Background and aim: Neuromuscular diseases encompass a range of conditions affecting muscle cells, nerves, or the interaction between the two. A common pathological feature of these conditions is the pro-gressive replacement of muscle tissue with fat, which can be visualised using magnetic reso-nance imaging (MRI). MRI-based fat quantification serves as a key biomarker for disease characterisation, progression tracking, and treatment assessment. Currently, manual segmenta-tion of MRI scans for fat quantification is very time-consuming, requiring individual muscle delineation. Therefore, an artificial intelligence (AI) model is being developed to automate the segmentation. The aim of this study is to validate this AI model and assess its possibilities and limitations. Method: The study is ongoing. Retrospective MRI scans of patients with four different muscle diseases (anoctaminopathy, Becker muscular dystrophy, facioscapulohumeral muscular dystrophy, and hypokalemic periodic paralysis) are collected and manual delineation used for training the AI-model is being performed. The intramuscular fat fraction of individual muscles of the pelvis, thigh, and calf will be analysed using the AI model. The performance of the AI model will be compared to manual segmentation. The AI will be evaluated on metrics such as segmentation accuracy and time efficiency.
See this in plain English?
AI-rewrites the medical criteria so a patient or caregiver can understand them. Always confirm with the trial site.
Difference in fat fraction between manual and AI outlining.
Timeframe: Analysis of the muscle fat fraction takes 1 hour per patient.