The development and validation of a 3D subject-specific anthropometric model through merging DXA and 3D scans for musculoskeletal modelling.
To understand pathology, injury and sport performance from a biomechanical perspective, we need accurate estimates of joint forces and moments. To compute these joint dynamics, the body is modelled as a collection of interconnected rigid segments whose inertial properties are obtained through either direct or indirect estimation methods.
Direct estimation methods are medical imaging based, and though accurate, they have disadvantages such as being expensive, not widely available, data processing is labor intensive and/or exposing participants to radiation. Indirect estimation methods use anthropometric values such as girth, length and breadth as predictors to compute the segments inertial parameters, and despite being easy to use, they may yield inaccurate results especially when applied to participants of a different population from which the equations were devised.
Moreover, the rigidity assumption of the segments, specifically for the trunk, may lead to reduced accuracy of the body dynamics.
Therefore, the present study proposes a 3D subject-specific anthropometric model devised from merging mass distribution data from DXA with the volumetric information from 3D surface scans. This model also uses skeletal animation technique to enable deformation and mass re-distribution of the trunk segment specific to the motion under analysis.
The model will then be validated through (i) verifying its accuracy when computing the inertial properties of objects and compare it against criterion measures (ii) assessing the accuracy of simulated kinematics of those objects during airborne motion (under the angular momentum conservation law) using the inertial properties gathered from the proposed model. Finally, the application of the method will be demonstrated in two studies, the first determining the errors arising from indirect estimation methods when applied to populations of elite athletes and the second assessing whether the anthropometric model reduces residual errors between the model dynamics and the experimental ground reaction forces.
The proposed method combines the information gathered using low cost medical imaging technology (DXA) and 3D surface scans to create relatively inexpensive and accessible 3D subject-specific anthropometric models. With more realistic BSIP, specifically at the trunk, this method is expected to provide more accurate estimates of internal and external forces applied to the body during low and high-velocity dynamic tasks.
It is also the first anthropometric model to assume the trunk as a non-rigid segment, but rather as a deformable segment.
If successful, the development of a 3D subject-specific anthropometric model represents another meaningful step towards more realistic musculoskeletal models for the assessment of human motion.