Speckle noise reduction and segmentation of kidney regions from ultrasound image

Ultrasound imaging plays a crucial roles in medical field to estimate kidney size, position, appearance and helps to detect structural abnormalities as well as the presence of cysts, stones, cancer, congenital anomalies, swelling, blockage of urine flow etc. But presence of speckle noise and low contrast in ultrasound images, detection of kidney is a difficult as well as challenging task. In this paper we develop and implement a system, which can segment human kidney from ultrasound images, usable during surgical operations like punctures. First, we take input image and perform restoration on that image. Then we reduce speckle noise and smooth resultant image using Gabor filter. Histogram equalization is used to enhance the image quality. For this study, two segmentation techniques were chosen to be compared consist of cell segmentation and region based segmentation. For better result we use region based segmentation to extract kidney regions. Lastly we perform refinement and crop the segmented kidney region from the original image.