Prostate ultrasound image segmentation and registration

The overall objective of this research is to improve the quality of the widely used prostate cancer treatment—the transrectal ultrasound (TRUS) guided transperineal interstitial permanent prostate brachytherapy (TIPPB). We focus on the development of image segmentation and registration algorithms specifically for medical ultrasound images to solve two challenging problems: (1) prostate ultrasound image segmentation and (2) prostate ultrasound and fluoroscopy image registration.; Prostate boundaries are routinely outlined in parallel transverse TRUS images at various stages of TIPPB. To reduce the effort required for manual delineation and to increase the robustness of automated segmentation, we have developed a deformable model-based Bayesian framework to integrate multiple sources of information, such as prior knowledge of the prostate shape as well as the image edge information. The algorithm has been tested on 125 prostate TRUS images from 16 patients, and its performance was shown to be robust and as accurate as human observers.; TRUS and X-ray fluoroscopy fusion provides a promising way of performing post-implant dosimetric analysis, the principal measure of implant quality of the TIPPB. To this end, we have proposed a fiducial-based solution and preliminarily validated it on two tissue mimicking prostate phantoms. The results show that it is a feasible and accurate alternative to the currently used CT-based post-implant dosimetry.