From medical scans to muscle biomarkers: MuscleMap drives early disease detection

Musculoskeletal and neurological conditions such as osteoporosis, Parkinson’s Disease or chronic fatigue syndrome affect 4 billion people worldwide. As populations grow and age, the prevalence of these conditions will rise sharply.

Quantitative muscle imaging offers a powerful but underused opportunity to detect disease earlier.

Using NIF’s expertise and MRI and CT instruments – plus other instruments around the world – NIF user and collaborator Prof James Elliott (Director of the Kolling Institute, and Academic Director of Allied Health and Public Health at The University of Sydney) is addressing this opportunity along with collaborative colleagues A/Prof Marnee McKay (University of Sydney), and Dr Kenneth Weber and Dr Eddo Wesselink (Stanford University).

Changes in muscle size, shape, and composition often occur years before obvious clinical symptoms emerge in a wide range of conditions. Those conditions can include musculoskeletal disorders, neurological disease, metabolic and systemic illness, and ageing-related functional decline.

Early diagnosis and treatment for these conditions can drastically minimise the health burden on Australians. But MRI and CT scans – which can detect declines – generally take too long.

Putting numbers to muscle health via muscle fibres, size, fat content, and other measures could help alert clinicians earlier.

MuscleMap, a productive collaborative co-lead by Weber, Wesselink, McKay and Elliott, is filling this gap. It is an open-source platform for whole-body muscle segmentation and muscle-composition quantification.

Its innovative model has now outgrown the imaging facilities at Stanford and the University of Sydney.

MuscleMap draws on data from tens of thousands of scans and analyses using AI, enabling muscle biomarkers to be interpreted relative to population-specific normative ranges.

This allows clinicians and researchers to distinguish between healthy variation and clinically meaningful muscle decline – an essential step toward precision medicine.

“From athletes to astronauts and the general population, MuscleMap will accurately inform the journey from injury to repair, recovery and return to physical activity, informing personalised strategies to develop healthier muscle mass and improve general health and wellbeing,” said Prof Elliott.

The long-term vision is for MuscleMap to operate as a true “push-button system” – where a clinician will perform the scan, and within minutes receive a comprehensive set of muscle-health metrics that can easily be interpreted in a clinical context.

The planned accessibility of those metrics for clinicians breaks down barriers to the comprehensive assessment of muscle health.

MuscleMap’s format supports this goal: it is open source and freely accessible.

Over the past 5 years, it has grown into a global consortium of more than 60 collaborators. Contributors upload their own datasets, helping to model healthy muscle across diverse populations worldwide.

Recently, MuscleMap has just released an updated v1.2 of the model, trained on an expanded community-sourced dataset, that can automatically quantify 100 muscles and bones.

The team are encouraging users to report any issue on a GitHub page, created specifically to report and solve issues with MuscleMap v1.2, or to share data and provide general feedback.

Prof Elliott says: “In this tutorial, we show the basics of using MuscleMap’s muscle quantification tool. You’ll learn how to run the mm_extract_metrics command-line utility to automatically extract quantitative muscle metrics—such as volume, cross-sectional area, intramuscular fat, and density from MRI or CT scans after segmentation. We cover the essential inputs, options, and workflow so you can start analyzing your imaging data with MuscleMap.”