Localized convolutional function regression: A computational method for measuring interstitial fluid flow and perfusion in DCE-MRI

Motivation: Aggressive gliomas are known to migrate in the direction of interstitial fluid flow (IFF), though it is difficult to measure interstitial fluid flow using MRI. Goal(s): Our goal is to develop a computational method for measuring IFF using DCE-MRI. Approach: We developed localized convolutional function regression (LCFR), validated in silico, in porous hydrogel, and apply it to in vivo tumors. Results: LCFR accurately measures fluid flow and perfusion to less than 10% error in silico, and measures IFF in a mouse model of glioma to be 1.63E-3 mm/s. In a case study, the method tentatively predicts invasion across the corpus collosum. Impact: This method will allow physicians and researchers to investigate how highly aggressive gliomas invade healthy tissue, and can be further used to predict how therapeutic agents or cells will disperse throughout the tumor, predicting disease progression and response.