SAT0260 A Novel MRI-Based Analysis Workflow for Chronic Recurrent Multifocal Osteomyelitis (CRMO). (15th July 2016)

• Record Type: Journal Article
• Authors: Chu, Y.; Roettger, D.; Trentin, C.; Hinton, M.; Kubassova, O.; Hargunani, R.; Boesen, M.; O'Donnell, P.
• Is Part Of: Annals of the rheumatic diseases. Volume 75(2016)Supplement 2
• Page Start: 762
• Page End: 763
• Publication Date: 2016-07-15
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2016-eular.4182
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Chronic recurrent multifocal osteomyelitis (CRMO) is an auto-inflammatory bone disorder affecting children and adolescents, characterised by sterile bone inflammation. There is a clinical spectrum, ranging from presentation with a single, non-recurrent bone lesion to a chronic relapsing course affecting multiple, often symmetrical bone sites. Whole body MR imaging may reveal multiple lesions and therefore is promising for assessing progression of CRMO and effectiveness of treatment. Nevertheless, no tool has been available for quantitative measurement of multiple lesions in whole body MRI scans, and a fast, reproducible scoring system for CRMO patient assessment is warranted. Objectives: Methods: Whole-body static MRI STIR images from 4 patients were analysed by 2 different readers. The workflow for each reading was: 1) Define reference/baselines in fat, bone and muscle region in both left and right side of the body, respectively. 2) Draw regions of interests (ROIs) around lesions defined as areas of marrow hyper intensity. 3) Compute, Norml parametric maps by comparison with corresponding baseline. NormI have previously shown to be useful in semi-automatic quantification of inflammatory lesions in the spine on STIR or DWI-MRI sequences [1]. 4) Group the selected 2D ROIs across slices into volume of interests (VOIs). 5) Generate and save report - automated generation of lesion ROI parameters, including the sum, maximum and mean of all positive pixel values in the VOIs (SumNormI, MaxNormI and MeanNormI). Inter-reader agreement for SumNormI and MeanNormI was calculated using the intraclass correlation coefficient for each VOI. Results: Single or multiple lesions seen in whole body MRI scans of CRMO patients were identified and quantified using NormI parametric maps. As shown in figure, quantification parameters for the lesions identified in right femur and right clavicle have been computed by normalising to the baseline in muscle. Reproducibility assessed by ICC is shown in table. Conclusions: The suggested workflow and postprocessing technique provides a relatively fast and reproducible scoring method for both single and multiple lesions in CRMO from wholebody STIR MRI scans. Larger trials to test day-to-day variation and sensitivity to change following treatment response are warranted and currently ongoing. References: D. Roettger, et al. A novel MRI-based analysis workflow for spinal lesions in Ankylosing Spondylitis, in Proceeding of 15th Annual European Congress of Rheumatology EULAR, June 11–14, 2014. Disclosure of Interest: Y. Chu Employee of: Image Analysis Ltd., D. Roettger Employee of: Image Analysis Ltd., C. Trentin Employee of: Image Analysis Ltd., M. Hinton Employee of: Image Analysis Ltd., O. Kubassova Employee of: Image Analysis Ltd., R. Hargunani: None declared, M. Boesen Consultant for: Image Analysis Limited as Head of Clinical Advisory Board, P. O'Donnell: None declared