The Impact of Fatty Infiltration on MRI Segmentation of Lower Limb Muscles in Neuromuscular Diseases: A Comparative Study of Deep Learning Approaches

Abstract

BackgroundDeep learning methods have been shown to be useful for segmentation of lower limb muscle MRIs of healthy subjects but, have not been sufficiently evaluated on neuromuscular disease (NDM) patients.PurposeEvaluate the influence of fat infiltration on convolutional neural network (CNN) segmentation of MRIs from NMD patients.Study TypeRetrospective study.SubjectsData were collected from a hospital database of 67 patients with NMDs and 14 controls (age: 53 ± 17 years, sex: 48 M, 33 F). Ten individual muscles were segmented from the thigh and six from the calf (20 slices, 200 cm section).Field Strength/SequenceA 1.5 T. Sequences: 2D T1‐weighted fast spin echo. Fat fraction (FF): three‐point Dixon 3D GRE, magnetization transfer ratio (MTR): 3D MT‐prepared GRE, T2: 2D multispin‐echo sequence.AssessmentU‐Net 2D, U‐Net 3D, TransUNet, and HRNet were trained to segment thigh and leg muscles (101/11 and 95/11 training/validation images, 10‐fold cross‐validation). Automatic and manual segmentations were compared based on geometric criteria (Dice coefficient [DSC], outlier rate, absence rate) and reliability of measured MRI quantities (FF, MTR, T2, volume).Statistical TestsBland–Altman plots were chosen to describe agreement between manual vs. automatic estimated FF, MTR, T2 and volume. Comparisons were made between muscle populations with an FF greater than 20% (G20+) and lower than 20% (G20−).ResultsThe CNNs achieved equivalent results, yet only HRNet recognized every muscle in the database, with a DSC of 0.91 ± 0.08, and measurement biases reaching −0.32% ± 0.92% for FF, 0.19 ± 0.77 for MTR, −0.55 ± 1.95 msec for T2, and − 0.38 ± 3.67 cm3 for volume. The performances of HRNet, between G20− and G20+ decreased significantly.Data ConclusionHRNet was the most appropriate network, as it did not omit any muscle. The accuracy obtained shows that CNNs could provide fully automated methods for studying NMDs. However, the accuracy of the methods may be degraded on the most infiltrated muscles (>20%).Evidence Level4.Technical EfficacyStage 1.

A Comprehensive Review of the Impact of Machine Learning and Omics on Rare Neurological Diseases

“…Radiomics studies, enhanced by ML, have improved the glioblastoma biopsy guidance and differentiated brain metastases from glioblastoma [79,80] Convolutional neural networks have been used to detect fatty infiltration in neuromuscular diseases, with HRNet being the most effective [81]. Finally, ML regression models have been employed to predict the muscle fat fraction in FSHD, aiding in disease progression assessment [82]. These advancements in AI and ML are transforming the landscape of diagnosis, prognosis, and treatment in rare diseases.…”

Editorial for “The Impact of Fatty Infiltration on MRI Segmentation of Lower Limb Muscles in Neuromuscular Diseases: A Comparative Study of Deep Learning Approaches”

“…In this issue of JMRI , Hostin et al utilize four different convolutional neural network (CNN) methods to automatically segment the same lower extremity muscle MRI dataset, which includes both control subjects and patients with several disparate NMD 8 . Additionally, they measure several qmMRI parameters, namely fat fraction (FF), magnetic transfer ratio (MTR), and T2 values 8 .…”

“…In this issue of JMRI , Hostin et al utilize four different convolutional neural network (CNN) methods to automatically segment the same lower extremity muscle MRI dataset, which includes both control subjects and patients with several disparate NMD 8 . Additionally, they measure several qmMRI parameters, namely fat fraction (FF), magnetic transfer ratio (MTR), and T2 values 8 . Their work is important for a few reasons: 1) it utilizes completely automatic segmentation, which is required to decrease the time extract important quantitative data, allowing future implementation in a clinical setting or large clinical trial; 2) it includes a varied population of NMD patients to more realistically assess the accuracy of segmentation with different degrees of disease; 3) it suggests an objective measure of FF, namely greater than 20%, where the error rates in segmentation and measurement of muscle parameters are likely too great for accurate automated assessment 8 .…”

“…Additionally, they measure several qmMRI parameters, namely fat fraction (FF), magnetic transfer ratio (MTR), and T2 values 8 . Their work is important for a few reasons: 1) it utilizes completely automatic segmentation, which is required to decrease the time extract important quantitative data, allowing future implementation in a clinical setting or large clinical trial; 2) it includes a varied population of NMD patients to more realistically assess the accuracy of segmentation with different degrees of disease; 3) it suggests an objective measure of FF, namely greater than 20%, where the error rates in segmentation and measurement of muscle parameters are likely too great for accurate automated assessment 8 . Although their data will need to be confirmed in other populations of NMD patients, Hostin et al are helping to refine qmMRI and move it closer to replacing total muscle volume, which has been used in clinical trials, 9 as a measure of muscle pathology severity in future clinical trials.…”