Receiver operating characteristic (ROC) curve analysis of transrectal ultrasonographic prostate cancer detection

PURPOSE: To provide an overview of receiver operating characteristics (ROC) curve theory and create an ROC curve describing transrectal ultrasound (TRUS) detection of prostate cancer (PCa). METHODS: One hundred thirty-two patients with either an abnormal digital rectal exam (DRE) or a prostate-specific antigen (PSA) above 4 ng/ml or both underwent TRUS and biopsy of the peripheral zone (PZ) of each quadrant. ROC software was used to create an ROC curve. RESULTS: One hundred seventy-nine of 528 quadrants (34%) harbored PCa. The performance of TRUS in detection of PCa as defined by the area below the ROC is 0,809. CONCLUSIONS: Future investigators are encouraged to employ ROC analysis of TRUS to permit more meaningful comparisons of performance.