A novel CAD system for analyzing cardiac first-pass MR images

We propose a novel framework for the analysis of myocar-dial transit of contrast agent on cardiac first-pass magnetic resonance imaging (FP-MRI). Significant shape changes of the left ventricle (LV) due to heart contraction and respiratory motion limit accurate quantification of perfusion parameters on time series data. To account for the rigid and nonrigid deformations of the heart, we developed a registration methodology for the segmented LV that includes: (i) an initial affine-based registration, (ii) a local B-splines based alignment to maximize a similarity metric that accounts for the 1
^st
- and 2
^nd
-order normalized mutual information (NMI) between the globally aligned frames, followed by (iii) a refinement registration step that deforms each pixel of the target wall over evolving iso-contours to closely match the reference wall based on solving Laplace equation. Finally, from contrast agent kinetic curves, obtained from the co-aligned frames, four perfusion indexes - peak signal intensity, time-to-peak, initial up-slope, and the average signal change of the slowly varying agent delivery phase - are determined. To depict regional perfusion, we computed pixel-by-pixel parametric maps for the derived indexes. We have tested our framework on 24 FP-MRI data sets that have been collected from 8 patients with ischemic damage from heart attacks, who are undergoing a novel myoregeneration therapy and have documented an improvement in the visualization and display of treatment effects.