Predicting Intratumoral Fluid Pressure And Liposome Accumulation Using Physics Informed Deep Learning

This paper presents a deep learning model based on physics informed machine learning, which is capable of predicting the spatiotemporal accumulation of liposomes and spatial IFP in vivo from a single image of liposome accumulation. The model requires imaging of the spatial liposome accumulation within the tumour at a time post-administration as well as estimates of the parameters present in the mathematical model. This approach is an improvement over methods in previous studies, as it requires less data and can make predictions of spatial IFP and liposome accumulation simultaneously. The model was applied to an animal dataset from a previously published liposome accumulation study and the results showed that the model was able to accurately predict IFP.