Title: Integration of Multiple Imaging Data for improved Volumetric Cardiac Motion Analysis

Abstract:
We present our recent efforts for the improved Volumetric Cardiac Motion Analysis based on data from multiple imaging modalities. First, we will present our framework for the automated spatiotemporal analysis of the heart's ventricles based on Ultrasound, CT and tMRI data. Recent advances in CT have allowed the acquisition of high spatial resolution data that based on our deformable modeling methods we can built a detailed model of the ventricles. We then estimate the cardiac motion for a full cardiac cycle using tagged data, which is hard to achieve with a model constructed from only sparse clinical tagged MR images. Our accurate estimation algorithms compute two sets of cues from tagged MRI, the intersections of the three tagging planes, and the intersections of the cardiac boundary with the tagging planes. The image forces on the intersections are interpolated onto the cardiac mesh vertices by tessellation and meshless FEMs. The LV motion reconstruction provides information for further analysis of cardiac mechanisms. Results on normal and pathologic hearts will be presented. Finally, we will present recent results on the accuracy of 2D ultrasound-based cardiac analysis by comparing it to tMRI based analysis. These methods are based on collaborative research with NYU medical school and Columbia Medical School.