CT Prostate Segmentation Based on Random Forest

Due to the low contrast of CT image, uncertain position of prostate and variable bowel gas, automatic and accurate CT prostate segmentation is a challenging task. In this paper, a prostate identification method is proposed based on random forest. Using the auto-context model, a sequence of Random Forest classifiers is trained. Then the trained classifiers are applied on the testing image to improve the classification response map iteratively. The experimental results show that, compared with other state-of-the-art methods, our method achieves a better performance.