Three-dimensional ultrasound shape analysis reveals distinct structural patterns in muscle aging and sarcopenia

BACKGROUND: Preserving muscle quality is critical for healthy longevity. Aging and sarcopenia induce structural muscle alterations, yet the impact of muscle shape changes remains poorly understood, despite evidence from other clinical contexts linking shape to force production. We aimed to explore methodologies to characterize age- and sarcopenia-related muscle shape alterations using medical imaging. METHODS: Using 3D ultrasound acquisitions of 3 lower-limb muscles in 74 young and older adults, we developed and evaluated complementary muscle shape analysis approaches, ranging from simple geometric descriptors to advanced anomaly detection using Statistical Shape Models and Implicit Neural Representations (INR). RESULTS: Simple geometric descriptors revealed significant group differences in 3D Shape Factor for all muscles studied distinguishing young from older adults but not healthy from sarcopenic older adults. Population mean shapes showed proximal volume reductions in sarcopenic participants compared with healthy older adults, particularly in Rectus Femoris and Tibialis Anterior. Reconstruction errors from Statistical Shape Models were higher in sarcopenic participants (Rectus Femoris p = 0.03; Tibialis Anterior p < 0.001), with strong group discrimination on multivariate analysis (p < 0.001). INR-based reconstructions confirmed these patterns, with lower Dice scores in sarcopenic than in healthy older adults (0.93 vs 0.95; p < 0.001). CONCLUSION: Aging and sarcopenia induce distinct and measurable muscle shape alterations. Advanced shape modeling, particularly using INR, showed higher sensitivity. Rectus Femoris emerged as a promising candidate for a shape-based biomarker. Although the metrics presented here should not be interpreted as classification metrics, this study provides a reproducible, open-source framework for muscle shape analysis relevant to aging research.