SAT0539 A New Method for The Quantitative Analysis of Hand Muscle Volume for Patients with Rheumatoid Arthritis. (15th July 2016)

• Record Type: Journal Article
• Authors: Friedberger, A.; Museyko, O.; Figueiredo, C.; Bäuerle, T.; Chaudry, O.; Grimm, A.; Schett, G.; Engelke, K.
• Is Part Of: Annals of the rheumatic diseases. Volume 75(2016)Supplement 2
• Page Start: 864
• Page End: 864
• Publication Date: 2016-07-15
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2016-eular.5519
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Rheumatoid arthritis (RA) is characterized by bone, cartilage and muscle loss in context of chronic inflammation. While bone and cartilage damage have been extensively studied in the past, the effects of RA on the muscles of the hand have not been studied. This situation is surprising, since visible hand muscle atrophy is a hallmark of RA and hand muscle composition determines the functional state of the patients function. Objectives: Development of a method to quantitatively assess hand muscles in patients with RA, using magnetic resonance imaging (MRI) and 3D segmentation of the hand muscles. Methods: T1 weighted MRI scans of healthy subjects and RA patients were used to develop a new segmentation method for hand muscles. Three groups (20 subjects each) were studied: (1) healthy subjects, (2) untreated RA patients and (3) DMARD-treated RA patients. The new approach uses a graph cuts based segmentation, which is initialized by fuzzy clustering. In a second step the segmented muscle volume of interest (VOI) is further refined and limited to the interosseal muscles and the thenar and hypothenar eminence. This segmentation procedure is fully automated. Nevertheless an extensive set of 3D editing tools such as hole filling, morphological operators and 3D surface manipulation is available to the operator, should the automatic approach not be satisfactory. Typically this happens in elderly diseased subjects with pathologic muscle and bone changes. The results of the automatic hand muscle segmentation were checked and confirmed by two experienced MRI readers. Results: The automatic segmentation procedure took about 3 min per patient. In about 70% of the cases the automatic procedure resulted in an accurate muscle segmentation, as judged by the experts. In 30% of the cases additional operator interaction was necessary, primarily to correct for over segmentation. On average, the user interaction could be completed within 2 min. Final segmentation results were confirmed again by the experts. The figure shows a slice of a T1 weighted fat suppressed MRI image in the metacarpal region. Top without and bottom with overlaid muscle VOI (red). The quantitative analysis of muscle volume and the determination of among group differences is currently being performed. Conclusions: We successfully developed a new accurate 3D-segmentation for hand muscles, which allows quantitative analysis of hand muscles in RA patients. Disclosure of Interest: None declared