Comparison of PI-RADS and Likert Scoring Systems in the Diagnosis of Prostate Cancer and the Contribution of Radiologist Experience

Abstract

Purpose: The aim of this study was to investigate the concordance of these two scoring systems with histopathological data and the relationship between this concordance and radiologist experience. Materials and Methods: A total of 347 patients who underwent multiparametric prostate MRI (mpMRI) with a preliminary diagnosis of prostate cancer were retrospectively reviewed. The assessors independently scored the images according to PI-RADS v2.1. Two weeks later, they independently scored the images using the LIKERT system while blinded to their previous PI-RADS v2.1 scores. The study investigated the correlation of these scores with the pathology results and the inter-reader agreement. Results: The mean age of the patients was 65.5 +/- 7.7 years. In the kappa analysis, which evaluated the concordance of both scoring systems with the reference standard pathology, it was observed that concordance increased with radiologist experience. For the entire gland, the kappa values for readers 1, 2, 3, and 4 with PI-RADS v2.1 were found to be 0.669, 0.669, 0.711, and 0.771, respectively, and with the LIKERT system, they were 0.589, 0.669, 0.701, and 0.771, respectively. The AUC values were 0.901 (0.893-0.921) for PI-RADS and 0.895 (0.871-0.922) for LIKERT. Conclusion: The PI-RADS v2.1 and LIKERT scoring systems provided similar inter-reader agreement in evaluating mpMRI. Among less experienced radiologists, PI-RADS v2.1 demonstrated higher concordance with pathology, whereas no difference was observed between more experienced radiologists.