Oblique needle segmentation for 3D TRUS-guided robot-aided transperineal prostate brachytherapy

Abstract

3D TRUS-guided robot-aided prostate brachytherapy provides tools for dynamic re-optimization of a dose plan by freeing needle insertions from parallel trajectory constraints, i.e., needle trajectories can be positioned with considerable flexibility including oblique. However, oblique insertion results in the needle intersecting the 2D TRUS image and appearing as a dot, leading to blind guidance. Here, we propose a method for oblique needle segmentation and tracking to be used in a 3D TRUS guided and robot aided prostate brachytherapy system. This algorithm applies a grey-level change detection technique to find the location and orientation of needles from 3D images. Three 2D images containing the needle (oblique sagittal, coronal and transverse planes) are extracted and displayed in near real-time. Testing showed that our algorithm can find 3D needle orientation within 0.54/spl deg/ for a chicken tissue phantom, and 0.58/spl deg/ for agar phantoms, over a /spl plusmn/15/spl deg/ insertion orientation. The execution time averaged 0.13s on a 1.2 GHz computer.