A new parallel method for medical image segmentation using watershed algorithm and an improved gradient vector flow

Abstract

Image segmentation is a fundamental task in image analysis responsible for partitioning an image into multiple sub-regions based on a desired feature. This paper presents a parallel approach for fast and robust object detection in a medical image. First, the proposed approach consists to decompose the image into multiple resolutions by a Gaussian pyramid algorithm. Then, the object detection in the higher pyramids levels is done in parallel by a Hybrid model combining Watershed algorithm, GGVF (Generic Gradient Vector Flow) and NBGVF (Normally Biased Gradient Vector Flow) models where the initial contour is subdivided into sub-contours, which are independent from each other. Each sub-contour converges independently in parallel. The last step of our approach consists to project the sub-contours detected in the low resolution image to the high-resolution image. The experimental results were performed using a number of synthetic and medical images. Its rapidity is justified by runtime comparison with a conventional method.