Analyzing ligament prestrain in a multibody model of an ankle joint with random sampling

Abstract

Purpose Modeling an ankle joint is a challenge, especially when considering complex phenomena such as prestrain. In the literature two main approaches to ligament prestrain can be found in ankle modeling. The first one assumes a strain-free configuration, effectively omitting the prestrain, while in the second one the slack lengths are obtained by shortening the ligament lengths in the rest configuration by 2%. These approaches were not compared directly in a controlled environment. Methods The aim of the study is to compare the two common approaches to ligament prestrain in ankle joint modeling. The approaches are compared on a collection of models generated by random sampling from a 6-link, 2-contact pair multibody model of the ankle. Random sampling includes perturbation of slack lengths, which makes the generated models prestrained and with known output characteristics. Their resemblance to the original model and the ankle joint makes them viable for using in a prestrain comparison. Each generated model is prestrain with the two approaches, then the outputs are compared to determine, which approach returns results closer to reality. Results The comparison was performed on 592 generated models. On average, the strain-free approach significantly outperformed the 2% shortening. Conclusions The method for testing prestrain proposed in the paper is an effectively tool for exploring the solution space of the model. The obtained results were interesting, but should be taken with caution as they are connected to the test condition. However, the method is general and could used with any other biomechanical model.