Abstract
The current approaches to imaging the tissue blood flow index (BFI) from diffuse correlation tomography (DCT) data are either an analytical solution or a finite element method, both of which are unable to simultaneously account for the tissue heterogeneity and fully utilize the DCT data. In this study, a new imaging concept for DCT, namely NL-DCT, was created by us in which the medical images are combined with light Monte Carlo simulation to provide geometrical and heterogeneous information in tissue. Moreover, the DCT data at multiple delay time are fully utilized via iterative linear regression. The unique merit of NL-DCT in utilizing the medical images as prior information, when combined with a split Bregman algorithm for total variation minimization (Bregman-TV), was validated on a realistic human head model. Computer simulation outcomes demonstrate the accuracy and robustness of NL-DCT in localizing and separating the flow anomalies as well as the capability to preserve edges of anomalies.