To evaluate the potential of grating-based phase-contrast computed-tomography (gb-PCCT) to classify human carotid and coronary atherosclerotic plaques according to modified American Heart Association (AHA) criteria. Experiments were carried out at a laboratory-based set-up consisting of X-ray tube (40kVp), grating-interferometer and detector. Eighteen human carotid and coronary artery specimens were examined. Histopathology served as the standard of reference. Vessel cross-sections were classified as AHA lesion type I/II, III, IV/V, VI, VII or VIII plaques by two independent reviewers blinded to histopathology. Conservative measurements of diagnostic accuracies for the detection and differentiation of plaque types were evaluated. A total of 127 corresponding gb-PCCT/histopathology sections were analyzed. Based on histopathology, lesion type I/II was present in 12 (9.5%), III in 18 (14.2%), IV/V in 38 (29.9%), VI in 16 (12.6%), VII in 34 (26.8%) and VIII in 9 (7.0%) cross-sections. Sensitivity, specificity and positive and negative predictive value were ≥0.88 for most analyzed plaque types with a good level of agreement (Cohen's kappa = 0.90). Overall, results were better in carotid (kappa = 0.97) than in coronary arteries (kappa = 0.85). Inter-observer agreement was high with kappa = 0.85, p < 0.0001. These results indicate that gb-PCCT can reliably classify atherosclerotic plaques according to modified AHA criteria with excellent agreement to histopathology. • Different atherosclerotic plaque types display distinct morphological features in phase-contrast CT. • Phase-contrast CT can detect and differentiate AHA plaque types. • Calcifications caused streak artefacts and reduced sensitivity in type VI lesions. • Overall agreement was higher in carotid than in coronary arteries.
Read full abstract