Intra-operative Registration Methods for Image-Guided Kidney Surgery

Abstract

In this work, I examined, developed, and validated methods for intra-operative registration in minimally invasive kidney surgery. One of these methods uses a conoscopic laser that can be inserted through a small trocar port to scan the kidney and obtain a surface point cloud. To enable intra-operative feature tracking, a novel method for texture-mapping the conoscopic surface was developed using laparoscopic video. The feasibility and accuracy of this texture-mapping method was evaluated. In addition, a registration method using features tracked from the textured-mapped conoscopic surfaces was investigated for use in-vivo, under laparoscopic conditions in swine. The feasibility and accuracy of the registration method was investigated and shown to be promising.

Finally, to evaluate the need for non-rigid deformation correction in an image guidance system for the kidney, a preliminary study of non-rigid deformation in the kidney was performed. In this study, ex-vivo porcine kidneys were perfused, the renal vessels clamped, and an incision was made to simulate surgical conditions. The resulting deformation was measured, and the fiducial-tracked displacements were interpolated using a spline. The accuracy of this method was assessed, and the results show this method could be used with tracked features from intra-operative conoscopic scans to correct for non-rigid deformation.