Hypoechoic Nodules Detection and Classification in TRUS Prostate Images Using Active Contours and Parabolic Zone Division

In this paper, a semi-automatic system for the detection and classification of hypo echoic nodules in prostate transrectal ultrasound (TRUS) images using active contours identification and parabolic zone division is presented. First, a TRUS image is acquired, then the image is preprocessed using a median filter. After preprocessing, the image is segmented using active contours, then the different zones of the prostate are divided by a novel parabolic algorithm in two zones: transitional and peripheral. Finally, the stage of feature extraction is performed and the hypo echoic nodules are classified as low or high cancer risk. The experiments and results showed the ability of the system to detect hypo echoic nodules which can represent prostate cancer.