MiniMAP: An automatic pipeline for paediatric lower limb musculoskeletal model generation

MiniMAP generates custom paediatric OpenSim lower-limb models from CT-derived articulated shape models, predicting skin padding with RMSE 4.68mm and bone surfaces with low millimetre errors.

• Why it matters: Paediatric musculoskeletal model building must balance limited data, geometric accuracy, and computation time, and needs reliable bone and skin-to-bone geometry for fitting to measurements.
• What they did: Using CT scans from over 300 children aged 4–18, the pipeline predicts bone geometry from demographics, linear bone measures, and 12 skin landmarks, while a random forest estimates skin padding. Leave-one-out testing quantified landmark, bone surface, and clinical measurement errors, and two additional paediatric cases (motion capture and MRI) evaluated unseen performance.
• The result: Skin padding ranged from 3.4mm (lateral malleolus) to 25.9mm (PSIS), with best prediction using age, height, mass, and specific widths (average RMSE 4.68mm). Bone surface RMSEs were 2.64 {+/-} 0.99mm (pelvis), 1.91 {+/-} 0.59mm (femora), and 1.69 {+/-} 0.56mm (tibiae/fibulae), with unseen-case errors of 1.64 {+/-} 1.2mm to 3.41 {+/-} 3.1mm.

AI summary of the abstract · may contain errors · not medical advice · read the paper.