Premium
Computer‐Aided Assessment of Spinal Inflammation on Magnetic Resonance Images in Patients With Spondyloarthritis
Author(s) -
Griffith James F.,
Wang Defeng,
Shi Lin,
Yeung David K. W.,
Lee Ryan,
Shan Tam Lai
Publication year - 2015
Publication title -
arthritis and rheumatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 4.106
H-Index - 314
eISSN - 2326-5205
pISSN - 2326-5191
DOI - 10.1002/art.39126
Subject(s) - medicine , magnetic resonance imaging , sagittal plane , ankylosing spondylitis , intraclass correlation , nuclear medicine , radiology , surgery , clinical psychology , psychometrics
Objective To evaluate a computer‐aided approach to the assessment of spinal inflammation on magnetic resonance imaging (MRI) in spondyloarthritis as compared to visual assessment. Methods Following institutional ethics approval, 32 patients (mean ± SD age 35.3 ± 10.2 years) with active ankylosing spondylitis underwent treatment with methotrexate and infliximab for 30 weeks. Spinal MRI examination (T1‐weighted, T2‐weighted fat‐suppressed, and postcontrast T1‐weighted fat‐suppressed sagittal sequences using a 1.5T MRI system) was performed at baseline and at 30 weeks. Following manual identification of the vertebral corners, vertebral body segmentation was performed using a deformable model that automatically isolated the 4 vertebral corner areas, allowing automatic measurement of the relative mean corner intensity of each vertebral corner before and after treatment. Quantitative computer‐aided assessment of spinal inflammation was compared with a semiquantitative visual assessment of spinal inflammation (the Berlin method). Results Computer‐aided quantification was quick and highly reliable, and it identified increases in vertebral corner edema or enhancement that were significantly decreased following treatment ( P < 0.5). For computer‐aided analysis, there was excellent inter‐ and intrarater correlation of both corner edema and enhancement (intraclass correlation coefficients [ICCs] >0.99), and the correlations were better than those for visual analysis (ICCs 0.83–0.96). For computer‐aided analysis, the standardized response mean was 1.67 for corner edema and 1.64 for enhancement, as compared to 1.20 and 1.18, respectively, for visual analysis. Computer‐aided quantification of MRI data correlated better (r = 0.50–0.53, P < 0.01) with clinical features of spinal disease activity pre‐ and posttreatment than did visual analysis of spinal inflammation (r = 0.37–0.43, P > 0.02). Conclusion Computer‐aided assessment of spinal inflammation in spondyloarthritis is quick, reliable, and sensitive and correlates better with clinical disease activity than does visual assessment by the Berlin method.