Fractal analysis of dynamic, four-dimensional computed tomography myocardial perfusion (4D-CTP) imaging might have potential for noninvasive differentiation of microvascular ischemia and macrovascular coronary artery disease (CAD) using fractal dimension (FD) as quantitative parameter for perfusion complexity. This multi-center proof-of-concept study included 30 rigorously characterized patients from the AMPLIFiED trial with nonoverlapping and confirmed microvascular ischemia (n(micro) = 10), macrovascular CAD (n(macro) = 10), or normal myocardial perfusion (n(normal) = 10) with invasive coronary angiography and fractional flow reserve (FFR) measurements as reference standard. Perfusion complexity was comparatively high in normal perfusion (FDnormal = 4.49, interquartile range [IQR]:4.46-4.53), moderately reduced in microvascular ischemia (FDmicro = 4.37, IQR:4.36-4.37), and strongly reduced in macrovascular CAD (FDmacro = 4.26, IQR:4.24-4.27), which allowed to differentiate both ischemia types, p < 0.001. Fractal analysis agreed excellently with perfusion state (kappa = 0.96, AUC = 0.98), whereas myocardial blood flow (MBF) showed moderate agreement (kappa = 0.77, AUC = 0.78). For detecting CAD patients, fractal analysis outperformed MBF estimation with sensitivity and specificity of 100% and 85% versus 100% and 25%, p = 0.02. In conclusion, fractal analysis of 4D-CTP allows to differentiate microvascular from macrovascular ischemia and improves detection of hemodynamically significant CAD in comparison to MBF estimation.
