A Novel Mobile Application to Assess and Track Fatigue in Multiple Sclerosis

Abstract

Background: Multiple sclerosis (MS) is a disabling neurologic disease, and fatigue is one of the most common symptoms. White matter hyperintensities (WMH) on magnetic resonance imaging (MRI) serve as a clinical marker for disease, but numerous widespread and microstructural changes develop throughout the disease course. When WMH acutely appear, they damage the surrounding white matter, affecting bundles of axons throughout their course. Objectives: My internship has 3 main objectives: (1) assessment of patient compliance and its relation to disease severity in a mobile application to assess real-time fatigue in MS, (2) determining when and how remodeling of callosal white matter tracts occurs following an active MS lesion occurring in the hemispheres, (3) creation of instructional materials for lesion identification, segmentation, and characterization workflows to be included in a web-based platform, SPINE, designed for experiments, with a specific focus on medical images. Methods: (1) In collaboration with Mobilengine, the Center for Neurological Imaging (CNI) created a mobile application containing 10 questionnaires focusing on fatigue and possible confounders, as well as 14 days of routine visual analog scales (VAS) and sleep diaries. Utilizing the medical record as well as patient submitted responses, we performed statistical analysis to determine whether severity of fatigue or disability affected compliance with our application. (2) In a cohort of 37 MS patients who underwent yearly MRI over a period of 3 years, we performed detailed lesion segmentation of all MS lesions in their brains. We then used a pipeline developed at the CNI to identify and assess the signal intensities of individual voxels of the midsagittal corpus callosum that contained white matter tracts that were: a) unaffected by MS lesions, b) affected by lesions in the corpus callosum, c) affected by lesions in the white matter tract, d) affected by cortical white matter lesions, e) affected by active lesions in the hemisphere. We have analyzed the first timepoint for any significant differences in T1-weighted signal intensity between these groups. (3) SPINE is currently under active development by the CNI. There are tools for identifying and segmenting lesions of interest currently implemented. Additionally, our group drafted scripts for videos to explain MS lesion identification, segmentation, and qualitative characterization using SPINE’s platform. Results: (1) We successfully developed and tested our application in a cohort of 64 MS patients. Fifty-six (87.5%) completed at least one question, with 47 (73.4%) completing the entire collection period. Fifty-one (79.7%) completed the entire one-time questionnaire, with 44 (86.3%) completing this within 72 hours. Nearly 4 VAS were entered per patient per day. No significant differences were found in age, sex, ethnicity, disease duration, or Expanded Disability Status Scale (EDSS) score between patients who did not begin, who began but did not complete the one time questionnaire, did not complete the entire 14-day VAS tracking period, or completed the entire duration of our study. No significant difference in Fatigue Severity Scale (FSS) score was found between the different groups of patients who completed this questionnaire in the application. (2) We show no consistently significant difference in T1-weighted signal intensity between voxels in the corpus callosum containing tracts that are unaffected by white matter lesions and those that are affected by hemispheric, cortical, or active lesions. (3) We have implemented tools for area of interest identification, segmentation, and recording of qualitative characteristics. We have drafted scripts for videos detailing the use of these tools in order to instruct other individuals to perform these tasks on white matter hyperintensities in MS. Conclusion: We describe here the development of a mobile application that is readily used by MS patients to assess and track fatigue over a 2-week period. We determined that severity of disability or fatigue in MS patients does not appear to affect compliance rates. In assessment for remodeling of white matter tracts affected by MS lesions, we have not found consistent differences in signal intensity between tracts affected by active lesions, chronic lesions, or unaffected by either. Following the modification of the pipeline to allow for assessment at two time-points, we will continue our analysis into whether tracts containing active lesions two years previously, or tracts that contain newly developed lesions in the prior two years, show evidence of remodeling. We also are continuing development of numerous workflows to be integrated into SPINE. For the MS lesion identification, segmentation, and characterization modules, we are beginning to create the videos detailing how to use SPINE’s built-in tools to accomplish these functions.