An automated workflow for generating finite element models of the knee

We present an automatic workflow for generating patient specific finite element models of the knee from MRI. A neural network was trained on 20 segmented MRI, and tested on 5 MRIs. A custom python workflow was used to automatically create FE models. The workflow had a mean error of 0.83 mm, and on average produced a FE model in under 20 minutes. The workflow shows promise in reducing human error.