Population based approaches to computational musculoskeletal modelling

OpenHands: An Open-Source Statistical Shape Model of the Finger Bones

This paper presents statistical shape models of the four fingers of the hand, with an emphasis on anatomic analysis of the proximal and distal interphalangeal joints. A multi-body statistical shape modelling pipeline was implemented on an exemplar training dataset of computed tomography (CT) scans of 10 right hands (5F:5M, 27-37 years, free from disease or injury) imaged at 0.3 mm resolution, segmented, meshed and aligned. Model generated included pose neutralisation to remove joint angle variation during imaging. Repositioning was successful; no joint flexion variation was observed in the resulting model. The first principal component (PC) of morphological variation represented phalanx size in all fingers. Subsequent PCs showed variation in position along the palmar-dorsal axis, and bone breadth: length ratio. Finally, the models were interrogated to provide gross measures of bone lengths and joint spaces. These models have been published for open use to support wider community efforts in hand biomechanical analysis, providing bony anatomy descriptions whilst preserving the security of the underlying imaging data and privacy of the participants. The model describes a small, homogeneous population, and assumptions cannot be made about how it represents individuals outside the training dataset. However, it supplements anthropometric datasets with additional shape information, and may be useful for investigating factors such as joint morphology and design of hand-interfacing devices and products. The model has been shared as an open-source repository ( https://github.com/abel-research/OpenHands ), and we encourage the community to use and contribute to it.

Machine Learning for Optical Motion Capture-Driven Musculoskeletal Modelling from Inertial Motion Capture Data

Marker-based Optical Motion Capture (OMC) systems and associated musculoskeletal (MSK) modelling predictions offer non-invasively obtainable insights into muscle and joint loading at an in vivo level, aiding clinical decision-making. However, an OMC system is lab-based, expensive, and requires a line of sight. Inertial Motion Capture (IMC) techniques are widely-used alternatives, which are portable, user-friendly, and relatively low-cost, although with lesser accuracy. Irrespective of the choice of motion capture technique, one typically uses an MSK model to obtain the kinematic and kinetic outputs, which is a computationally expensive tool increasingly well approximated by machine learning (ML) methods. Here, an ML approach is presented that maps experimentally recorded IMC input data to the human upper-extremity MSK model outputs computed from ('gold standard') OMC input data. Essentially, this proof-of-concept study aims to predict higher-quality MSK outputs from the much easier-to-obtain IMC data. We use OMC and IMC data simultaneously collected for the same subjects to train different ML architectures that predict OMC-driven MSK outputs from IMC measurements. In particular, we employed various neural network (NN) architectures, such as Feed-Forward Neural Networks (FFNNs) and Recurrent Neural Networks (RNNs) (vanilla, Long Short-Term Memory, and Gated Recurrent Unit) and a comprehensive search for the best-fit model in the hyperparameters space in both subject-exposed (SE) as well as subject-naive (SN) settings. We observed a comparable performance for both FFNN and RNN models, which have a high degree of agreement (ravg,SE,FFNN=0.90±0.19, ravg,SE,RNN=0.89±0.17, ravg,SN,FFNN=0.84±0.23, and ravg,SN,RNN=0.78±0.23) with the desired OMC-driven MSK estimates for held-out test data. The findings demonstrate that mapping IMC inputs to OMC-driven MSK outputs using ML models could be instrumental in transitioning MSK modelling from 'lab to field'.

A biophysically guided constitutive law of the musculotendon-complex: modelling and numerical implementation in Abaqus

BACKGROUND AND OBJECTIVE

Many biomedical, clinical, and industrial applications may benefit from musculoskeletal simulations. Three-dimensional macroscopic muscle models (3D models) can more accurately represent muscle architecture than their 1D (line-segment) counterparts. Nevertheless, 3D models remain underutilised in academic, clinical, and commercial environments. Among the reasons for this is a lack of modelling and simulation standardisation, verification, and validation. Here, we strive towards a solution by providing an open-access, characterised, constitutive relation (CR) for 3D musculotendon models.

METHODS

The musculotendon complex is modelled following the state-of-the-art active stress approach and is treated as hyperelastic, transversely isotropic, and nearly incompressible. Furthermore, force-length and -velocity relationships are incorporated, and muscle activation is derived from motor-unit information. The CR was implemented within the commercial finite-element software package Abaqus as a user-subroutine. A masticatory system model with left and right masseters was used to demonstrate active and passive movement.

RESULTS

The CR was characterised by various experimental data sets and was able to capture a wide variety of passive and active behaviours. Furthermore, the masticatory simulations revealed that joint movement was sensitive to the muscle's in-fibre passive response.

CONCLUSIONS

This user-material provides a "plug and play" template for 3D neuro-musculoskeletal finite element modelling. We hope that this reduces modelling effort, fosters exchange, and contributes to the standardisation of such models.

Deep Learning Approaches to Automated Video Classification of Upper Limb Tension Test

The purpose of this study was to classify ULTT videos through transfer learning with pre-trained deep learning models and compare the performance of the models. We conducted transfer learning by combining a pre-trained convolution neural network (CNN) model into a Python-produced deep learning process. Videos were processed on YouTube and 103,116 frames converted from video clips were analyzed. In the modeling implementation, the process of importing the required modules, performing the necessary data preprocessing for training, defining the model, compiling, model creation, and model fit were applied in sequence. Comparative models were Xception, InceptionV3, DenseNet201, NASNetMobile, DenseNet121, VGG16, VGG19, and ResNet101, and fine tuning was performed. They were trained in a high-performance computing environment, and validation and loss were measured as comparative indicators of performance. Relatively low validation loss and high validation accuracy were obtained from Xception, InceptionV3, and DenseNet201 models, which is evaluated as an excellent model compared with other models. On the other hand, from VGG16, VGG19, and ResNet101, relatively high validation loss and low validation accuracy were obtained compared with other models. There was a narrow range of difference between the validation accuracy and the validation loss of the Xception, InceptionV3, and DensNet201 models. This study suggests that training applied with transfer learning can classify ULTT videos, and that there is a difference in performance between models.