis.

COAPT

TEER in selected secondary MR on maximal medical therapy: all-cause death at 24 months 29.1% vs 46.1% (HR 0.62, 95% CI 0.46–0.82) and heart-failure hospitalization 35.8% vs 67.9% per patient-year (HR 0.53, 0.40–0.70).

MITRA-FR

The same device, neutral — 54.6% vs 51.3% death or unplanned heart-failure hospitalization at one year (OR 1.16, p=0.53).

EVEREST II

Randomised surgical candidates, and surgery won on efficacy — 55% vs 73% at 12 months (p=0.007), with less complete MR reduction — while the device arm had far fewer 30-day major adverse events (15% vs 48%). The prohibitive-risk indication comes from the High Risk Study and REALISM registry, not the randomised trial.

VARC-3

Not a trial — the endpoint dictionary that makes two TAVR papers comparable at all. Worked through at the end of this section.

Going deeper — the entry criteria that produced the disagreement

MITRA-FR counted an effective regurgitant orifice above 20 mm² as severe, took an EF as low as 15%, and set no ceiling on LV size. COAPT used the higher American threshold for severe, EF 20–50%, an end-systolic dimension no greater than 70 mm, and maximal medical therapy first. Those numbers are what the phrase COAPT criteria means.

Proportionality is a reconciliation, not a proven mechanism. COAPT's own secondary analysis found that the COAPT patients who resembled MITRA-FR's still gained quality of life and walk distance — just not survival (Grayburn, JACC Cardiovasc Imaging 2019; Lindenfeld, JAMA Cardiol 2021).

Educational summary — decisions are individualized by the Heart Team using current guidelines.

Shunt lesions — and the oximetry run that finds them

Every left-to-right shunt does the same thing: it adds arterial blood to the right heart. Where the saturation jumps tells you where the hole is.

How the run works

Screen first. A rise of 8% or more from the SVC to the pulmonary artery is the trigger to sample properly, chamber by chamber.

Then localise. The first chamber whose saturation is above the one before it is the chamber the shunt enters. The thresholds differ because the mixing is better the further downstream you go.

Then quantify. The step-up gives you Qp:Qs, and Qp:Qs ≥ 1.5:1 with right-sided enlargement is the number that makes an ASD a Class 1 indication to close. It is 1.5, not 2.

Lesion	What decides management	The trap
Secundum ASD	Adequate septal rims to anchor a device, on transoesophageal or intracardiac echo, plus Qp:Qs ≥ 1.5:1 with right atrial and right ventricular enlargement.	Only secundum defects are device candidates. Primum (an AV canal variant, with a cleft mitral valve), sinus venosus (almost always with anomalous pulmonary venous drainage) and coronary sinus defects all go to surgery.
VSD	Shunt size and pulmonary vascular resistance.	A small restrictive VSD makes a loud murmur and little else. A large unrestricted one drives pulmonary vascular disease.
PDA	Step-up at the pulmonary artery, with a continuous murmur.	The only shunt whose step-up appears distal to the ventricles — which is exactly how you tell it from a VSD.
PFO	A right-to-left shunt on agitated saline, in a patient with a cryptogenic embolic stroke and high-risk anatomy (large shunt, atrial septal aneurysm).	A PFO is a flap, not a defect — present in about a quarter of adults. It is normally not a left-to-right shunt and does not enlarge the right heart, so it never appears on an oximetry run. Closure is for stroke prevention, not haemodynamics.
Eisenmenger syndrome	A long-standing left-to-right shunt whose pulmonary vascular resistance has risen until PVR exceeds SVR and the shunt reverses — cyanosis, clubbing, erythrocytosis.	Closure is contraindicated once the disease is fixed. The shunt has become the right ventricle's pop-off valve, and closing it causes acute RV failure. Vasoreactivity testing and pulmonary vasodilator therapy replace closure.

Aortic coarctation

Not a shunt — a pressure gradient in the aorta. Pull a catheter across the narrowed segment and record above and below: a peak-to-peak systolic gradient ≥ 20 mmHg (or less with extensive collaterals), especially with upper-extremity hypertension. Treat by balloon angioplasty and stenting or by surgery, and look for the associated bicuspid aortic valve and intracranial aneurysms.

Anomalous coronary origin

What matters is whether the vessel takes an interarterial course between the aorta and pulmonary artery, with a slit-like ostium, acute take-off and an intramural segment. The malignant pattern is an anomalous left main from the right sinus — a cause of exertional sudden death in young athletes. An anomalous right coronary from the left sinus is far more common and usually more benign; retroaortic, prepulmonic and septal courses are generally benign. CT angiography defines the course; intravascular imaging or FFR characterises the intramural segment.

Educational summary. Congenital decisions in adults follow current adult congenital heart disease guidelines and are made in specialist centres.

Transseptal puncture — the door to the left atrium

Mitral TEER, balloon mitral valvuloplasty, left atrial appendage occlusion and any direct left atrial pressure all start here. Where you cross determines how well the rest of the case goes.

The target

The fossa ovalis — thin, mid-posterior. Not the muscular limbus around it, and certainly not what lies beyond: the aortic root anteriorly, the free wall posteriorly.

Site it for the procedure ahead

Posterior and superior for mitral work, so the device has height above the annulus — 3.5 to 4 cm is what a clip delivery system needs. Posterior and inferior for the left atrial appendage, because the appendage points antero-superiorly and a low crossing is what lets the sheath turn up into it coaxially.

The kit

A Brockenbrough needle inside a Mullins sheath and dilator, dragged down from the SVC until the assembly steps off the limbus into the fossa.

Confirming the spot

Watch for tenting of the septum on echo in two orthogonal views: bicaval gives you superior versus inferior, short-axis at the aortic valve gives you anterior versus posterior. Fluoroscopy alone is not enough for a structural case.

Tamponade in roughly 1% — from puncturing the aorta, the atrial free wall or the coronary sinus. If the needle is in the wrong place, recognise it before you advance the dilator: a needle hole is survivable, a sheath hole often is not.

A naming trap worth defusing early: the Brockenbrough needle and the Brockenbrough–Braunwald–Morrow sign (the post-extrasystolic beat in hypertrophic obstructive cardiomyopathy) share a surname and nothing else.

The Wilkins score — who gets a balloon instead of a surgeon

Four features of the mitral apparatus on echo, each graded 1 to 4, total 4 to 16. 8 or less is favourable for percutaneous balloon mitral valvuloplasty.

Component	Grade 1 — best	Grade 4 — worst
Leaflet mobility	Only the leaflet tips are restricted	Almost no forward movement in diastole
Leaflet thickening	Near normal, 4–5 mm	Marked throughout, 8–10 mm
Subvalvular thickening	Minimal, just below the leaflets	Extensive chordal shortening to the papillary muscles
Calcification	A single bright area	Extensive brightness through much of the leaflet

A low score is necessary, not sufficient. Valvuloplasty is off the table whatever the score if there is a left atrial thrombus or more than mild mitral regurgitation. Heavy commissural calcium predicts a poor result even when the total looks acceptable — and the score never grades the commissures, which are precisely what the balloon splits.

“Adequate rims” — what the phrase actually specifies

Every account of ASD device closure turns on adequate rims, and the phrase is usually left to stand on its own. A rim is the collar of septal tissue between the edge of the defect and the structure next to it — the shelf the device sits on. Each margin is measured on transoesophageal or intracardiac echo, sweeping through the septum so that each is seen in turn.

The working numbers

About 5 mm or more at each margin is the convention operators use.

The aortic (anterosuperior) rim is the accepted exception — the one most often deficient, and the one implicated in device erosion. Deficiency here is not by itself a contraindication in experienced hands, but it changes the conversation and the follow-up.

The postero-inferior rim is the one you cannot do without. A deficient one is the usual reason a device will not stay where it was put.

A defect beyond roughly 38 mm stretched exceeds the largest septal occluder made, so very large defects go to surgery whatever the rims look like.

Erosion is the complication the rims are really about. In the manufacturer’s complication registry, 28 erosion events were reported and the United States incidence was 0.1% — rare, but it presents as tamponade days to years after an elective procedure. All of them occurred at the dome of the atria near the aortic root, and a deficient aortic rim was present in 89%. Practical consequence: a small pericardial effusion on the 24-hour study is not a shrug, it is a reason to follow the patient more closely.

Do not oversize. The temptation with a floppy rim is to reach for a bigger device. In the erosion registry the ratio of device size to unstretched defect size was significantly larger in the cases that eroded — and overstretching at balloon sizing is what produces the inflated number that justifies the bigger device.

Amin Z, et al. Erosion of the Amplatzer septal occluder after closure of secundum ASDs. Catheter Cardiovasc Interv 2004. The 5 mm and 38 mm figures are device-selection convention, not trial results.

PFO closure after cryptogenic stroke

A PFO is present in about a quarter of adults, so finding one after a stroke proves nothing on its own. Three randomised trials reported together in 2017 and settled the question for a narrow group: young patients, a stroke with no other cause after a full work-up, and an anatomy that makes the shunt plausible. Within that group closure reduces recurrent stroke; outside it, the trials are silent.

The size of the benefit

Recurrent stroke roughly halved to eliminated depending on how tightly the anatomy was selected: RESPECT 0.58 vs 1.07 per 100 patient-years (HR 0.55); REDUCE 1.4% vs 5.4% (HR 0.23); CLOSE zero strokes in 238 closure patients versus 14 of 235 (HR 0.03).

The trade you are making

Procedural risk of a few percent and new atrial fibrillation in roughly 1 in 20. Much of it is early and self-limited, but it has to be disclosed, and it has to be looked for.

What it is not for

Migraine, decompression illness in divers and platypnoea–orthodeoxia are separate indications with their own, much thinner, evidence. A PFO found incidentally is left alone.

Read the entry criteria, not the headline. CLOSE produced the most dramatic result in the set because it randomised only the anatomies most likely to be causal — an atrial septal aneurysm or a large shunt. REDUCE selected the same way, with 81% moderate or large shunts. That is why a small shunt in a 68-year-old with vascular risk factors is not what these trials studied, and why the decision belongs to a neurologist and a cardiologist together rather than to whoever ordered the bubble study.

The three trials in full

Trial	Who was randomised	Stroke result	What it cost
RESPECT long-term follow-up	980 patients aged 18–60 (mean 45.9) at 69 sites; median follow-up 5.9 years	Recurrent ischaemic stroke 0.58 vs 1.07 per 100 patient-years (HR 0.55, 95% CI 0.31–0.999, P=0.046).	Venous thromboembolism was more common in the closure group.
CLOSE	663 patients aged 16–60, randomised 1:1:1, entry restricted to a PFO with an atrial septal aneurysm or a large interatrial shunt; mean follow-up 5.3 years	Zero strokes among 238 closure patients versus 14 of 235 on antiplatelet therapy alone (HR 0.03, 95% CI 0–0.26, P<0.001)	Procedural complications 5.9%. Atrial fibrillation 4.6% vs 0.9% (P=0.02), though serious adverse events overall did not differ (P=0.56).
REDUCE	664 patients (mean age 45.2), randomised 2:1; 81% had moderate or large shunts; median follow-up 3.2 years	Clinical ischaemic stroke 1.4% (6/441) vs 5.4% (12/223), HR 0.23 (95% CI 0.09–0.62, P=0.002). New brain infarction 5.7% vs 11.3% (RR 0.51, P=0.04).	Serious device-related events 1.4%. Atrial fibrillation in 29 patients (6.6%) after closure.

Saver JL, et al. N Engl J Med 2017 (RESPECT) · Mas JL, et al. N Engl J Med 2017 (CLOSE) · Søndergaard L, et al. N Engl J Med 2017 (REDUCE). The RESPECT extended-follow-up analysis was explicitly exploratory.

Left atrial appendage occlusion — and an honest reading of its evidence

In non-valvular atrial fibrillation most left atrial thrombus forms in the appendage, so sealing it is an alternative to anticoagulating the whole patient. It is a transseptal procedure — the puncture is sited posterior and inferior, for the reasons above — with the landing zone sized on echo and the device released only once position, anchoring, size and seal have all been checked.

State this one carefully. It is often said that PROTECT AF and PREVAIL established non-inferiority. PREVAIL’s first coprimary endpoint did not meet it. What PREVAIL did establish, convincingly, is that the procedural hazard falls with operator experience — which is the honest case for the device and a fair thing to tell a patient. Note too that both comparators were warfarin, not a direct oral anticoagulant.

Who it is for

A patient with non-valvular atrial fibrillation and a stroke risk high enough to warrant anticoagulation, who has a durable reason not to take it — usually recurrent or life-threatening bleeding. Not simply a patient who would rather not.

The complication to fear

Pericardial effusion and tamponade. It is the dominant early event in both trials, it comes from the transseptal puncture or from the device engaging thin appendage tissue, and its rate is the number that fell as operators learned — early safety events 7.4 per 100 patient-years in PROTECT AF, 2.2% in PREVAIL.

Afterwards

Antithrombotic therapy continues until imaging confirms the appendage is sealed and the device endothelialised; a residual peri-device jet more than a few millimetres wide is the usual reason to carry on. Device-related thrombus is uncommon but is looked for on every follow-up study.

The two trials in full

Trial	Design	Efficacy	Safety
PROTECT AF 2009	707 patients randomised 2:1 — 463 device, 244 warfarin at INR 2.0–3.0; 1,065 patient-years	Primary efficacy event rate 3.0 vs 4.9 per 100 patient-years (RR 0.62)	Primary safety events were more frequent with the device — 7.4 vs 4.4 per 100 patient-years (RR 1.69, 1.01–3.19) — and mostly periprocedural
PREVAIL 2014	407 patients (269 device / 138 control), designed to retest the device in newer hands	The first coprimary efficacy endpoint did not meet non-inferiority; the second coprimary — stroke or systemic embolism more than 7 days after randomisation — did	Early safety events 2.2%. On the broader definition, 4.2% versus 8.7% in PROTECT AF (p=0.004). Effusion needing surgical repair fell 1.6% to 0.4% (p=0.027)

Holmes DR, et al. Lancet 2009 (PROTECT AF) · Holmes DR, et al. J Am Coll Cardiol 2014 (PREVAIL).

Alcohol septal ablation — a deliberate, targeted infarct

In obstructive hypertrophic cardiomyopathy the basal septum bulges into the outflow tract and the mitral leaflet is drawn against it. Ablation infarcts that bulge on purpose, by injecting ethanol into the septal perforator that supplies it. The septum thins and remodels over weeks to months, so the gradient falls further after the patient has gone home.

The five steps

1 · Measure the gradient — resting and provoked, LV to aorta.

2 · Wire and balloon the candidate perforator, occlude it, and check that the gradient falls.

3 · Prove the territory. Inject echo contrast down the balloon lumen. Myocardial contrast echocardiography answers the only question that matters at that moment: does this vessel supply the point of leaflet–septal contact, or the right ventricular free wall, a papillary muscle, or the distal septum?

4 · Ablate — 1 to 3 mL of ethanol, slowly, with pacing capability in place.

5 · Watch the conduction. A temporary wire stays in.

The contrast step is not optional flourish. In Faber’s original 91-patient series, selecting the target vessel by contrast echo rather than by balloon occlusion alone raised acute success from 70% to 92% (P<0.01) and mid-term success from 64% to 94% (P<0.01). Same drug, same catheter, better aim.

Who qualifies

All four gates, not any one: severe symptoms — typically NYHA III–IV or exertional syncope — on maximally tolerated drug therapy; a dynamic LVOT gradient of 50 mmHg or more at rest or with provocation; a septal perforator that actually supplies the point of leaflet–septal contact, in a septum thick enough to take a controlled infarct; and a comprehensive centre doing enough of these to have a result.

Who does not

An asymptomatic gradient, however impressive. A mid-cavity or apical variant — a different problem. Obstruction driven by an anomalous papillary muscle or intrinsic mitral disease — a surgical problem. Too thin a septum, which risks a ventricular septal defect.

What it achieves

In Faber’s series the resting gradient fell from 73.8 to 16.6 mmHg and the provoked gradient from 149.3 to 61.9 mmHg (P<0.0001 each), with mean NYHA class falling from 2.8 to 1.1 by three months.

Versus myectomy

A meta-analysis of 12 observational studies — there are no randomised trials — found no significant difference in short- or long-term mortality, symptom improvement, ventricular arrhythmia, re-intervention or residual mitral regurgitation. Ablation left a small but significantly higher residual gradient.

What it costs

Conduction. Ablation raised right bundle branch block (pooled OR 56.3) and permanent pacemaker implantation (pooled OR 2.6) relative to myectomy; 10 of Faber’s 91 patients (11%) needed a pacemaker. Surgery does the mirror image and produces left bundle branch block.

Faber L, et al. Circulation 1998 · Agarwal S, et al. J Am Coll Cardiol 2010. Symptom, gradient and centre thresholds are guideline definitions (2020 AHA/ACC hypertrophic cardiomyopathy guideline), not trial endpoints. The conduction consequences are worked through in Procedure & Complications.

TAVR: how the outcomes are counted, and the pacemaker question

Two things separate a fellow who can quote TAVR trials from one who can read them. The first is knowing that the numbers come from a shared dictionary. The second is knowing the one complication the patient will ask about.

Check the dictionary before you quote the number. A major vascular complication rate of 4% and one of 8% may describe identical practice measured against different VARC generations. The Valve Academic Research Consortium has standardised transcatheter valve endpoints since 2010; the third iteration, published in 2021, revised bleeding, access-site complications, repeat hospitalisation, conduction disturbance, cardiac structural complications and bioprosthetic valve dysfunction, and replaced the three-grade paravalvular regurgitation scale with five.

The valve sits directly on the membranous septum, where the conduction system runs. A meta-analysis of 41 studies and 11,210 patients found that 17% required a permanent pacemaker, with a range across studies of 2% to 51% — a spread that is itself an argument for standardised definitions.

Predictor	Relative risk	Reading it
Intraprocedural AV block	3.49	The strongest signal, and the reason the temporary wire stays in.
Pre-existing right bundle branch block	2.89	The strongest baseline predictor. The valve injures the left bundle; if the right is already blocked, what is left is complete heart block.
Left anterior hemiblock	1.62	Same logic, weaker.
First-degree AV block	1.52	Conduction already marginal.
Male sex	1.23	Consistent, small.
Self-expanding vs balloon-expandable	2.5-fold (unadjusted)	Median pacemaker rate 28% for the self-expanding CoreValve versus 6% for the balloon-expandable SAPIEN in that analysis — a device choice with a consequence.

The mirror image worth carrying. In TAVR, right bundle branch block is the baseline risk factor that predicts a pacemaker. In alcohol septal ablation, right bundle branch block is the expected result. Same bundle, opposite end of the causal arrow — and it is a favourite examination hinge.

Généreux P, et al. Valve Academic Research Consortium 3. Eur Heart J 2021 · Siontis GCM, et al. J Am Coll Cardiol 2014. Device-specific rates predate the current valve generations and are quoted as reported.

Intravascular Imaging — IVUS & OCT

A camera inside the artery — what the angiogram cannot show you, and what to do about it.

By the end of this section you should be able to

Orient yourself on a frame — lumen, EEM, artefacts — and say what each modality cannot show you
Name the plaque from its signature, and tell calcium from lipid by its borders
Size a stent off a pullback: diameter, length, landing zones
Decide when calcium needs preparing, and what to reach for
Judge a stent result against the numbers — expansion, apposition, edges
Work out why a stent failed from when it failed

▸ The words first new to this? start here

Three terms carry most of it.

Term	What it is	Why it matters
Lumen	The open channel the blood runs through	Its cross-sectional area, in mm², is the number every threshold on this page is written in
EEM external elastic membrane	The outer border of the vessel, at the boundary between the muscular media and the adventitia	The outer edge you measure to. Lumen and EEM together give you the disease
Plaque burden	(EEM area − lumen area) ÷ EEM area	How much of the vessel is disease. An angiographically normal segment still runs about 40 per cent

Using the tool below: pick a lesion, switch between IVUS and OCT on the same one, and drag along the long view to travel down the vessel. The numbers redraw as you move.

An angiogram is a silhouette: contrast fills the lumen and you infer the disease from the shape of the shadow. So a vessel packed with plaque can look almost normal, and a vessel that looks tight may not be. Intravascular imaging puts the camera inside the artery, on the wire, and returns a cross-section of the wall itself — an ultrasound one (IVUS) or a light one (OCT). For complex lesions it now carries a Class 1 recommendation.

Measurements

Drag along the long view — distal left, proximal right — to move the cross-section down the vessel, or press Play pullback. Schematic teaching models drawn to a true millimetre scale, not clinical images.

How to read a frame

Where you are

The catheter is at the centre and the image is built outward from it, so you read the frame like a clock face and report findings as arcs in degrees. The catheter lies in the lumen, so eccentric plaque makes the EEM look off-centre, not the lumen.

The three layers

Bright intima, dark media, bright adventitia. The EEM is the media–adventitia border, and every area measurement runs to it. Find it first.

What is hidden, and why

Behind calcium on IVUS: nothing. Sound reflects off the surface — bright leading edge, total dropout, no EEM. Arc measurable, thickness never. Past about 1.5 mm on OCT: nothing. Light is absorbed, so the EEM shows only where the wall is thin, and not at all behind lipid. Light does pass through calcium, which is why OCT alone measures its thickness.

Artefacts, not disease

The guidewire casts a dark stripe on both, every strut casts one on OCT, and IVUS shows a ring-down halo around the catheter. Exclude all three.

Which view gives which number

Areas, arcs and cap thickness come from the cross-section. Lesion length, stent length and where the edges land come only from the pullback.

So which one do you run?

Neither is the better scan. They fail in opposite directions, and that is the whole basis for choosing: ultrasound loses detail, light loses depth.

	IVUS (ultrasound)	OCT (light)
Resolution	~100–150 µm	~10–20 µm — roughly ten times finer
Blood clearing	Not required	Needs a contrast flush; light is blocked by blood
Reach for it for	Vessel sizing, the left main and aorto-ostial lesions, and any patient in whom contrast is the enemy	Thin caps, dissection, thrombus, strut apposition, bifurcations — anywhere detail near the lumen decides the next move

Four signatures — what the plaque is made of

Four patterns, told apart by two features: whether the borders are sharp (calcium) or diffuse (lipid), and whether anything is hidden behind. Get that wrong and you take the wrong strategy — one lesion needs preparing, the other needs its landing zone moved.

Plaque	On IVUS	On OCT	What it changes
Fibrous	Bright, homogeneous, nothing shadowed behind it	Signal-rich and uniform — you see through to the EEM	Nothing. Stent it as it is
Calcium	Bright leading edge with an acoustic shadow behind. Arc measurable, thickness never	Signal-poor with sharp borders on every side, so the far border shows and thickness is measurable	Everything. It decides whether you prepare the lesion, and how hard
Lipid / necrotic core	Echolucent and attenuating, no defined outer border	Signal-poor with diffuse borders and strong attenuation; the cap over it is measurable	Cap under 65 µm over an arc over 90° is a thin-cap fibroatheroma — the rupture-prone phenotype
Thrombus	A mobile or layered low-echoic mass, hard to tell from soft plaque	Red (cell-rich): highly backscattering, shadow behind. White (platelet-rich): signal-rich, barely any shadow	Names the mechanism in an acute presentation

A workflow you can run every time — MLD MAX

The commonest reason a pullback changes nothing is that nobody decided in advance what to look for. MLD before the stent, MAX after it — same six questions, same order, every case.

Run	What you read	What it decides
M — Morphology	What the plaque is made of, and how much calcium	Whether the lesion is prepared before the stent, and with what
L — Length	Reference to reference on the long view, not the angiogram	Stent length, and where the edges will land
D — Diameter	Distal and proximal reference lumen, and the EEM at each	Stent diameter — size to the distal reference and you rarely oversize
M — Medial dissection	The edges: is there a flap, and how deep	Whether an edge needs covering before you leave
A — Apposition	Struts on the wall, or standing off it	Whether to post-dilate the segment that is not touching
X — eXpansion	Minimal stent area, and that against the reference	The one worth going back for. Everything else is secondary

Numbers to hit

A pullback only earns its time if you act on it.

Target	Number to hit
Minimal stent area — outside the left main	≥ 5.5 mm² by IVUS · ≥ 4.5 mm² by OCT
Expansion index (MSA ÷ reference lumen area)	over 80%
Stent-edge plaque burden	under 50% — land the edge in normal-looking vessel
Edge dissection	Treat it if it reaches into or beyond the media, spans over 60°, runs over 2 mm, carries an intramural haematoma, or limits flow. A superficial flap with none of those heals — leave it
Malapposition	Correct a gap over 0.4 mm across a long segment; small acute gaps usually heal

Left main — segmental, because the vessel tapers

Segment	Minimal stent area
Left main proper	≥ 8.2 mm²
Polygon of confluence	≥ 7.2 mm²
Ostial LAD	≥ 6.3 mm²
Ostial circumflex	≥ 5.0 mm²

The IVUS cut-offs that best predicted restenosis in 403 patients with unprotected left main disease — segment by segment, because one number cannot describe a vessel that steps down twice.

Minimal lumen area is not a stenosis test. An MLA under 6.0 mm² in the left main, or under 4.0 mm² in an epicardial vessel, is the classic imaging cut-off — but where you can interrogate a vessel physiologically, physiology decides whether to treat and imaging decides how. Reach for MLA when FFR or iFR is unavailable or unreliable, and in the left main, where physiology is technically hardest.

Calcium — the pullback says prepare, not atherectomy

Calcium is the finding that most reliably changes the plan — and each modality scores it its own way.

Modality	The red flag
IVUS	A superficial arc over 270° running 5 mm or more — one of four elements of the IVUS calcium score, with circumferential 360° calcium, a calcified nodule, and a vessel under 3.5 mm
OCT	A calcium score of 4 out of 4: arc over 180° scores 2, thickness over 0.5 mm scores 1, length over 5 mm scores 1. Only the maximum marks a lesion that expands badly — 78 against 96 per cent

What the evidence supports is balloon first — non-compliant, scoring or cutting, at high pressure — escalating only for the lesion the balloon cannot cross or cannot dilate. Routine upfront atherectomy has been randomised twice against exactly that strategy and beat it neither time. It is not a better opening move; it is the answer to a specific failure. What imaging changes is that you knew to prepare at all, and that you can see whether the preparation worked before the stent goes in rather than nine months later.

The three trials, one line each · 2,676 patients

Trial	What it found	Why it matters
ROTAXUS 2013 · n=240	Routine rotational atherectomy before a paclitaxel-eluting stent gave a larger acute gain but worse late lumen loss at 9 months — 0.44 against 0.31 mm. Restenosis, target-lesion revascularisation and major adverse events were all the same. Crossover ran the other way: 12.5% of the balloon arm needed the burr against 4.2%, so strategy success favoured rotablation, 92.5% against 83.3%	The trial's own conclusion was balloon first, rotablation on demand — the burr rescues, it does not improve
ECLIPSE 2025 · n=2,005	Routine orbital atherectomy against a balloon-based strategy, in lesions suitable for either. Neither co-primary endpoint favoured atherectomy: target-vessel failure at one year 11.5% against 10.0% (p=0.28), and minimal stent area at the site of maximal calcium 7.67 against 7.42 mm² (p=0.078)	Confirmed ROTAXUS in a trial eight times the size, with imaging itself as the referee
Disrupt CAD III 2020 · n=431	Intravascular lithotripsy, single-arm against a pre-specified performance goal. Procedural success 92.4% and 30-day freedom from major adverse events 92.2%, both above goal; OCT showed calcium fracture in 67.4%. Mean calcium arc was 292° and thickness 0.96 mm — these were not borderline lesions	Approved the device. It did not beat a comparator, because it never had one

Why did this stent fail? Timing names the mechanism

Ask when before you ask what to do. The interval since implantation narrows the differential faster than any single frame.

When it failed	What it usually is	What the pullback shows
Acute — first 24 h Subacute — out to 30 days	Mechanical, nearly always — under-expansion, an untreated edge dissection, a landing zone left in plaque, tissue or thrombus between struts. The rest of the differential is the antiplatelet therapy: stopped, never absorbed, or never working	Small stent area, a flap at the edge, plaque burden over 50% where the stent ends
Late — 30 days to 1 year	Delayed healing — uncovered struts, unresolved malapposition, early neointimal disease. Interrupted dual antiplatelet therapy still matters	Struts with no neointima over them, a persisting strut-to-wall gap
Very late — beyond 1 year	Neoatherosclerosis, or late acquired malapposition from positive remodelling or a thrombus resolving behind the struts	Lipid-laden or calcified neointima, sometimes a ruptured cap; struts standing off a wall they used to touch

Restenosis has phenotypes too, and each wants a different tool

Under-expanded stent

The commonest reason a stent restenoses, and the one another layer of metal will not fix. High-pressure non-compliant balloon, or lithotripsy if calcium is holding it.

Hyperplastic neointima

A well-expanded stent filled with homogeneous tissue. Drug-coated balloon, or a second drug-eluting stent.

Neoatherosclerosis

New atherosclerosis growing inside the neointima years out, and it ruptures exactly as native plaque does. Treat it — and the risk factors that grew it — as fresh disease.

Edge disease or geographic miss

Disease at or just past the stent margin. Extend the coverage; re-dilating what is already stented does nothing.

This is the argument for imaging before you re-treat a failed stent. Four mechanisms, four different answers, and an angiogram that shows the narrowing but never the reason for it.

Does it change outcomes?

Imaging improves outcomes by changing the stent, not the diagnosis. Every positive trial worked through one mechanism — a larger, better-expanded stent landed in a healthier segment — so the benefit is biggest exactly where that is hardest: long lesions, bifurcations, chronic occlusions, the left main. It is also why ILUMIEN IV improved the stent without moving its two-year clinical endpoint. On the strength of these trials, imaging-guided PCI of complex lesions now carries a Class 1 recommendation.

The five trials, one line each

Trial	What it found	Why it matters
IVUS-XPL 2015 · long lesions, stent ≥ 28 mm	Fewer major adverse cardiac events at 1 year with IVUS guidance	The separation was still there at 5 years — a procedural decision with a durable consequence
ULTIMATE 2018 · all-comers	Lower target-vessel failure at 1 year, and again at 3	Showed the benefit is not confined to a selected lesion subset
RENOVATE-COMPLEX-PCI 2023 · complex lesions	Target-vessel failure 7.7% against 12.3% over a median 2.1 years	The clearest modern demonstration that complexity is where the benefit lives
ILUMIEN IV 2023 · high-risk PCI or complex lesions	Larger minimal stent area and fewer peri-procedural events — but the 2-year target-vessel failure endpoint was not significantly reduced	The honest counterweight: a better stent does not automatically become a better outcome in every population
OCTOBER 2023 · bifurcation lesions	Fewer target-lesion events at 2 years with OCT guidance	Bifurcations are where the extra detail earns its cost — carina, re-wiring, the link-free strut after kissing inflation

Sources — eighteen papers, one line each · every row links to its DOI

Study	Citation
Shlofmitz E, Croce K, Bezerra H, et al. The MLD MAX OCT algorithm: An imaging-based workflow for percutaneous coronary intervention	Catheter Cardiovasc Interv 2022;100(Suppl 1):S7-S13	DOI
Räber L, Mintz GS, Koskinas KC, et al. Clinical use of intracoronary imaging. Part 1: guidance and optimization of coronary interventions. An expert consensus document of the European Association of Percutaneous Cardiovascular Interventions	Eur Heart J 2018;39(35):3281-300	DOI
Johnson TW, Räber L, di Mario C, et al. Clinical use of intracoronary imaging. Part 2: acute coronary syndromes, ambiguous coronary angiography findings, and guiding interventional decision-making	Eur Heart J 2019;40(31):2566-84	DOI
Rao SV, O’Donoghue ML, Ruel M, et al. 2025 ACC/AHA/ACEP/NAEMSP/SCAI Guideline for the Management of Patients With Acute Coronary SyndromesWhere the Class 1 recommendation for imaging-guided PCI comes from	Circulation 2025;151(13):e771-e862	DOI
Fujino A, Mintz GS, Matsumura M, et al. A new optical coherence tomography-based calcium scoring system to predict stent underexpansionThe 4-point OCT calcium score	EuroIntervention 2018;13(18):e2182-e2189	DOI
Zhang M, Matsumura M, Usui E, et al. Intravascular Ultrasound-Derived Calcium Score to Predict Stent Expansion in Severely Calcified LesionsWhere the 270° arc over 5 mm comes from	Circ Cardiovasc Interv 2021;14(10):e010296	DOI
Fujii K, Carlier SG, Mintz GS, et al. Stent underexpansion and residual reference segment stenosis are related to stent thrombosis after sirolimus-eluting stent implantation: an intravascular ultrasound study	J Am Coll Cardiol 2005;45(7):995-8	DOI
Kang SJ, Ahn JM, Song H, et al. Comprehensive intravascular ultrasound assessment of stent area and its impact on restenosis and adverse cardiac events in 403 patients with unprotected left main diseaseThe four left main segmental cut-offs	Circ Cardiovasc Interv 2011;4(6):562-9	DOI
Hong SJ, Kim BK, Shin DH, et al. Effect of Intravascular Ultrasound-Guided vs Angiography-Guided Everolimus-Eluting Stent Implantation: The IVUS-XPL Randomized Clinical Trial	JAMA 2015;314(20):2155-63	DOI
Hong SJ, Mintz GS, Ahn CM, et al. Effect of Intravascular Ultrasound-Guided Drug-Eluting Stent Implantation: 5-Year Follow-Up of the IVUS-XPL Randomized Trial	JACC Cardiovasc Interv 2020;13(1):62-71	DOI
Zhang J, Gao X, Kan J, et al. Intravascular Ultrasound Versus Angiography-Guided Drug-Eluting Stent Implantation: The ULTIMATE Trial	J Am Coll Cardiol 2018;72(24):3126-37	DOI
Gao XF, Ge Z, Kong XQ, et al. 3-Year Outcomes of the ULTIMATE Trial Comparing Intravascular Ultrasound Versus Angiography-Guided Drug-Eluting Stent Implantation	JACC Cardiovasc Interv 2021;14(3):247-57	DOI
Lee JM, Choi KH, Song YB, et al. Intravascular Imaging-Guided or Angiography-Guided Complex PCIRENOVATE-COMPLEX-PCI	N Engl J Med 2023;388(18):1668-79	DOI
Ali ZA, Landmesser U, Maehara A, et al. Optical Coherence Tomography-Guided versus Angiography-Guided PCIILUMIEN IV	N Engl J Med 2023;389(16):1466-76	DOI
Holm NR, Andreasen LD, Neghabat O, et al. OCT or Angiography Guidance for PCI in Complex Bifurcation LesionsOCTOBER	N Engl J Med 2023;389(16):1477-87	DOI
Abdel-Wahab M, Richardt G, Joachim Büttner H, et al. High-speed rotational atherectomy before paclitaxel-eluting stent implantation in complex calcified coronary lesions: the randomized ROTAXUS trial	JACC Cardiovasc Interv 2013;6(1):10-9	DOI
Kirtane AJ, Généreux P, Lewis B, et al. Orbital atherectomy versus balloon angioplasty before drug-eluting stent implantation in severely calcified lesions eligible for both treatment strategies (ECLIPSE): a multicentre, open-label, randomised trial	Lancet 2025;405(10486):1240-51	DOI
Hill JM, Kereiakes DJ, Shlofmitz RA, et al. Intravascular Lithotripsy for Treatment of Severely Calcified Coronary Artery DiseaseDisrupt CAD III	J Am Coll Cardiol 2020;76(22):2635-46	DOI

Pressure–Volume Loop Lab

The single most powerful picture in hemodynamics. Drag the sliders and watch the loop change in real time.

▸ Concept primer new to this? start here

A pressure–volume loop plots left-ventricular pressure against its volume across one beat, so a single closed loop captures the whole mechanical performance of the ventricle. Read it counter-clockwise: filling along the bottom, isovolumic contraction up the right side, ejection across the top, isovolumic relaxation down the left.

Three properties set its shape. Preload is how full the ventricle is before it contracts (the right-hand end-diastolic corner). Afterload is the pressure it must overcome to eject (roughly the loop's height). Contractility is the intrinsic squeeze, captured by the slope of the end-systolic pressure–volume relationship. The width of the loop is the stroke volume; its area is the stroke work.

Every disease state moves the loop in a recognizable way — tall and narrow with high afterload, wide and short in a failing ventricle, small and left-shifted when preload is low.

Using the tool: drag the preload, afterload, and contractility sliders and watch the loop reshape live, with stroke volume, ejection fraction, and stroke work updating.

Pressure–Volume loop

Loop ESPVR (contractility) EDPVR (stiffness) Ea (afterload)

Pressure vs time — LV & aorta (2 beats)

LV pressure Aortic pressure ● marker = same instant on both

Ventricle & valves

End-diastole

aortic valve mitral valve

Tip: drag across the loop or the tracing to scrub through the beat by hand — the marker and ventricle follow.

Presets

Physiologic model (time-varying elastance, Suga–Sagawa). Teaching approximation, not a monitoring device.

The Cardiac Cycle — Wiggers Diagram

Every pressure, the ventricular volume, the ECG, and the heart sounds — one heartbeat, all moving in sync. Watch the sweep line cross the cycle and see exactly when each valve opens and closes.

▸ Concept primer new to this? start here

The cardiac cycle is the sequence of electrical, pressure, and volume events in a single heartbeat. The Wiggers diagram stacks them on one time axis so you can see how they line up — and once you know it, almost every hemodynamic finding becomes "a departure from this."

Two labels to carry in: on the ECG the P wave is atrial contraction and the QRS is ventricular contraction; the heart sounds mark the valve closures, S1 at mitral closure (the start of systole) and S2 at aortic closure (the end of systole).

Using the tool: play the cycle and watch the sweep line — the pressures, the LV volume, the ECG, and the valve states all move together, so you can see exactly when each valve opens and closes. The seven phases are named and defined in the table below the diagram.

Beating heart · synced to the tracing

—

oxygenated (left heart) deoxygenated (right heart)

Drag the tracing to move the heart at your own pace — the heart stays in view as you scrub.

LA pressure drag the graph to scrub through the beat

The phases of one heartbeat

1 · Atrial systole

The P wave fires; the atria contract and top off the ventricle (the "atrial kick" → the a wave and end-diastolic volume). Ends when the mitral valve closes (S1).

2 · Isovolumic contraction

All valves shut; the ventricle contracts against a fixed volume, so pressure rises steeply with no change in volume.

3 · Rapid ejection

LV pressure exceeds aortic → the aortic valve opens and blood is ejected; volume falls fast as aortic and LV pressures rise to their peak.

4 · Reduced ejection

Ejection slows; pressures begin to fall as the ventricle repolarizes (T wave).

5 · Isovolumic relaxation

The aortic valve closes (S2, the dicrotic notch); all valves shut again and pressure falls with no volume change.

6 · Rapid filling

LV pressure drops below LA → the mitral valve opens and the ventricle fills quickly (the y descent).

7 · Reduced filling (diastasis)

Filling slows as the ventricle approaches its resting volume, until the next atrial kick.

Right Heart Catheterization Simulator

Float a Swan–Ganz catheter from the right atrium to the wedge and watch the pressure waveform transform at each step — the exact transitions you use to know where your catheter tip is without looking at fluoro.

Right atrium

Respiratory Dynamics — Interdependence & Pulsus Paradoxus

Static numbers can't teach this. Watch simultaneous LV and RV pressures breathe, and see how inspiration separates constriction, restriction, and tamponade in real time.

▸ Concept primer new to this? start here

Breathing changes the pressures inside the heart, and the pattern of that change is diagnostic. Ventricular interdependence is the reason: the two ventricles share a septum and, in some diseases, a fixed space — so when one fills more, the other must fill less.

Disease exaggerates the normal respiratory effect in specific ways, and the pattern names the diagnosis.

Condition	What inspiration does	The name for it
Normal	Slightly increases right-heart filling and slightly decreases left-heart filling.	A small effect.
Constrictive pericarditis	A rigid pericardium forces the ventricles to compete, so RV pressure rises while LV pressure falls.	Discordance
Cardiac tamponade	The same competition produces an exaggerated inspiratory fall in systemic pressure.	Pulsus paradoxus — over 10 mmHg supports tamponade (LR 3.3, 95% CI 1.8–6.3), but sensitivity is only ~82%, and asthma, COPD, PE and hypovolaemia produce it too
Restrictive cardiomyopathy	With no pericardial constraint, the two pressures fall together.	Concordance — this is how you separate it from constriction

Using the tool: switch between normal, constriction, restriction, and tamponade and watch the simultaneous LV/RV tracings breathe — the discordance, the pulsus, and the blunted or steep descents appear in real time.

LV RV inspiration systolic peaks

How inspiration separates the three

Constriction

Discordance — RV systolic ↑ while LV systolic ↓ on inspiration (ventricular interdependence through a rigid pericardium). Steep y, equalized diastolic pressures, and a Kussmaul sign.

Restriction

Concordance — LV and RV systolic pressures fall together on inspiration (no pericardial constraint). LVEDP > RVEDP; higher PA systolic.

Tamponade

Exaggerated pulsus paradoxus — a >10 mmHg inspiratory fall in systolic pressure as the RV bows into the LV; blunted y; equalized diastoles.

Case Simulator

Pick a case. You'll move through it decision by decision — the data reveals itself as you'd collect it in the lab.

▸ Faculty · build a custom case data-driven authoring

CathSim cases are plain data. Paste a case in the schema below and it becomes a fully playable branching case — a way for your division to author its own teaching cases without touching code.

Where the case lives.

No server. A case you load plays in this browser session and is gone when you close the tab.
The JSON text itself is the portable artifact. Save it to a file and circulate it, or send it to be folded into the built-in library.
Nothing is retained. Nothing you paste here is stored, uploaded, or transmitted anywhere.
Not yet. A shared case repository with versioning and attribution is a roadmap item, not a current feature.

Hemodynamics Lab

The calculations you'll be asked to do on the boards and in the lab, live. Change any input and the interpretation updates.

By the end of this section you should be able to

Calculate cardiac output by Fick, and say when it is least trustworthy
Size a shunt, and localise it from a saturation run
Read a valve area knowing the four ways Gorlin misleads you
Separate pre- from post-capillary pulmonary hypertension on the PVR
Stage a shock patient, and say which ventricle is failing

Pick a calculation below — each opens with a short primer, then works live as you change the inputs.

▸ Systemic disease in the cath lab carcinoid · sarcoid · scleroderma · amyloid · high-output

A handful of multisystem illnesses have hemodynamic signatures specific enough that the diagnosis is suspected from the numbers on the table before anyone orders a biopsy. The clue is always the pattern, never a single value.

What is on the table	What to think
Right-sided regurgitant lesions with liver metastases	Carcinoid
High-grade AV block under 60 with clean coronaries	Sarcoid
Thick heart, restrictive filling, low ECG voltage, and no hypertension or aortic stenosis to explain it	Amyloid
Pulmonary hypertension with Raynaud and sclerodactyly	Systemic sclerosis — confirm pre-capillary before treating
High output with a high SvO₂	A vasodilated or shunted circulation, not a failing pump

Restriction versus constriction — the cath discriminator is ventricular interdependence. Both give elevated, equalised diastolic pressures and a square-root sign. In constriction the rigid pericardium makes the two ventricles compete for one fixed volume, so with inspiration RV systolic pressure rises while LV systolic pressure falls — discordant peaks. In restriction both feel the same drop in intrathoracic pressure, so the peaks stay concordant. Quantified, a systolic area index above 1.1 identified constriction with 97% sensitivity. Supporting signs — RVEDP over a third of RV systolic, RV systolic under 50 mmHg, LVEDP and RVEDP within 5 mmHg — are nudges, not verdicts: a quarter of patients cannot be classified on haemodynamics at all.

The five patterns — why each looks the way it does

Carcinoid. Midgut tumours release serotonin into the portal circulation, where the liver inactivates it — so the heart is spared until hepatic metastases drain mediators straight into the IVC. The lung then clears them, which is why disease is right-sided: fibrous deposits tether the tricuspid and pulmonic leaflets open, and severe TR fuses the c and v waves into one tall systolic wave with no x descent, a ventricularized right atrial tracing. Left-sided disease is the exception, and should send you looking for a PFO or a bronchial primary that bypasses the pulmonary filter.

Sarcoid. Non-caseating granulomas seed the basal septum, exactly where the conduction system runs. In adults under 60 with unexplained second-degree or higher block, sarcoid is the cause in up to a third — making it an MRI-and-PET workup, not just a pacemaker. Because the other face is scar-mediated ventricular tachycardia, the device decision is not a simple one either.

Amyloid. Fibrils stiffen the interstitium and give restrictive filling — a deep sharp y descent and a dip-and-plateau contour. Macroglossia and periorbital purpura point to AL. Type it before you treat it: screen with a serum free light chain assay plus serum and urine immunofixation, and all three negative effectively excludes AL. Bone scintigraphy identifies ATTR only after a plasma cell dyscrasia is excluded, because AL takes up tracer too. Typing decides everything: ATTR has disease-modifying therapy — tafamidis, acoramidis, vutrisiran — while AL needs urgent plasma-cell-directed treatment.

Systemic sclerosis. Echo cannot separate the three possibilities — pre-capillary (Group 1 pulmonary arterial hypertension), post-capillary (Group 2, myocardial fibrosis) or lung-driven (Group 3, interstitial disease); the cath is where the ambiguity is resolved. Confirming PAH takes all three of: mean PA above 20 mmHg, PCWP 15 or less, and PVR above 2 Wood units. If the wedge is above 15 the problem is on the left, and pulmonary vasodilators can make it worse by driving flow into a circuit that cannot drain.

High output. Above roughly 8 L/min with a low SVR and a high mixed venous saturation, the tissues are not extracting — flow is excessive, or shunted past them. In low-output failure SvO₂ is low, because wider extraction is compensating. Sustained long enough, high output causes heart failure on its own. Think: severe anaemia, sepsis, thyrotoxicosis, an arteriovenous fistula including a dialysis access, advanced liver disease, Paget disease, thiamine deficiency. At the other end of the spectrum, iron overload gives an early restrictive phenotype that later dilates — treatable by phlebotomy or chelation if caught before fibrosis.

▸ Aortic regurgitation in the cath lab chronic vs acute · pulse pressure · AR index

Aortic regurgitation is the lesion where the tracing tells you the time course, not just the severity — and the one where a reassuring-looking number can mean the patient is about to die.

	Chronic severe AR	Acute severe AR
Cause	Regurgitant volume accumulates over years	Endocarditis, a dissection into the root, or a failed prosthesis
The ventricle	Dilates eccentrically, and its compliance rises with it	Normal-sized and non-compliant — no time to remodel
Filling pressure	Accepts an enormous volume at a modestly elevated pressure	LVEDP rises steeply — it can reach 40 mmHg or more
Pulse pressure	Wide, often 100 mmHg or more — high systolic from the large stroke volume, low diastolic from runoff back into the ventricle	Narrow or normal, because aortic diastolic pressure falls as LV diastolic pressure climbs until the two equilibrate before end-diastole
The murmur	Long	Short and soft — the gradient driving it has gone by mid-diastole. Tachycardia is the only compensation left

Every classic chronic sign is a consequence of that wide pulse pressure: the water-hammer (Corrigan) pulse, head bobbing (de Musset), the bisferiens pulse, pistol-shot femoral sounds (Traube). Invasively, chronic AR is graded 1+ to 4+ on root angiography, by how densely and persistently contrast opacifies the LV.

The acute trap: a patient in cardiogenic shock with a normal pulse pressure and an unimpressive murmur. The reassuring exam is itself the sign of decompensation. Two echo findings say the ventricle is at its limit and argue for emergency surgery rather than observation — premature mitral valve closure (LV diastolic pressure climbs above LA pressure before systole begins, so the valve shuts early) and its corollary, diastolic mitral regurgitation. An intra-aortic balloon pump is contraindicated: diastolic augmentation drives more blood backwards through the incompetent valve.

AR after TAVR — paravalvular leak is the modality-specific complication, and eyeballing a root angiogram in a hybrid room is unreliable. The aortic regurgitation index quantifies it from pressures already on the screen:

AR index = (aortic diastolic pressure − LVEDP) ÷ aortic systolic pressure × 100

It works on exactly the acute-AR logic: a worsening leak drops the aortic diastolic pressure and lifts the LVEDP, so the numerator collapses. An index below 25 predicted markedly higher 1-year mortality after TAVI — 46.0% versus 16.7%, independent of the echocardiographic grade. It prompts post-dilatation, a second valve or repositioning before the patient leaves the room.

Read it alongside the heart rate and the pre-procedural filling pressures. Tachycardia shortens diastole, and a stiff ventricle already running a high LVEDP pushes the index down independently of the leak.

One lesion, opposite tracings. Wide pulse pressure with a modest LVEDP = chronic AR the ventricle has adapted to. Narrow pulse pressure with a very high LVEDP, early diastolic equilibration and a short soft murmur = acute AR it has not adapted to — a surgical emergency, and no balloon pump.

▸ Staging cardiogenic shock — the SCAI A–E ladder what stage is this patient, right now?

“Cardiogenic shock” covers everything from a warm, well-perfused patient with a large infarct to one being coded on the table. Those are not the same escalation decision or the same prognosis, so a shock patient needs a stage before any number below means anything.

Stage	Name	The patient in front of you	What defines the step
A	At risk	Not hypotensive, not hypoperfused — a large infarct or decompensated heart failure that could become shock.	Risk, not physiology.
B	Beginning	Hypotensive or tachycardic, but still perfusing: warm, lactate normal, making urine, mentating.	Hypotension without hypoperfusion.
C	Classic	Hypoperfusion has appeared, and it takes an intervention to reverse.	The arrival of hypoperfusion.
D	Deteriorating	A stage C patient whose first interventions have not restored perfusion after at least 30 minutes.	Trajectory — what you did has not worked.
E	Extremis	Circulatory collapse — refractory arrest, ongoing CPR, ECMO-supported.	The top rung.

The line that matters is B → C, and it is hypoperfusion, not blood pressure. Lactate climbing, extremities cooling, urine falling, the patient going quiet — any one of those is stage C. And a patient can be stage C with a normal blood pressure, which is precisely the error the ladder was written to prevent.

The 2022 update made cardiac arrest a modifier rather than an automatic stage E, and requires the stage to be re-assigned after every intervention. It is a trajectory, not a label — which is what separates C from D.

Stage the patient, then decide on the physiology. The letter tells you how sick they are, and lets two clinicians mean the same thing by the word shock. It does not tell you which pump to put in.

No trial has randomised patients to stage-triggered escalation of mechanical support. In infarct-related shock exactly one device has ever improved survival; VA-ECMO is a rescue for refractory or biventricular failure, taken as a Shock-Team decision rather than the next rung up.

The five shock trials, one line each

Trial	Result	At the bedside
IABP-SHOCK II 2012 · routine balloon pump	Neutral — no reduction in 30-day mortality	Counterpulsation is not definitive therapy, and calling it that is a classic board trap
CULPRIT-SHOCK 2017 · culprit-only vs multivessel PCI	Positive for culprit-only — 30-day death or renal-replacement therapy 45.9% vs 55.4% (RR 0.83, p=0.01)	The opposite of the non-shock STEMI setting. Treat the culprit and stop
ECMO-CS 2022 · immediate VA-ECMO	Neutral — 63.8% vs 71.2% (p=0.21)	Waiting is defensible; 39% of the conservative arm crossed over and did the same
ECLS-SHOCK 2023 · routine early ECLS	Neutral, with harm — 30-day death 47.8% vs 49.0% (p=0.81); bleeding 23.4% vs 9.6%; vascular complications 11.0% vs 3.8%	Routine escalation to ECMO buys complications, not survival
DanGer Shock 2024 · microaxial flow pump	Positive — 180-day mortality 45.8% vs 58.5%, with more device-related complications	The only mechanical support device so far to improve survival, in a narrowly selected population

Baran et al., SCAI expert consensus 2019; Naidu et al., SCAI SHOCK stage update 2022. The staging numbers live in Quick Reference. To watch the ladder used on a live patient, run Cold and Wet at 3 AM.

Waveform Trainer

Every tracing is drawn from a model, not a screenshot — labels on demand. Study mode teaches the signature; Quiz mode hides it and asks you to call it.

▸ Concept primer new to this? start here

A pressure waveform is just the pressure inside a chamber or vessel plotted against time, traced out as the catheter tip sits there. Every site has a characteristic shape, and learning to recognize those shapes — and how disease distorts them — is one of the core skills of invasive cardiology.

Where the tip is	The signature
Atria, and the wedge (which mirrors the left atrium)	A repeating set of small waves and descents: the a wave (atrial contraction), the v wave (atrial filling against a closed valve), and the x and y descents between them
Ventricles	A tall systolic peak dropping to a low diastolic baseline
Arteries	A brisk upstroke, a dicrotic notch (valve closure), then a gradual runoff

Disease shifts these predictably — a giant v wave in mitral regurgitation, a blunted y descent in tamponade, a dip-and-plateau ("square-root sign") in constriction. Reading a tracing is really about spotting the departure from normal.

Using the tool: in Study mode the tracing streams live — freeze it to label each wave and descent; in Quiz mode the labels hide and you call the tracing yourself.

Board Room

Board-style single-best-answer questions with worked explanations. Pick a topic or take a mixed set.

Quick Reference

The numbers and formulas worth memorizing, in one place.

Send feedback

If something here is wrong, broken, or harder to use than it ought to be, say so. The people using this in the lab find things the person who built it cannot.

What kind of feedback is this?

Which part of the app?

Where you are in training optional

What happened, or what would you change?

0 / 4000

Your email optional, and only used to reply to you

Please leave patient identifiers out.

▸ What gets sent along with this nothing hidden

Cath Simulator BETA

The case run end to end, one step at a time.

▸ Concept primer · How to use the simulatorConcept primer new to this? start here

Nothing here is a recording. The catheter travels the real arterial course, in the same body coordinate frame as the 3D projections module; the C-arm swings to the angle the operator would actually be in for that step; and the pressure tracing is generated, so every number in the panel underneath is measured off that waveform rather than typed in.

Control	What it changes	Why it matters
Access	Femoral or right radial.	The route to the aorta, the catheter shapes that work from it, and where the pressure is being measured.
Case	Normal, severe AS, intermediate mid-LAD, severe proximal LAD.	The first two cross the aortic valve; the second two engage the left main and interrogate an LAD lesion with a pressure wire.
Course	Steps through the case one action at a time.	Each step is a decision an operator makes, not a frame of an animation.
Run pullback	Withdraws the catheter across the valve or the lesion.	The gradient is read off the pullback, exactly as it is in the lab.

Predict first, confirm second. Before you advance a step, say what the tracing should do and where the C-arm should end up. The simulator is only worth the time if you are wrong occasionally.

ACCESS

CASE

COURSE

THE LAB drag to orbit · wheel or pinch to zoom 100%

arterial course catheter beam axis

WHAT THE DETECTOR SEES AP 100%

the view the operator would be in for this step

PRESSURE—

Cardiac catheterization, explained and simulated.

Six stages, novice to mastery

Perform the catheterization yourself

Follow the path, or explore in any order.

Introduction to cardiac catheterisation

Anatomy of a case

Predict, then reveal

Checkpoint · the key calls

Vascular access — radial and femoral

Why the distal radius is the landmark

Checkpoint · name the complication

Zeroing, levelling & the phlebostatic axis

Predict, then reveal

What the transducer reports

Checkpoint · correct the reading

The manifold, the fluid column & damping

What the tracing reads — against the pressure that is actually in the artery

Checkpoint · read three unlabelled tracings

Catheters, curves & torque

Predict, then reveal

Choosing one

Checkpoint · pick the fix

The C-arm & the angiographic views

Reading the label — a view is named for where the detector is

Predict, then reveal

The two ways a view lies

Which view lays out which vessel

Checkpoint · pick the view

Radiation & the C-arm

Four views, measured — scatter at eye height, per unit of what the machine reports

Predict, then reveal

Why LAO costs and RAO does not — three rooms, seen from the foot of the table

The protection ladder — measured factors, not claims

What it adds up to — over a career, not a case

The patient's dose — watch the skin

Checkpoint · price the room

Introduction to right heart catheterisation

Floating the balloon — where am I?

The four signatures — four traces, one pressure scale

Three patients, one catheter — and the number that separates them

Checkpoint · name the trace

Reading the wedge — a, v, and when the number lies

Telling a from v — two different events, not two shapes to memorise

Predict, then reveal

Three ways the number lies

Checkpoint · three numbers about to be acted on

Cardiac output — two methods, one patient and they disagree

What you are measuring, and the only two ways to get it

Method 1 Thermodilution — where the number comes from, and why the injectate matters

Method 2 Fick — what it needs, and where it bends

Two methods, and only two — there is no third way to get a cardiac output here

Predict, then reveal

Checkpoint · three sets of curves

Oximetry — finding and sizing a shunt

What you are looking for, and how a saturation finds it

The comparison Normal against a shunt — the whole module in one picture

Predict, then reveal

How to run one

Checkpoint · four runs

PCI Technique — Doing the Case

The sequence

Guide, wire, support

Preparing the lesion — and what the calcium arc costs you

Sizing — off the reference, not off the lesion

Optimisation — four numbers, not an impression

Bifurcations — provisional until it is not

What goes wrong that is specific to doing this

The Procedure & Its Complications

Curriculum Map

Where you are

Foundations — the path

The instruments

Topics covered

Design principles

Guideline & reference sources

Version & review status

Change log

Angiography

3D Angiographic Projections BETA

Structural Interventions — Decision Pathways