MRI-derived 3D lower limb muscle shape: a biomarker for disease severity in Duchenne muscular dystrophy?

Duchenne muscular dystrophy (DMD) is characterized by progressive muscle degeneration leading to loss of gait. Identification of predictive biomarkers of gait loss is crucial, yet analysis of muscle shape remains underexplored. Using MRI data from 17 children with DMD (10 followed by gait loss) and 10 healthy controls, we analyzed the muscle shapes and volumes of the lower limbs. Correlations with functional metrics (gait tests, dynamometric forces) were evaluated, and predictive modeling tasks (classification and regression) were performed via random forests. Compared with those of the controls, the key muscle groups, particularly the soleus and gastrocnemius, of DMD patients presented distinct patterns, including increased thickness and reduced extensibility, without significant volume differences. Compared with traditional biomarkers, predictive models based on shape descriptors achieved 100\% accuracy in distinguishing controls from DMD patients and reduced the error in predicting time to gait loss to 110 days. This study highlights the potential of muscle shape analysis as a reproducible and acquisition-independent biomarker. This study provides novel insights into the progression of DMD, highlighting early changes in the superficial muscles of the posterior compartment as robust indicators of functional decline. Despite the small cohort, these findings support the integration of geometric biomarkers into clinical monitoring.