Disease duration and disability in dysfeRlinopathy can be described by muscle imaging using heatmaps and random forests.
Introduction
The manner in which imaging patterns change over the disease course and with increasing disability in dysferlinopathy is not fully understood.
Methods
Fibroadipose infiltration of 61 muscles was scored based on whole-body MRI of 33 patients with dysferlinopathy and represented in a heatmap. We trained random forests to predict disease duration, Motor Function Measure dimension 1 (MFM-D1), and modified Rankin scale (MRS) score based on muscle scoring and selected the most important muscle for predictions.
Results
The heatmap delineated positive and negative fingerprints in dysferlinopathy. Disease duration was related to infiltration of infraspinatus, teres major-minor, and supraspinatus muscles. MFM-D1 decreased with higher infiltration of teres major-minor, triceps, and sartorius. MRS related to infiltration of vastus medialis, gracilis, infraspinatus, and sartorius.
Discussion
Dysferlinopathy shows a recognizable muscle MRI pattern. Fibroadipose infiltration in specific muscles of the thigh and the upper limb appears to be an important marker for disease progression. Muscle Nerve 59:436-444, 2019.
© 2018 Wiley Periodicals, Inc.
Overview publication
Title | Disease duration and disability in dysfeRlinopathy can be described by muscle imaging using heatmaps and random forests. |
Date | 2019-04-01 |
Issue name | Muscle & nerve |
Issue number | v59.4:436-444 |
DOI | 10.1002/mus.26403 |
PubMed | 30578674 |
Authors | |
Keywords | disability, dysferlin, heatmap, machine learning, muscle imaging, random forest |
Read | Read publication |