Segmentation of implanted radioactive seeds in 3D TRUS images for intraoperative prostate brachytherapy

Intraoperative prostate brachytherapy has potential to solve the problems in current brachytherapy, and verification of implanted seed locations in ultrasound (US) images is a necessary step in an intraoperative procedure. However, due to low quality of US images, and small size of radioactive seeds, it is difficult to achieve seed segmentation in US images. In this paper, we introduce an intraoperative seed segmentation algorithm in three-dimensional transrectal US images. A volumetric cone-beam flat-panel micro-CT scanner was used to evaluate the accuracy of the algorithm. We found that the rms error in determining seed locations using the algorithm was 0.98 mm in agar phantoms, and 1.02 mm in chicken phantoms