Quantitative Analysis of Muscle Ultrasound Images in Adult Patients With Pompe Disease Using Radiomics

Lucie Le Calvé, Loïc Duron, Luc Defebvre, Romain Thomas, Loic Danjoux, Céline Tard, Jean‐Baptiste Davion
Muscle and Nerve · Wiley · 2026-11-19 · DOI: 10.1002/mus.70462

Abstract

ABSTRACT Introduction/Aims Pompe disease is a genetic myopathy caused by acid alpha‐glucosidase enzyme deficiency. Treatment response is mainly evaluated using motor scores, for instance the 6‐min walk test (6MWT) or the Motor Function Measure‐32 (MFM‐32), which can be time‐consuming and biased by fatigue and motivation. Quantitative muscle ultrasound may be an additional biomarker for monitoring disease progression. Our objective was to study the performance of a large number of quantitative features derived from muscle ultrasound images to predict motor involvement in late‐onset Pompe disease. Methods We retrospectively collected muscle ultrasound images from 19 patients with late‐onset Pompe disease and 28 control subjects, and calculated 93 quantitative features for each image. Using a radiomics approach, we selected reproducible features that were significantly different between patients and controls, and then assessed their predictive power for the 6MWT and the MFM‐32 score. Results Among the 93 features, 36 (38.7%) showed good reproducibility, and 18 (19.3%) were significantly different between Pompe patients and controls after adjustment for age, sex, and body mass index. The most predictive features of motor functions were the grayscale 10th percentile in the rectus femoris, biceps brachii, deltoid, and extensor carpi radialis muscles, explaining 76% of the variance of the 6MWT, and 84% of the variance of the MFM‐32 score. Discussion These findings confirm the association between muscle ultrasound and motor function in Pompe disease found by previous studies and suggest that radiomics can identify additional predictive features, such as the grayscale 10th percentile.

Read at publisher →

Similar articles