A Novel Method for Preoperative 6D Pose Estimation of Rat Skull Based on 3D Vision Techniques

Abstract

Optical stereoscopic imaging has the advantages of smaller size, less complicated structure, lower cost, better real-time performance in brain surgeries over MRI and CT. In this paper, we present a novel method based on 3D vision techniques to estimate the 6-DOF pose of a rat skull semi-automatically, which can be applied in guiding surgical robots to perform brain surgeries on rats. This method contains three steps. First, a binocular camera system is used to capture images and pointcloud of a rat skull. Next, the figures of anatomical sutures of the skull in images are fitted with proper curves to reconstruct the 3D positions of the bregma and lambda points, which are used for confirming the anteroposterior (AP) axis. Then a pair of symmetrical points on the skull surface is selected by processing the skull pointcloud to compute the orientation of the mediolateral (ML) and dorsoventral (DV) axis eventually. An experimental rotating platform was developed to adjust the rat skull to different positions to validate the proposed method. By calculating the angular difference of the spatial orientation of the skull in different positions and comparing it with the operating angle of the platform, the rotation errors around the X, Y and Z axes of the predefined world coordinate system were obtained, which were 0.17±0.10°, 0.10±0.04° and 0.14±0.06° respectively. The results show that this method can provide skull pose estimation with high accuracy and precision for position guidance during brain surgeries.