SegmentAnyMuscle: A universal muscle segmentation model across different locations in MRI.
Key claim: SegmentAnyMuscle segments skeletal muscle across diverse anatomical regions and MRI sequences (mean DSC 88.45%; 86.21% on challenging cases) and yields automatic muscle-mass measures that correlate >0.99 with expert manual assessment.
Abstract
Muscle mass and muscle quality, characterized by intramuscular adiposity and fiber composition, are clinically established biomarkers significantly associated with obesity, sarcopenia, frailty and cardiometabolic disorders. Current muscle quantification efforts are predominantly based on CT, often focusing on single anatomical region such as the L3 vertebral level, due it its relatively standardized acquisition protocols and the overall simplicity of developing muscle segmentation algorithms. However, MRI offers superior soft-tissue contrast for visualizing intramuscular fat and enables muscle assessment across a wider range of body regions. In this study, we developed SegmentAnyMuscle, a model capable of segmenting muscles across diverse anatomical regions and various MRI sequences, addressing the barriers to clinical muscle assessment. By developing upon a robust, expert-curated dataset of 316 MRI exams from 160 patients covering 11 anatomical regions, SegmentAnyMuscle demonstrated high segmentation performance with an averaged Dice Similarity Coefficient (DSC) of 88.45% on routinely acquired MRI sequences and maintained strong performance (DSC of 86.21%) under clinical challenging cases, including less common sequences, abnormalities such as muscular atrophy, implant-associated artifacts, and significant noise. Furthermore, automatic skeletal muscle mass measurement provided by SegmentAnyMuscle showed excellent correlation with expert manual assessments (Pearson correlation coefficient >0.99, p < 0.001). By providing the first reliable, automatic muscle segmentation tool across diverse anatomical regions and MRI acquisition settings, SegmentAnyMuscle can open new opportunities for MRI-based muscle health assessment and encourages broader indicators of metabolic and functional health. We made codes and model publicly available at: https://github.com/mazurowski-lab/SegmentAnyMuscle.
Why this matters
Muscle-mass and muscle-quality imaging is a candidate frailty/sarcopenia biomarker layer, but adoption has been bottlenecked by manual segmentation. A publicly released model that generalizes across 11 anatomical regions and multiple MRI sequences with near-perfect agreement to expert muscle-mass measures lowers that bottleneck — but accuracy against expert tracings is not the same as predicting outcomes, and the clock-movement trap applies here in an imaging form: automating a measurement does not validate its prognostic meaning.