Accelerating Monte Carlo forward model with structured light illumination via “photon sharing” (Conference Presentation)

With the rapid development of spatial light modulators, structured light strategies have been readily implemented for efficient diffuse optical tomography (DOT) and fluorescent molecular tomography (FMT) applications. Compared to traditional pencil-beam sources, wide-field illumination enables larger imaging field-of-view (FOV), higher signal-to-noise ratios (SNR), and faster data acquisition speed, making it attractive for small-animal whole-body imaging. As the gold-standard for simulating photon propagations inside complex biological tissues, the Monte Carlo (MC) method is one of the most accurate approaches for imaging general media. Recently, thanks to parallel hardware such as graphics processing units (GPUs), MC simulations can be computed with high efficiency even with personal computers. While we have added support for wide-field illumination patterns in our widely distributed MC platform (http://mcx.space), the computation for multiple patterns is currently performed sequentially. To further accelerate forward modeling of large number of wide-field patterns, we propose a new method, referred to as “photon sharing”, to simultaneously simulate multiple structure-light sources. We demonstrate a 5- to 10-fold reduction of the MC simulation time. This technique is particularly valuable in DOT or FMT applications using structured light illumination and/or single-pixel-camera based systems. The proposed algorithm has been implemented in our open-source MC simulation platforms, supporting both CPUs and GPUs.