A computer-aided detection system for rheumatoid arthritis MRI data interpretation and quantification of synovial activity.

RATIONAL AND
OBJECTIVE: Disease assessment and follow-up of rheumatoid arthritis (RA) patients require objective evaluation and quantification. Magnetic resonance imaging (MRI) has a large potential to supplement such information for the clinician, however, time spent on data reading and interpretation slow down development in this area. Existing scoring systems of especially synovitis are too rigid and insensitive to measure early treatment response and quantify inflammation. This study tested a novel automated, computer system for analysis of dynamic MRI data acquired from patients with RA, Dynamika-RA, which incorporates efficient data processing and analysis techniques.
MATERIALS AND METHODS: 140 MRI scans from hands and wrists of 135 active RA patients and 5 healthy controls were processed using Dynamika-RA and evaluated with RAMRIS. To reduce patient motion artefacts, MRI data were processed using Dynamika-RA, which removed motion in 2D and 3D planes. Then synovial enhancement was visualised and qualified using a novel fully automated voxel-by-voxel analysis based algorithm. This algorithm was used to replace traditional region-of-interest (ROI) and subtraction methods, yielding observer independent quantitative results.
RESULTS: Conventional scoring performed by an observer took 30-45 min per dataset. Dynamika-RA reduced motion artefacts, visualised inflammation and quantified disease activity in less than 3 min. Data processing allowed increasing signal to noise ratio by a factor 3. Due to fully automated procedure of data processing, there was no interest variation in the results.
CONCLUSIONS: Algorithms incorporated into Dynamika-RA allow for the significant enhancement of data quality through eliminating motion artefacts and reduction of time for evaluation of synovial inflammation. Copyright (c) 2009 Elsevier Ireland Ltd. All rights reserved.