Quantifying synovitis in rheumatoid arthritis using computer-assisted manual segmentation with 3 Tesla MRI scanning

Area of Science:

• Rheumatology
• Radiology
• Medical Imaging

Background:

• Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by joint inflammation.
• Accurate assessment of synovitis is crucial for monitoring RA disease activity and treatment response.
• Current methods for quantifying synovitis may have limitations in reliability or accessibility.

Purpose of the Study:

• To evaluate the reliability, validity, and feasibility of a computer-assisted manual segmentation technique for measuring synovial membrane volume in RA patients.
• To assess synovial membrane volume as a surrogate marker for synovitis.
• To compare the segmentation method with the established Rheumatoid Arthritis MRI Scoring System (RAMRIS).

Main Methods:

• Acquired 3 Tesla (T) MRI scans from 22 early RA and 16 established RA patients.
• Two non-radiologist observers manually segmented synovial membrane volumes in postcontrast T1w axial images of the wrist joint.
• Calculated intraobserver and interobserver reliability using intraclass correlation coefficients (ICCs).

Main Results:

• Excellent intraobserver reliability (ICC = 0.99) was observed in both early and established RA cohorts.
• Moderate to excellent interobserver reliability (ICCs = 0.87-0.88) was achieved.
• Strong correlations were found between segmented synovial membrane volumes and RAMRIS synovitis scores (Spearman rho = 0.85-0.96).
• The segmentation process averaged 19-21 minutes per patient.

Conclusions:

• Computer-assisted manual segmentation offers excellent intraobserver and very good interobserver reliability for measuring MRI-detected synovitis in RA.
• The method demonstrates strong content validity, correlating well with RAMRIS synovitis scores.
• The technique shows moderate feasibility for clinical application in assessing RA synovitis.