A Parallel Optimized Approach for Prostate Boundary Segmentation from Ultrasound Images

This paper presents an automated multi-step algorithm for segmenting the prostate boundary from ultrasound images. As the acquired ultrasound images are usually affected due to the presence of Speckle and other noise artifacts, the first step is to pre-process these images using Sticks Filtering. Further, in stage two an initial contour is determined from these pre-processed images. Finally, Ant Colony Optimization is used to segment the prostate boundary to determine the volume of the prostate. In the last section the performance of both pre-processing stage as well as segmentation stage is presented and is compared with existing algorithms.