An implementation of Otsu thresholding and the Chan–Vese method on the PCO segmentation of ultrasound images

Medical practitioners have been using ultrasound images to diagnose and monitor polycystic ovarian syndrome (PCOS) manually. However, manual segmentation is laborious and time-consuming due to the disturbance of speckle noise in ultrasound images. In addition, manual segmentation could produce errors. Thus, researchers have been implementing image processing for a fast and accurate diagnosis of PCOS. Image processing consists of steps, amongst which the most crucial is image segmentation. Before segmentation, the median filter is applied for preprocessing. For the segmentation step, this study proposes combining Otsu’s threshold method and the Chan–Vese method. In this study, the application of the different greyscale levels of Otsu’s thresholding is compared with that of the Chan–Vese method. The proposed method is also compared with the classic Chan–Vese method quantitatively. The comparison table reveals that the proposed method shows superiority over the classic Chan–Vese method.