Recognition Algorithm for Prostate Calculus Based on Transition Region and PCA-SVM

Recognition of prostate calculus is an important step to determine the source of pathological organ, and is of great importance for further diagnosis of prostate cancer. In this paper, due to some tissues are similar to calculus, and prostate calculus usually adheres to other tissues, a recognition algorithm for prostate calculus based on transition region and PCA-SVM is proposed. Firstly, local entropy, Otsu method and morphology are used to extract transition region. Then suspicious calculus regions are segmented from each independent transition region. Finally, taking domain knowledge and texture features of each suspicious calculus into account, a PCA-SVM model is implemented for the automatic prostate calculus recognition. Our experiment result shows that the proposed algorithm for prostate calculus segmentation and recognition has a satisfactory robustness and accuracy.