Comparison of quantitative myocardial perfusion imaging CT to fluorescent microsphere-based flow from high-resolution cryo-images

摘要: Myocardial perfusion imaging using CT (MPI-CT) has the potential to provide quantitative measures of myocardial blood flow (MBF) which can aid diagnosis coronary artery disease. We evaluated accuracy MPI-CT in a porcine model balloon-induced LAD ischemia guided by fractional reserve (FFR). quantified MBF at baseline (FFR=1.0) and under moderate (FFR=0.7) compared fluorescent microsphere-based from high-resolution cryo-images. Dynamic, contrast-enhanced images were obtained spectral detector (Philips Healthcare). Projection-based mono-energetic reconstructed processed obtain MBF. Three quantification approaches evaluated: singular value decomposition (SVD) with fixed Tikhonov regularization (ThSVD), SVD determined L-Curve criterion (LSVD), Johnson-Wilson parameter estimation (JW). The three over-estimated JW produced most accurate MBF, average error 33.3±19.2mL/min/100g, whereas LSVD ThSVD had greater over-estimation, 59.5±28.3mL/min/100g 78.3±25.6 mL/min/100g, respectively. Relative as assessed ratio LAD-to-remote myocardium was strongly correlated between cryo-imaging, R2=0.97, R2=0.88 0.78 for ThSVD, tissue impulse response functions (IRFs) each approach sources error. While constrained physiologic solutions, both IRFs non-physiologic properties due noise. L-curve provided noise-adaptive but did not eliminate IRF or optimize accuracy. These findings suggest that model-based may be more appropriate cryo-imaging support development providing spatial distributions