Segmentation of Ultrasound Medical Image Using A Hybrid Method

In this paper, a hybrid method for the segmentation of medical ultrasound images is proposed. The method is composed of two parts as region growing and region merging. First, we initially segment the images into many small regions using a region growing method. The region growing algorithm is based on some pixels we specially choose as growing centers. The initially segmented images acquired from the region growing algorithm have many small pseudo regions caused by the speckle noises. To get better segmentations, we secondly combine this initial segmentation algorithm with a region merging algorithm based on region energy. The experimental result shows that the hybrid method can perform well in ultrasound medical image segmentation.