Artificial Intelligence-based Coronary Stenosis Quantification at Coronary CT Angiography versus Quantitative...

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Medical Diagnostics

Background:

• Coronary artery disease (CAD) diagnosis relies on assessing stenosis severity.
• Invasive quantitative coronary angiography (QCA) is the reference standard but is invasive.
• Non-invasive CT angiography (CCTA) offers a less invasive alternative, but accurate quantification is crucial.

Purpose of the Study:

• To evaluate the diagnostic performance of a novel AI-based coronary stenosis quantification (AI-CSQ) software.
• To compare AI-CSQ quantified stenosis severity on CCTA against invasive QCA.
• To assess AI-CSQ accuracy on both per-vessel and per-patient bases.

Main Methods:

• A post hoc analysis of 120 participants from three clinical trials who underwent both CCTA and QCA.
• Coronary stenosis severity was quantified using AI-CSQ software on CCTA images.
• Blinded comparison of AI-CSQ results with QCA reference standard.

Main Results:

• AI-CSQ demonstrated high diagnostic performance for diameter stenosis (DS) ≥50% and ≥70%.
• Per-vessel sensitivity, specificity, and accuracy for DS ≥50% were 80%, 88%, and 86%.
• Areas under the ROC curve (AUCs) were high, ranging from 0.92 to 0.93 for per-vessel and 0.88 to 0.93 for per-patient analyses.

Conclusions:

• AI-CSQ applied to CCTA shows high diagnostic performance for quantifying coronary stenosis.
• The AI tool is comparable to QCA, offering high sensitivity for stenosis detection.
• AI-CSQ represents a promising non-invasive method for assessing coronary artery stenosis severity.