Inverse Rendering (IR) for high-resolution PET image reconstruction

The XEIL Lab has developed a new high-resolution image reconstruction framework based on a physically-based differentiable rendering of the emission image, which aims to produce the highest-resolution PET images due to accurate Monte Carlo (MC)-based forward modeling of the imaging system. We will use a differentiable objective function and apply a gradient-based optimization method in the inverse rendering path to update the estimated image.