INTEGRATING VALIDATION INCREMENTAL NEURAL NETWORK AND RADIAL-BASIS FUNCTION NEURAL NETWORK FOR SEGMENTING PROSTATE IN ULTRASOUND IMAGES

Prostate hyperplasia usually affects male adults in developed countries. Transrectal ultrasound (TRUS) imaging is widely used to diagnose prostate disease. Ultrasonic images have primitive echo perturbations and speckle noise, which may confuse physicians during inspection. Therefore, this study proposes an automatic prostate segmentation system for TRUS images to eliminate the process of manual outlining the prostate region. The proposed automatic segmentation system combines the active contour model (ACM) with a prostate classifier. The prostate classifier consists of a validation incremental neural network (VINN) and a radial-basis function neural network (RBFNN). Experimental results show that the proposed method has higher accuracy than that of the regular ACM method.