high performance computing on graphics processing units: hgpu.org

In this paper we cast the problem of tomography in the realm of computer graphics. By using PBRT (physically based rendering toolkit) we create a scripting environment that simplifies the programming of tomography algorithms such as maximum-likelihood expectation maximization (ML-EM) or simultaneous algebraic reconstruction technique (SART, a deviant of ART). This allows the rapid development and testing of novel algorithms with a variety of parameter configurations. Additionally, it takes advantage of speed-up techniques that are common and well-researched in the graphics community, such as multi-resolution techniques based on octrees or similar space-partitioning data structures as well as algorithms accelerated through graphics hardware (GPU). Using our framework, we have evaluated different attenuation correction schemes during the back projection of ML-EM and SART