Improved operator agreement and efficiency using the minimum area contour change method for delineation of hyperintense multiple sclerosis lesions on FLAIR MRI

BMC Medical Imaging

Research Article

Sara Hussein¹  Christopher Magnano¹  Niels Bergsland¹  Deepa Ramasamy¹  Michael G Dwyer¹  David S Wack²  Robert Zivadinov³  Laura Ranza⁴  Carol Di Perri⁴  Kevin Seals⁵ 

[1] Buffalo Neuroimaging Analysis Center, Department of Neurology, School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York at Buffalo, Buffalo, NY, USA;Buffalo Neuroimaging Analysis Center, Department of Neurology, School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York at Buffalo, Buffalo, NY, USA;Department of Nuclear Medicine, University at Buffalo, State University of New York at Buffalo, Buffalo, NY, USA;Buffalo Neuroimaging Analysis Center, Department of Neurology, School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York at Buffalo, Buffalo, NY, USA;MR Imaging Clinical Translational Research Center, School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York, Buffalo, NY, USA;Department of Neuroradiology, IRCCS, C. Mondino, University of Pavia, Pavia, Italy;School of Medicine and Biomedical Sciences, State University of New York at Buffalo, Buffalo, NY, USA;

关键词: Multiple sclerosis;    Detection error;    Outline error;    Rater agreement;    Operator agreement;    Metric;    Jaccard index;    Similarity index;    Kappa;    Lesion;    Assessment;    Minimum area contour change;   

DOI  :  10.1186/1471-2342-13-29

 received in 2013-03-20, accepted in 2013-08-19,  发布年份 2013

来源: Springer

【 摘 要 】

BackgroundActivity of disease in patients with multiple sclerosis (MS) is monitored by detecting and delineating hyper-intense lesions on MRI scans. The Minimum Area Contour Change (MACC) algorithm has been created with two main goals: a) to improve inter-operator agreement on outlining regions of interest (ROIs) and b) to automatically propagate longitudinal ROIs from the baseline scan to a follow-up scan.MethodsThe MACC algorithm first identifies an outer bound for the solution path, forms a high number of iso-contour curves based on equally spaced contour values, and then selects the best contour value to outline the lesion. The MACC software was tested on a set of 17 FLAIR MRI images evaluated by a pair of human experts and a longitudinal dataset of 12 pairs of T2-weighted Fluid Attenuated Inversion Recovery (FLAIR) images that had lesion analysis ROIs drawn by a single expert operator.ResultsIn the tests where two human experts evaluated the same MRI images, the MACC program demonstrated that it could markedly reduce inter-operator outline error. In the longitudinal part of the study, the MACC program created ROIs on follow-up scans that were in close agreement to the original expert’s ROIs. Finally, in a post-hoc analysis of 424 follow-up scans 91% of propagated MACC were accepted by an expert and only 9% of the final accepted ROIS had to be created or edited by the expert.ConclusionWhen used with an expert operator's verification of automatically created ROIs, MACC can be used to improve inter- operator agreement and decrease analysis time, which should improve data collected and analyzed in multicenter clinical trials.

【 授权许可】

CC BY   
© Wack et al.; licensee BioMed Central Ltd. 2013