Interobserver Reproducibility of the PI-RADS Version 2 Lexicon: A Multicenter Study of Six Experienced Prostate
Andrew B Rosenkrantz1, Luke A Ginocchio1, Daniel Cornfeld1
1From the Department of Radiology, NYU School of Medicine, NYU Langone Medical Center, 660 1st Ave, Third Floor, New York, NY 10016 (A.B.R., L.A.G., J.S.B.); Department of Radiology, Yale School of Medicine, New Haven, Conn (D.C.); Department of Radiology, Mayo Clinic, Rochester, Minn (A.T.F.); Department of Radiology, Duke University Medical Center, Duke Cancer Institute, Durham, NC (R.T.G.); Molecular Imaging Program, National Cancer Institute, National Institutes of Health, Bethesda, Md (B.T.); Departments of Radiology and Biomedical Imaging and Urology, University of California-San Francisco, San Francisco, Calif (A.C.W.); and Department of Radiology, David Geffen School of Medicine at UCLA, Los Angeles, Calif (D.J.M.).
Experienced radiologists achieved moderate reproducibility using the Prostate Imaging Reporting and Data System (PI-RADS) version 2 lexicon for prostate MRI. Agreement was better in the peripheral zone than the transition zone, with diffusion-weighted imaging showing higher reliability.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Prostate cancer diagnosis relies on accurate interpretation of MRI.
- The Prostate Imaging Reporting and Data System (PI-RADS) version 2 provides a standardized lexicon for reporting prostate MRI findings.
- Assessing interobserver reproducibility is crucial for validating diagnostic tools.
Purpose of the Study:
- To evaluate the interobserver reproducibility of the PI-RADS version 2 lexicon among experienced radiologists.
- To compare reproducibility between different prostate zones (peripheral and transition zones).
- To identify features with higher or lower agreement for potential lexicon refinement.
Main Methods:
- Retrospective analysis of prostate MRI examinations from a single center.
- Six experienced radiologists independently assessed lesions using the PI-RADS v2 lexicon.
- Two assessment sessions were conducted with an intersession training period.
- Interobserver agreement was quantified using percent agreement and kappa (κ) coefficients.
Main Results:
- Overall moderate interobserver agreement was achieved for PI-RADS version 2 (κ values not specified).
- Agreement for PI-RADS score ≥4 was 0.593 in the peripheral zone (PZ) and 0.509 in the transition zone (TZ).
- Reproducibility was moderate to substantial for diffusion-weighted imaging features in PZ (κ = 0.535-0.619), but fair to moderate for dynamic contrast-enhanced (DCE) imaging (κ = 0.266-0.439).
- Agreement for T2-weighted image features in TZ varied widely (κ = 0.136-0.529).
- Majority PI-RADS ≥4 classification correctly identified high-grade tumors (Gleason score ≥3+4) in 45.9% of cases without missing any.
Conclusions:
- Experienced radiologists demonstrate moderate reproducibility with the PI-RADS v2 lexicon.
- No significant benefit was observed from the training session.
- Reproducibility is generally better in the PZ than TZ, with notable weakness in DCE imaging assessment in the PZ.
- Findings support potential areas for PI-RADS lexicon refinement.
More Related Videos
06:48Lexical Decision Task for Studying Written Word Recognition in Adults with and without Dementia or Mild Cognitive Impairment
Published on: June 25, 2019
08:17A Semantic Priming Event-related Potential ERP Task to Study Lexico-semantic and Visuo-semantic Processing in Autism Spectrum Disorder
Published on: April 12, 2018