Novel automated system for Magnetic Resonance Imaging quantification of the inflamed synovial membrane volume in patients with juvenile idiopathic arthritis.

Objective To introduce a novel automated method for the quantification of the inflamed synovial membrane volume (SV) using magnetic resonance imaging (MRI), and to investigate its feasibility and validity in patients with juvenile idiopathic arthritis (JIA). Methods The tool was tested on 58 patients with JIA and wrist involvement. Thirty-six patients had a 1-year MRI followup. MRI of the clinically more affected wrist was performed using a 1.5T scanner and a Flex small coil. An algorithmic approach, based on supervised voxel classification for automatic estimation of SV in a 3-dimensional MRI, was developed. The SV was estimated as the number of positively classified voxels and then normalized by the patient's body surface (NSV). Validation procedures included the analysis of reliability, construct validity, responsiveness to change, discriminant validity, and the predictive value. Results The agreement between the automated estimation of NSV and the manual measurements was excellent (intraclass correlation coefficient 0.93, 95% confidence interval 0.79-0.98). The automatic NSV demonstrated good construct validity by yielding strong correlations with local signs of disease activity and a moderate correlation with global physician assessment of disease activity and with the Rheumatoid Arthritis Magnetic Resonance Imaging Scoring system synovitis score. NSV showed a strong responsiveness to clinical change (standardized response mean values >1) and satisfactory discriminant validity. High baseline NSV (>4.6) had high predictive value (100%) with respect to erosive progression. Conclusion The proposed automated method allowed reliable quantification of NSV, which represents a promising imaging biomarker of disease activity in JIA. The automated system has the potential to improve the longitudinal assessment of JIA and to predict progressive joint destruction.