Noise Sensitive Trajectory Planning for MR Guided TAVI

Image-guided, pre-operative planning is fast becoming the gold standard for navigating real-time robotic cardiac surgeries. Planning helps the surgeon utilize the amended quantitative information of the target area and assess the suitability of the offered intervention technique prior to surgery. In apex access aortic valve replacements, safe zone generation for the penetration of delivery module along the left ventricle (LV) is a crucial step to prevent untoward cases from emerging. To address this problem, we propose a computational core, which is to locate left ventricle borders and specifically papillary muscles (PM), create an obstacle map along the left ventricle (LV), and ultimately extract a dynamic (off-line) trajectory for tool navigation. To this end, we first applied an isotropic diffusion on short-axis (SA) cardiac magnetic resonance (CMR) images. Second, we utilized an active contour model to determine the LV border. Third, we clustered the LV crops to locate the PM. Finally, we computed the centroids of each of the LV segments to determine the safest path for an aortic delivery module.