Automated TIMI frame counting using 3-d modeling

Three dimensional coronary modeling and reconstruction can assist in the quantitative analysis of coronary flow velocity from 2-d coronary images. In this paper a novel method to assess coronary flow velocity is proposed. First, 3-d models of the coronary arteries are estimated from bi-plane X-ray images using epipolar constraint energy minimization for the selected fiducial points like bifurcations, and subsequently 3-d B-spline energy minimization for the arterial segments. A 4-d model is assembled from a set of 3-d models representing different phases of the cardiac cycle. The 4-d model is fitted to the 2-d image sequences containing basal or hyperemic blood flow information. Then, by counting the frames in analogy with TIMI frame counting, an index of the mean coronary flow velocity can be estimated. Our experimental results show that the algorithm correlates with r = 0.98 (P<0.0001, 95% CI 0.92–0.99) to the clinical measurements of the TFC.