Advanced CT analysis
Changing any of these mid-trial creates a second algorithm. Bump the engine version, record the changeover date, and analyse by version.
Pre-procedural recommendation
—
Length — lesion, landing zones and the rack i
Diameter — permitted window and the shelf i
Reference
Reference lumen diameters are taken at face value by default. Where the reference cross-section itself carries calcium, blooming narrows the apparent lumen, so a small upward correction (k, default 1.05) is applied. k is a placeholder until you calibrate it against your own scanner, kernel and analysis software — a systematically wrong k biases the whole CCTA arm in one direction.
Five published strategies are selectable; pre-specify one rather than choosing per case. The default anchors to the distal lumen reference and rounds up, which is the EAPCI consensus approach. The vessel / EEL option transposes the ILUMIEN III–IV protocol to CT: take the outer-wall diameter at both references, use the smaller, and round down. Rounding direction is not cosmetic — lumen understates the vessel so it rounds up; the outer wall overstates what the artery tolerates so it rounds down. Vessel-based sizing selects larger devices, so the strategy choice moves mean stent diameter systematically.
Whichever strategy runs, three bounds hold: at most 1.20× the distal reference (perforation ceiling), at least 0.95× (malapposition floor), and at most 1.10× the proximal reference. If nothing satisfies all three, the malapposition floor is waived first — the perforation ceiling is never waived silently — and the case is flagged for intraprocedural confirmation.
Target coverage is the lesion plus a landing zone each side (default 2 mm, absolute minimum 1 mm). Candidates are scored with an asymmetric cost: falling short is penalised 1.6× harder than overshooting, because geographic miss is worse than a few millimetres of excess metal. Where no single device reaches the minimum, overlapping stents are planned with 3 mm of overlap, capped at three devices.
Calcium does not change the diameter — it triggers a preparation plan, because the failure mode with heavy calcium is under-expansion, not undersizing. A CT-adapted score weights circumferential calcium, arc ≥180°, calcified length >5 mm and calcified nodule. Thickness is deliberately omitted: blooming makes CT calcium thickness unreliable. The thresholds are transposed from intravascular imaging and are not externally validated.
Lesions in the same vessel interact. Two separated by 10 mm or less become one treatment unit spanning both plus the segment between — one stent beats two stents and a fragile bridge. The merged unit inherits the outermost references, the worst image quality, and the union of all calcium and anatomy flags. Because the distal reference of a proximal lesion and the proximal reference of the one below it describe the same artery, a disagreement over 0.50 mm is flagged as a measurement problem.
Within a vessel the distal lesion goes first, so no device crosses a fresh stent. Across vessels, ordering follows prognostic priority — a convention for discussion, not evidence, and labelled as such. The sequence can be reordered by hand and the override is recorded.
Crossing comes before sizing. CT-RECTOR and the CCTA J-CTO score are implemented to their source definitions, and the crossing strategy follows the published hybrid algorithm from cap ambiguity, occlusion length, distal vessel quality and collaterals. Collateral suitability is not a CT measurement — enter it from angiography or the planner will say so.
Sizing fails in the opposite direction here. Distal to an occlusion the vessel is negatively remodelled from chronic low flow and is systematically under-measured, and it enlarges over the months after recanalisation. The diameter shown is therefore a floor, with the range imaging is likely to reveal shown alongside it. The planner never upsizes on the CT number alone, and intravascular imaging is mandatory. A donor-vessel lesion is always sequenced ahead of any CTO attempt.
Confidence starts high and is downgraded for moderate or poor image quality, motion, blooming or circumferential calcium, diffuse disease without a normal reference, a distal reference under 2.5 mm, and any multi-stent plan. A lesion can carry high confidence and still mandate verification — the two answer different questions.
Angles are signed: LAO positive, RAO negative; CRA positive, CAU negative. Every working and confirmatory pair is at least 30° of beam separation apart, so the second view is genuinely independent. These are population defaults — in the CCTA arm they are superseded by the patient-specific angle derived from the 3D centreline, and serve as the fallback when that angle falls outside the gantry envelope. Report predicted foreshortening alongside the working view: above about 20%, the angiographic length understates the truth and the CT-derived length should win.