Building the digital dummy: towards developing high resolution, predictive 3D models of musculoskeletal systems

The musculoskeletal (MSK) system is a complex, multi-organ system which performs a number of important functions in animals, not least of which is the ability to move. In humans, MSK injury and disease accounts for a large proportion of all medical treatment, and the high cost associated with that treatment make it an important health burden for many countries. Computational simulations offer the potential to both better understand the cause of MSK injury and disease, as well as to direct and optimise its treatment. Unfortunately, current computational techniques lack the required accuracy, and predictive power that would be required for confident clinical applications. In the past, this inadequate level of accuracy has been largely due to the complexity of the MSK system in combination with the cost of computational power, and limited availability of high-resolution imaging data to perform the simulations on; as a result of this, simulations by necessity had to be heavily simplified. While the MSK system remains complicated, the significant improvements to the other two factors (particularly computing costs) presents an opportunity to make radical improvements to current simulation techniques, that until recently were unfeasible. By leveraging these improvements, a number of simulation techniques related to the MSK system can be considerably improved, ultimately forming the foundation for highly predictive, 3D models of the MSK system. Upon validation, those improvements will form the framework that allows for the development of an integrated, accurate, virtual anatomical modelling system that can be used for biomedical research and which will directly lead to improved clinical treatment of MSK injury and disease. This thesis provides and extensive review of what is currently possible in the field of computational simulations. Specifically, the key area of interest is the simulation of biological structures, and tissues. Whilst it is framed from the perspective of using computational techniques to benefit applications in medical sciences and clinical treatment, many of the cutting edge techniques for computational simulations have foundations in other disciplines. As such, this thesis draws from many disciplines that otherwise might be considered unrelated, such as engineering, comparative biomechanics, computer science, biology, ecology, and even palaeontology. By identifying the techniques that prove most promising (across all disciplines), as well as those areas that require additional development, a more targeted effort can be made towards developing highly accurate MSK simulations. Accuracy of such simulations is paramount in a clinical/treatment context, and thus they require the highest level of scrutiny before being used in practice. In order to be confident in the results of simulations, there needs to be a good understanding of how different (and logically valid) assumptions can affect the simulation results. The bulk of this thesis and primary contribution to the literature focuses on the sensitivity of simulations to the boundary conditions (assumptions) applied to them. This published work specifically assesses the sensitivity of a number of common Finite Element Analysis (FEA) assumptions in a multiple specimen comparative context. The key finding being that while there is no ‘silver bullet’ procedure for accurate simulations, the largest differences arose from functionally different simulations, indicating that despite the noise of uncertainty in parameter space, biological hypotheses in a comparative context can still be tested with a reasonable level of certainty using techniques such as FEA.