Automatic muscle segmentation of MRI images of mice with nnU-net

Motivation: Muscle or muscle groups segmentation is a crucial step in quantitative MRI pipeline to provide biomarkers of muscle disorders or aging but it is a time consuming step. Goal(s): The aim of this project is to leverage automatic muscle segmentation models in clinical research to translate them in preclinical research. Approach: We hypothesize that nnU-net, a deep-learning automatic segmentation framework, can be effectively used for studying aging and muscle disorders in mouse models. Results: Our results show that nnU-net model can be translated to mouse hind legs muscle segmentation. Impact: Our automatic muscle segmentation tool significantly speeds up the processing duration of MRI data and therefore the development of new quantitative MRI outcomes in the field of neuromuscular disorders.