Automatic Segmentation and Detection System for Varicocele in Supine Position

Image analysis is an important technique that can help specialists localize, detect and segment objects in different types of medical images such as MRI, CTs, and ultrasounds (US). In this research, we use US images to identify and segment the enlarged veins in the pampiniform venous plexus, which is called varicocele. The proposed method aims to determine whether a potential patient is affected or not. This method was evaluated on 90 US images that were taken to the left testicles of 90 patients using the Supine position. This system analyzes US images in three stages which are; preprocessing, processing, and edge detection. The Region Of Interest (ROI) of the pampiniform plexus area was extracted using Otsu segmentation with different parameters 0.1, 0.2, and 0.17, and different color modes (Grayscale, YCbCr, RGB). In the processing stage, different denoising filters were used. Eventually, in the edge detection stage, four edge detectors were applied which are Canny, Soble, Prewitt, and Roberts. Results showed that the best accuracy in detecting varicocele was 78% when YCbCr color mode yellow (y) channel is used with 0.1 Otsu segmentation and the Canny edge detector. The system also showed a Sensitivity of 91%, the test was able to detect 91% of the people with Varicocele, and a Specificity of 39%.