Venkatesh B. Bhat BEng, Abstract Co-Author: Nothing to Disclose

Raj Shekhar PhD, Presenter: Nothing to Disclose

Image-guided interventions (IGIs) are minimally invasive procedures that offer improved outcomes with much faster recovery. On account of many factors (speed, spatial resolution, cost and ease of operation), computed tomography (CT) is used to guide many such procedures. However, radiation concerns limit CT’s potential use. We have developed two techniques for fast and high-quality reconstruction of CT data acquired at low radiation doses.  

We acquired low-dose CT scans and used iterative separable paraboloidal surrogates (SPS) algorithm for reconstruction. We overcame the slow speed of this iterative algorithm by implementing the algorithm on 1) a cluster of computers working in parallel and 2) a graphics processor unit (GPU). A phantom was scanned at radiation doses varying from 20mAs to 200mAs at 120 KVP. We used the reconstruction time of the SPS algorithm on a single-CPU computer as the benchmark to evaluate the speeds of the two faster techniques. The images reconstructed using the popular filtered backprojection (FBP) algorithm were used to compare the reconstruction quality.  

The first technique implemented on an 8-node cluster (four cores/node) was 30 times faster than the benchmark technique. The GPU-based technique was 100-150 times faster compared with the same benchmark. At low radiation doses, the quality of the reconstructed images was consistently better for our techniques as compared with the FBP technique.  

We have demonstrated acceleration of a popular iterative CT reconstruction technique on two non-custom hardware platforms. We have also demonstrated that the acceleration techniques result in high-quality reconstructed images for scans acquired at low doses.  

High-speed CT reconstruction from low dose scans using iterative techniques can make CT more amenable to IGIs.  

Bhat, V, Shekhar, R, High-speed Reconstruction of Low-dose CT for Image-guided Interventions.  Radiological Society of North America 2008 Scientific Assembly and Annual Meeting, February 18 - February 20, 2008 ,Chicago IL. http://archive.rsna.org/2008/6021898.html