A new framework for computational biomechanical models and 3Rs in musculoskeletal research.

该项目有两个总体目标：(1)调查肌肉骨骼计算机模型所需的实验输入数据的类型和数量，以提供准确的预测；(2)在此过程中，提供关于模型当前和未来潜力的定量数据，以促进科学研究中动物实验的减少、替代和改进（3Rs）。为了做到这一点，我们将开发和验证新的计算生物力学模型，使用兔子咀嚼作为我们的案例研究。验证计算机模型需要大量关于兔子解剖和喂养力学（例如肌肉和咬合力）的实验数据。这一数据并不存在于兔子身上，也不存在于任何其他实验动物身上。因此，需要对兔饲养进行系统的解剖学和生物力学研究，其中所有饲养力学的主要决定因素都是从一群兔子中测量的。根据这些兔子的医学成像数据构建的计算模型可以直接和立即用于改进和验证计算机模拟。只有这样，模型才能真正得到验证，并在未来的研究中展示其实现3r的潜力。因此，我们的具体目标是收集：兔骨和肌肉形态的解剖和图像数据；当它们吃不同种类的食物时，体内的骨骼运动和肌肉生理数据；并结合这些数据来建立和验证兔子饲养生物力学的新计算机模型。之所以选择兔子，是因为它们被广泛应用于各种研究领域。它们是牙种植体设计和骨（再）生长研究的首选实验动物，因为它们体积大，易于处理，在骨组成、愈合和解剖结构方面与人类相对相似。这些实验，像许多肌肉骨骼研究一样，是高度侵入性的，在动物被安乐死之前会给它们带来痛苦和痛苦。数字模型有可能完全取代（或最大限度地减少）在肌肉骨骼研究和/或医疗设备设计中使用动物。数字模型的解剖结构和行为可以无限制地改变和重新测试，不会对真实动物造成任何伤害或痛苦。这也可以允许，例如：通过对解剖学/行为的数字修改，将模型分析扩展到同一物种（或相似物种）的不同品系/品种；以多种方式修改解剖学元素（例如去除牙齿/骨头），以检查不同手术入路的后果；此外，还可以将数字化的植入设备植入模型中，并检查其对性能的影响，而无需在真实的动物身上进行任何有害的实验。但是改进生物力学模型不仅会减少动物在研究中的使用，而且有可能改善人类生物力学的建模。目前，模型被广泛用于研究健康的生物力学（如运动表现）、衰老（如骶肌萎缩症）和相关疾病（如膝关节骨关节炎）、牙科手术（如正畸治疗）和损伤（如髋部骨折）。在这些人体研究中，它们被用来估计或预测不能直接在人体中测量的参数，因此它们的准确性本质上难以评估。因此，我们在这里提出的这种研究显然是有必要的。首先，我们将使用我们详尽的和最先进的实验数据集生成迄今为止最全面的生物力学模型。这将为模型准确性提供最佳场景。然后，我们将逐步降低模型输入数据的分辨率，并观察其对精度的影响。这将告诉我们个体输入参数是如何影响准确性的，并帮助肌肉骨骼研究界确定哪些参数不需要通过真实动物实验来测量，以达到必要的准确性。

This project has two overarching goals: (1) to investigate the type and amount of experimental input data required for musculoskeletal computer models to deliver accurate predictions, and (2) in doing so provide quantitative data on the current and future potential of models to contribute to the reduction, replacement and refinement (3Rs) of animal experiments in scientific research. To do this we will develop and validate new computational biomechanical models using mastication in rabbits as our case study. Validating computer models requires a large amount of experimental data about rabbit anatomy and feeding mechanics (e.g. muscle and bite forces). This data does not exist for rabbits, or indeed any other experimental animal. Therefore a systematic anatomical and biomechanical investigation of rabbit feeding is required in which all the primary determinants of feeding mechanics are measured from a cohort of rabbits. Computational models constructed from medical imaging data of those same rabbits can then be directly and immediately used to improve and validate computer simulations. Only in this way can models be truly validated and their potential for achieving 3Rs in future studies be demonstrated. Our specific objectives are therefore to collect: anatomical and image data on bone and muscle morphology in rabbits; in vivo data on bone motion and muscle physiology as they eat various food types; and combine these data to build and validate new computer models of rabbit feeding biomechanics.Rabbits have been chosen because they are widely used in a variety of research areas. They are the first-choice experimental animal for dental implant design and bone (re)growth studies because of their size, easy handling and relative similarities to humans in terms of bone composition, healing and anatomy. These experiments, like many in musculoskeletal research, are highly invasive, causing pain and distress to the animals before they are euthanized. A digital model has the potential to completely replace (or maximally reduce) the use of animals in musculoskeletal research and/or medical device design. The anatomy and behaviour of a digital model can be altered and re-tested without limitation and without any harm or distress to a real animal. This can also allow, for example: a model analysis to be extended to a different strain/breed of the same species (or a similar species) by digital modification of the anatomy/behaviour; elements of anatomy to be modified in multiple ways (e.g. removal of teeth/bone) to examine the consequences of different surgical approaches; and for implant devices to be digitally inserted into the models, and their impact on performance examined, all without the need for any harmful experimentation on real animals. But improving biomechanical models will not only reduce animal use in research, but has the potential to improve modelling of human biomechanics. Currently models are used widely to study healthy biomechanics (e.g. sports performance), ageing (e.g. sacropenia) and related diseases (e.g. knee osteoarithitis), dental procedures (e.g. orthodontic treatment) and injury (e.g. hip fracture). In these human studies they are used to estimate or predict parameters that cannot be measured directly in people, thus their accuracy is inherently difficult to assess. Thus there is clear need for the type of study we propose here.In the first instance we will generate the most comprehensive biomechanical models produced to-date using our exhaustive and state-of-the-art experimental dataset. This will provide a best-case scenario for model accuracy. We will then incrementally reduce the resolution of input data given to the model and observe the effects on accuracy. This will tell us how individual input parameters effect accuracy and help the musculoskeletal research community identify which parameters do not need to be measured through experimentation in real animals to achieve the necessary accuracy.