Efficient inversion strategies for estimating optical properties with Monte Carlo radiative transport models

Abstract

Indirect imaging problems in biomedical optics generally require repeated evaluation of forward models of radiative transport, for which Monte Carlo is accurate yet computationally costly. We develop a novel approach to reduce this bottleneck which has significant implications for quantitative tomographic imaging in a variety of medical and industrial applications. Using Monte Carlo we compute a fully stochastic gradient of an objective function for a given imaging problem. Leveraging techniques from the machine learning community we then adaptively control the accuracy of this gradient throughout the iterative inversion scheme, in order to substantially reduce computational resources at each step. For example problems of Quantitative Photoacoustic Tomography and Ultrasound Modulated Optical Tomography, we demonstrate that solutions are attainable using a total computational expense that is comparable to (or less than) that which is required for a single high accuracy forward run of the same Monte Carlo model. This approach demonstrates significant computational savings when approaching the full non-linear inverse problem of optical property estimation using stochastic methods.