Augmented Structure Preserving Neural Networks for cell biomechanics
This work addresses cell biomechanics modeling for researchers, but it is incremental as it builds on existing methods.
The authors tackled the problem of predicting cell trajectories and mitosis events in biomechanics by combining Structure Preserving Neural Networks with other ML tools, achieving high accuracy in simulations and real cases.
Cell biomechanics involve a great number of complex phenomena that are fundamental to the evolution of life itself and other associated processes, ranging from the very early stages of embryo-genesis to the maintenance of damaged structures or the growth of tumors. Given the importance of such phenomena, increasing research has been dedicated to their understanding, but the many interactions between them and their influence on the decisions of cells as a collective network or cluster remain unclear. We present a new approach that combines Structure Preserving Neural Networks, which study cell movements as a purely mechanical system, with other Machine Learning tools (Artificial Neural Networks), which allow taking into consideration environmental factors that can be directly deduced from an experiment with Computer Vision techniques. This new model, tested on simulated and real cell migration cases, predicts complete cell trajectories following a roll-out policy with a high level of accuracy. This work also includes a mitosis event prediction model based on Neural Networks architectures which makes use of the same observed features.