A computer-aided detection system for automatic classification of prostate cancer from benign hyperplasia of prostate

Intelligent computer programs were implemented to aid physicians and other medical professionals in making difficulty medical decisions. In this paper, a computer aided detection (CAD) system is designed that classifies prostate neoplasia into two classes (cancer and hyperplasia) for supporting urologist. Early detection of Prostate Cancer (PCa) is crucial for patient survival. Automated diagnostic method could reduce diagnostic complications, financial and human resources. In presented system, data from 360 patients were used who diagnosed after biopsy with cancer and benign hyperplasia of prostate. In order to design present CAD, we have used; Bayesian regulation Backpropagation Neural networks (BRBNN). Performance of selected algorithm in detection of prostate cancer was evaluated by calculating sensitivity, specificity indicators and areas under receiver-operating characteristics (ROC) curve. The result of present study shows that, CAD based on BRBNN could serve as a strong diagnostic tool with 92.88% specificity, 90.62% sensitivity, 91.67% accuracy for early automatic detection of prostate cancer from benign prostate hyperplasia.