Effect of Reconstruction Kernel and Virtual Monoenergetic Imaging on Segmentation-Based Measurement of Coronary Plaque Volume With Photon-Counting CT

Objectives Coronary computed tomography angiography is the primary modality for noninvasive assessment of coronary artery disease. Photon-counting computed tomography (PCCT) offers superior spatial resolution and spectral imaging for detailed characterization of atherosclerotic plaques. This study aimed to evaluate the impact of virtual monoenergetic imaging (VMI) energy levels and reconstruction kernels on segmentation-based measurement of plaque volume in individuals with coronary atherosclerosis using PCCT. Materials and Methods Fifty study participants underwent coronary computed tomography angiography with ultra-high-resolution PCCT. Both polyenergetic, 120 kVp (T3D) images and spectral images at varying VMI energy levels were reconstructed using different kernels. Plaque volumes were measured using semiautomated attenuation-based segmentation, adjusting segmentation thresholds for each VMI energy level. In addition, absolute plaque volume measurements were conducted using a coronary phantom simulating different plaque types. Results Using a sharper kernel (Bv64 vs Bv48) significantly increased noncalcified plaque volume measurements ( P < 0.005) in study participants, whereas a 0.2-mm slice thickness reduced calcified plaque volumes compared with 0.4 mm ( P < 0.005). VMI energy level had no impact on measured volumes. Phantom measurements confirmed significant variability in measured volumes of calcified and noncalcified plaques depending on reconstruction method, as well as a minor effect of VMI level. Conclusions In PCCT, the reconstruction kernel predominantly affects noncalcified coronary plaque quantification, whereas slice thickness mainly impacts calcified plaque volumes. In study participants, measured plaque volumes were not affected by VMI energy level when energy-specific segmentation thresholds were used, although a minor effect of VMI was observed in the phantom model.