DICOM Depth Image Generation using GPGPU for Fast Position and Orientation Estimation on Liver Surgery Support System

We are developing a surgical support system for a liver laparotomy surgery. In our system, we aim to navigate the surgical tools during surgery adapted to the change of the position, orientation and deformation of the liver during operation. The position and orientation of the liver is measured by matching the surface shape of the real liver measured by depth camera and the depth image generated from the 3D model of the liver. In this paper, we present a fast depth image generation method from computed tomography images (DICOM) using a general-purpose computing on graphics processing units (GPGPU). The experiments and results to estimate the position and orientation of the liver from the generated DICOM depth image and the depth image of a real liver model acquired from a depth camera are also described.