Inverse modeling of cancellous bone using artificial neural networks

Artificial neural networks are used to solve different tasks of daily life, engineering and medicine. In this work, we investigate its suitability for the examination of simulation results of cancellous bone with the aim to evaluate whether the bone is affected by osteoporosis. This bone disease is characterized by a reduction of the cortical bone phase, one of the two main components of the bone. The neural network predicts the simulated volume fraction in different parts of a cylinder, which models the bone. As a basis for its calculations, the neural network gets the information about the magnetic field inside the cylinder from finite element simulations. Examinations show that it is possible to train neural networks on solving that task with very high accuracies.

Automated summary

Artificial neural networks are investigated for evaluating the effect of osteoporosis on cancellous bone using simulation results. The neural network predicts the simulated volume fraction in different parts of a cylinder, based on the magnetic field information obtained from finite element simulations. The study shows that neural networks can solve this task with very high accuracies.