Introduction: In multiparametric MRI (mpMRI), there is an inter-intrareader variability according to Prostate Imaging Reporting and Data System (PI-RADS) Version 2.1. We have evaluated the inter-intrareader assessments with PI-RADS to improve the detection rate of prostate cancer from MRI/TRUS software fusion prostate biopsy Methods: We evaluated 210 patients who underwent multiparametric MRI for MRI/TRUS software fusion prostate biopsy from October 2019 to February 2021. Examinations were interpreted by four readers including two radiologists and two urologists from our institution. The readers independently viewed all 210 examinations, and the 2.1 version of the PI-RADS grading scale was used to interpret MRI images. The interpretations were performed in two sessions. After the first session, a training intervention was conducted to improve inter-intrareader agreement with PI-RADS scores of the peripheral zone (PZ) and transitional zone (TZ). We used the Fleiss’s kappa coefficient (?) to evaluate the inter-intrareader reliability. Results: The ? between all readers were 0.427 (P < 0.001) in the first session, and 0.466 (P < 0.001) in the second session, which shows a moderate level of agreement. Higher agreement was observed among the two radiologists, with the ? of 0.563 (P < 0.001) in the first session, and 0.708 (P < 0.001) in the second session. Interreader agreement had a tendency to improve in the second session, after the training intervention took place. For all readers, the level of agreement was also stronger in the peripheral zone compared to the the transitional zone. Patients were categorized according to the PI-RADS score of 4 or more points and less than 3 points, and the two groups were compared. The ? between all readers for the peripheral zone was 0.761 (P < 0.001) in the first session, and 0.599 (P < 0.001) in the second session. This shows a higher agreement in the scoring of the peripheral zone. In the radiologists group, the level of agreement for PZ was 0.791 (P < 0.001) and 0.733 (P < 0.001), showing a substantial level of agreement. We evaluated the intrareader agreement of the two individual radiologists. The ? was higher when the PIRADS score was 4 or higher, achieving a substantial level of agreement, while interpretation became heterogeneous when PIRADS score was lower. Conclusions: Readers achieved moderate agreement for PI-RADS version 2.1 and benefitted from a training session. Scoring the PZ achieved better agreement than the TZ. Interreader agreement was stronger when PI-RADS score was 4 or higher. Interreader agreement can be improved with feedback and training. SOURCE OF Funding: None
