Photon-Counting Detector CTA in Standard- and Ultrahigh-Resolution Modes for Diagnosing Coronary Artery Stenosis Using Invasive Angiography as the Reference: A Prospective Study.

Abstract

<b>BACKGROUND</b>. The literature reports excellent diagnostic performance of coronary CTA using photon-counting detector (PCD) CT, albeit obtained using various acquisition and reconstruction protocols. <b>OBJECTIVE</b>. The purpose of this study was to assess the diagnostic performance for detecting significant stenosis of coronary CTA performed by PCD CT with various standard-resolution (SR) and ultrahigh-resolution (UHR) protocols, using invasive coronary angiography (ICA) as the reference standard. <b>METHODS</b>. This prospective study enrolled inpatients undergoing coronary CTA between October 2023 and October 2024. Participants underwent coronary CTA by PCD CT, sequentially alternating between SR (collimation: 144 × 0.4 mm) and UHR (collimation: 120 × 0.2 mm) modes across participants. SR examinations were reconstructed into normal (SR_normal) and virtual noncalcium (SR_VNCa) image sets, both using 0.6-mm slice thickness and Bv40 kernel. UHR examinations were reconstructed into normal (UHR_normal [0.6-mm slice thickness, Bv40 kernel]) and thin (UHR_thin [0.2-mm slice thickness, Bv64 kernel]) image sets. Two radiologists independently measured the diameter of stenoses. The final analysis included patients who underwent ICA after CTA; a cardiologist reviewed the ICA images to determine the reference standard. Stenoses were considered significant at a threshold of 50% or greater. <b>RESULTS</b>. The SR group included 61 patients (mean age, 67 ± 9 [SD] years; 46 men, 15 women; 788 segments analyzed). The UHR group included 61 patients (67 ± 11 years; 43 men, 18 women; 825 segments analyzed). Per-segment sensitivity, specificity, and accuracy for reader 1 were 92.9%, 89.9%, and 90.5% for SR_normal, respectively; 92.9%, 91.6%, and 91.8% for SR_VNCa; 96.0%, 92.4%, and 93.0% for UHR_normal; and 100.0%, 98.6%, and 98.8% for UHR_thin; and for reader 2 were 92.9%, 88.8%, and 89.6% for SR_normal; 93.5%, 92.3%, and 92.5% for SR_VNCa; 96.0%, 91.6%, and 92.2% for UHR_normal; and 100.0%, 98.9%, and 99.0% for UHR_thin. Significant (<i>p</i> < .05) differences included SR_VNCa versus SR_normal for specificity for both readers and accuracy for reader 2; UHR_thin versus UHR_normal for sensitivity, specificity, and accuracy for both readers; and UHR_thin versus SR_VNCa for sensitivity, specificity, and accuracy for both readers. <b>CONCLUSION</b>. Coronary CTA performed by PCD CT achieved high diagnostic performance in the SR or UHR mode. Performance was higher for SR_VNCa than SR_normal and for UHR_thin than either UHR_normal or SR_VNCa. <b>CLINICAL IMPACT</b>. The findings will help to optimize protocols for coronary CTA performed by PCD CT.