Positron Emission Tomography Biomedical Image Reconstruction on High-Performance Computing Systems

Biomedical imaging, such as X-ray computed tomography (CT) and magnetic resonance imaging (MRI), has proven to be invaluable in providing physicians and researchers with the ability to observe anatomical structures in living organisms. Another form of imaging technology, positron emission tomography (PET), has extended biomedical imaging to the observation of functional processes in addition to anatomical structures. However, positron emission tomography has been limited by its low photon counting statistics-due to the nature of its signal detection technique-which results in images with lower resolution and higher noise compared to computed tomogaphy or magnetic resonance imaging.

To improve positron emission tomography image quality and effectively utilize this technology, sophisticated reconstruction algorithms that use substantial computational power are required. The University of Minnesota has great strengths in both high-performance computing and biomedical imaging. Professor David Lilja and his research group in the Department of Electrical Engineering are capitalizing on these strengths by extending an ongoing interdisciplinary research project between the Departments of Electrical Engineering, Computer Science, and Radiology at the University of Minnesota, and the Positron Emission Tomography Imaging Laboratory at the Veterans Administration Medical Center in Minneapolis. They are focusing on the problem of reconstructing 3-D images of human physiology with data obtained from the PET scanner in the jointly funded Veterans Administration Medical Center/University of Minnesota (VAMC/UM) PET program. In particular, they are implementing a PET image reconstruction algorithm on the Cray T3D multiprocessor system using the standard PVM message-passing library. The outcome of this project is the capability to produce higher quality PET images within a reasonable time frame for clinical research applications. This capability benefits a range of research programs in Radiology, Neurology, Psychiatry, and Cardiology.

figure courtesy of Dr. Jeih-San Liow, Department of Veterans Affairs Medical Center These positron emission tomography-fluorodeoxyglucose images are from the brain of a normal volunteer. The intensity implies the uptake of fluorodeoxyglucose (FDG) in the brain. FDG is similar to the regular glucose consumed by the tissue except that one of its OH- is replaced by the positron-emitting isotope F-18 to make it radioactive. The purpose is to get a functional picture of the brain. This method is in contrast to other imaging techniques like CT and MRI which provide only anatomical information.

In This Issue:

International Conference on Parallel Computing

Positron Emission Tomography

Workshop on Modeling in Biochemical Engineering

Evolutionary Algorithms Workshop

Upcoming Workshop on Rational Drug Design

Seminar Synopses

Research Reports