Reliability and Reproducibility of a Novel MRI-Based Classification System for Lumbar Intervertebral Disc...

This study assess the interobserver and intraobserver reliability of a novel eight-grade MRI-based classification system for lumbar intervertebral disc degeneration and fix it as describe its reproducibility relative to the five-grade Pfirrmann classification. In this retrospective single-center reliability study, 250 consecutive adults undergoing lumbar MRI between May 2025 and May 2026 were included, providing 1250 intervertebral discs from L1-L2 through L5-S1. Four blinded observers with different levels of musculoskeletal and spine imaging experience independently graded all discs using both classification systems. All assessments were repeated after a 3-week washout period with randomized image order. Overall interobserver agreement was assessed using quadratically weighted Gwet's AC2 and intraobserver agreement using quadratically weighted Cohen's kappa. Confidence intervals were calculated using patient-level cluster bootstrap resampling. Each classification generated 5000 grading assessments. At the initial assessment, interobserver Gwet's AC2 was 0.767 (95% CI 0.756-0.779) for the novel classification and 0.817 (95% CI 0.808-0.826) for Pfirrmann. Intraobserver weighted kappa ranged from 0.436 to 0.810 for the novel classification and from 0.507 to 0.833 for Pfirrmann. The novel eight-grade classification showed reproducible grading in this cohort, although its interobserver agreement was numerically lower than that of Pfirrmann and intraobserver agreement varied across readers. These findings do not establish clinical validity or statistical equivalence between the systems.