Automated vessel-specific coronary artery calcification quantification with deep learning in a large multi-centre registry

Williams, MC; Shanbhag, AD; Zhou, J; Michalowska, AM; Lemley, M; Miller, RJH; Killekar, A; Waechter, P; Gransar, H; Van Kriekinge, SD; Builoff, V; Feher, A; Miller, EJ; Bateman, T; Dey, D; Berman, D; Slomka, PJ

doi:10.1093/ehjci/jeae045

Coronary disease & ACS

Automated vessel-specific coronary artery calcification quantification with deep learning in a large multi-centre registry

Eur Heart J Cardiovasc ImagingPublished 20 Feb 2024Diagnostic accuracy study

Read the original article ↗PubMed

Cite

Authors' conclusion

Vessel-specific CAC assessment with DL can be performed accurately and rapidly on gated CT and AC CT and provides important prognostic information.

Study context

Population: Patients undergoing gated computed tomography (CT) or attenuation correction (AC) CT in a large multi-centre registryDeep learning (DL) trained on 3000 gated CT
Participants: 2,094 vs 5,969

Comparison: Automated DL vessel-specific coronary artery calcification (CAC) vs expert ground truth
Gated CTn = 2,094
AC CTn = 5,969
Outcome: Vessel-specific agreement (linear weighted Cohen's Kappa) between DL and expert ground truth for CAC categories (zero, 1-100, 101-400, >400 Agatston units (AU)) in LM/LAD, LCX and RCA

Methodological detailsmulticentre, registry

Multicentre
Registry

Performance

As reported in the abstract

Test or model	LM/LAD Kappa	Left circumflex (LCX) Kappa	Right coronary artery (RCA) Kappa
Gated CT	0.90 (95% CI 0.89 to 0.92)	0.70 (0.68 to 0.73)	0.79 (0.77 to 0.81)
AC CT	0.78 (0.77 to 0.80)	0.60 (0.58 to 0.62)	0.70 (0.68 to 0.71)

LM/LAD Kappa

Gated CT0.90 (95% CI 0.89 to 0.92)

AC CT0.78 (0.77 to 0.80)

Left circumflex (LCX) Kappa

Gated CT0.70 (0.68 to 0.73)

AC CT0.60 (0.58 to 0.62)

Right coronary artery (RCA) Kappa

Gated CT0.79 (0.77 to 0.81)

AC CT0.70 (0.68 to 0.71)

Further findings

Major adverse cardiovascular events (MACE) · 2.4 ± 1.4 years follow-up

	Gated CT	AC CT	← Lower risk ←→ Higher risk →	HR (95% CI)	P value
LM/LAD CAC >400 AUGated CT; highest risk of MACE	12%	14%	0.050.21520	12.0 (7.96–18.0)	P < 0.001
LM/LAD CAC >400 AUAC CT; highest risk of MACE			0.050.21520	4.21 (3.48–5.08)	P < 0.001
			0.050.21520Hazard ratio (log scale)

Original publication

Automated vessel-specific coronary artery calcification quantification with deep learning in a large multi-centre registry. Eur Heart J Cardiovasc Imaging. Published 20 February 2024. doi:10.1093/ehjci/jeae045

Read the original article ↗

Composed by AI from the published abstract and checked against it. How Briefs are made →