Ronald J. Jaszczak
Professor Emeritus in the Department of Radiology

We currently have a large effort in quantitative nuclear medicine imaging, specifically in SPECT. This research covers basic instrumentation development, investigation and development of new image reconstruction algorithms, simulations, and evaluation of new quantitative techniques. Quantitation has been an ongoing topic of research in our group, and requires understanding of the physical phenomena involved in SPECT acquisition. Photon attenuation, scatter, limited spatial resolution, and image noise all contribute to the inherent quantitative inaccuracy in SPECT images, and must be compensated for. Our group is investigating ways to solve these problems through improved hardware, simulation and reconstruction.

A major thrust of research in our group concerns improving imaging methods for evaluating myocardial perfusion and breast tumor characterization using Technetium-99, the most widely used nuclide. Our group has investigated alternate collimator design (fan-beam, half cone-beam, pinhole and astigmatic geometries), simulation methods, iterative reconstruction techniques, visualization, receiver operating characteristics (ROC) analysis, and correlation of images with other imaging modalities. Alternate collimator designs can improve the number of photons detected and therby reduce image noise, or in the case of pinhole collimation, can be used to magnify images significantly, increasing the effective resolution for small objects.

We have developed and tested new filtered backprojection and iterative reconstruction methods that model non-uniform attenuation and distance-dependent detectors response. Iterative reconstruction methods, although computationally time consuming, allow proper correction for the degrading effects. We have performed Monte Carlo simulation to determine collimator and reconstruction properties which cannot be measured easily. Simulation provides a way to divide the degrading physical effects into components for separate study. Both converging beam and parallel beam collimator geometries have been modeled and novel ways of improving the efficiency of the computationally intensive methods have been introduced.

Current Appointments & Affiliations

Contact Information

  • 2919 Reynolda Circle, Durham, NC 27712
  • 2919 Reynolda Circle, Durham, NC 27712

Some information on this profile has been compiled automatically from Duke databases and external sources. (Our About page explains how this works.) If you see a problem with the information, please write to Scholars@Duke and let us know. We will reply promptly.