Diffusion MRI predicts transrectal ultrasound biopsy results in prostate cancer detection

Abstract

To evaluate the ability of diffusion tensor imaging (DTI) to predict the transrectal ultrasound (TRUS) biopsy outcomes in persons who have no history of previous TRUS biopsy and present with elevated prostate-specific antigen (PSA) levels. Thirty-seven participants underwent DTI, followed by 12-core TRUS-guided needle biopsy within 2 weeks. DTI was performed using endorectal coils on a 1.5 Tesla scanner at 1-mm(3) spatial resolution. By comparing with the TRUS biopsy results, the optimum thresholds of the trace apparent diffusion coefficient (tADC) and of the nodular size were investigated. The diagnostic performance of both criteria, the tADC threshold (Criteria A) and the tADC threshold combined with nodular size threshold (Criteria B), were evaluated. The optimum tADC threshold was 1.0 μm(2)/ms. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of prostate cancer (PCA) detection for Criteria A were 98%, 89%, 73%, 99%, and 91%, respectively, and were 97%, 98%, 92%, 99% and 98% for Criteria B. Owing to high negative predictive value, the tADC threshold could be used to exclude subjects with clinically undetectable PCA. Adding the nodular size threshold, the combined threshold could identify the tADC-positive segments that are likely to yield positive biopsy results.