Fractal Analysis of Dynamic Stress CT-Perfusion Imaging for Detection of Hemodynamically Relevant Coronary Artery Disease

Combined computed tomography–derived myocardial blood flow (CTP-MBF) and computed tomography angiography (CTA) has shown good diagnostic performance for detection of coronary artery disease (CAD). However, fractal analysis might provide additional insight into ischemia pathophysiology by characterizing multiscale perfusion patterns and, therefore, may be useful in diagnosing hemodynamically significant CAD. The purpose of this study was to investigate, in a multicenter setting, whether fractal analysis of perfusion improves detection of hemodynamically relevant CAD over myocardial blood flow quantification (CTP-MBF) using dynamic, 4-dimensional, dynamic stress myocardial computed tomography perfusion (CTP) imaging. In total, 7 centers participating in the prospective AMPLIFiED (Assessment of Myocardial Perfusion Linked to Infarction and Fibrosis Explored with Dual-source CT) study acquired CTP and CTA data in patients with suspected or known CAD. Hemodynamically relevant CAD was defined as ≥90% stenosis on invasive coronary angiography or fractional flow reserve <0.80. Both fractal analysis and CTP-MBF quantification were performed on CTP images and were combined with CTA results. This study population included 127 participants, among them 61 patients, or 79 vessels, with CAD as per invasive reference standard. Compared with the combination of CTP-MBF and CTA, combined fractal analysis and CTA improved sensitivity on the per-patient level from 84% (95% CI: 72%-92%) to 95% (95% CI: 86%-99%; P = 0.01) and specificity from 70% (95% CI: 57%-82%) to 89% (95% CI: 78%-96%; P = 0.02). The area under the receiver-operating characteristic curve improved from 0.83 (95% CI: 0.75-0.90) to 0.92 (95% CI: 0.86-0.98; P = 0.01). Fractal analysis constitutes a quantitative and pathophysiologically meaningful approach to myocardial perfusion analysis using dynamic stress CTP, which improved diagnostic performance over CTP-MBF when combined with anatomical information from CTA.