Toward automatic zonal segmentation of prostate by combining a deformable model and a probabilistic framework

This paper introduces an original method for automatic 3D segmentation of the prostate gland from Magnetic Resonance Imaging data. A statistical geometric model is used as a priori knowledge. Prostate boundaries are then optimized by a Bayesian classification based on Markov fields modelling. We compared the accuracy of this algorithm, free from any manual correction, with contours outlined by an expert radiologist. In 3 random cases, including prostates with cancer and benign prostatic hypertrophy (BPH), mean Hausdorff s distance (HD) and overlap ration (OR) were 8.07 mm and 0.82, respectively. Despite fast computing times, this new method showed satisfying results, even at prostate base and apex. Also, we believe that this approach may allow delineating the peripheral zone (PZ) and the transition zone (TZ) within the gland in a near future.