MRI reliability in classifying thoracolumbar fractures according to AO classification

Area of Science:

• Orthopedics
• Radiology
• Spine Surgery
• Medical Imaging

Background:

• The AO classification system is a standard for categorizing vertebral fractures.
• Traditional imaging methods like radiography and computed tomography (CT) have shown limited reproducibility in AO classification.
• Posterior ligamentous complex damage, crucial for fracture stability, is better visualized with MRI.

Purpose of the Study:

• To evaluate the inter- and intraobserver reliability of the AO classification for traumatic vertebral fractures using magnetic resonance imaging (MRI).
• To compare the reproducibility of MRI with historical data from radiography and CT.
• To assess the impact of evaluator experience and familiarity on classification accuracy.

Main Methods:

• Retrospective analysis of 37 traumatic vertebral fractures from a prospective cohort.
• Fractures were classified using both radiographs and fat-saturated MRI sequences according to the AO classification.
• Five spine surgeons, 2 orthopedic residents, 2 musculoskeletal radiologists, and 2 radiodiagnosis residents independently classified fractures in two separate sessions, 6 weeks apart.

Main Results:

• Mean interobserver agreement for all evaluators was kappa=0.53 (session 1) and kappa=0.47 (session 2).
• Senior spine surgeons demonstrated higher interobserver agreement (kappa=0.59-0.54) than residents (kappa=0.45-0.31).
• Orthopedic surgeons showed better agreement (kappa=0.71) than radiologists (kappa=0.48). Mean intraobserver agreement was kappa=0.58.
• MRI yielded moderate reproducibility, slightly superior to reported CT values.

Conclusions:

• Magnetic resonance imaging offers moderate reproducibility for AO classification of vertebral fractures.
• Reliability is influenced by evaluator experience, with senior spine surgeons showing better agreement.
• Familiarity with the AO classification system positively impacts diagnostic accuracy.