Whole gamma imaging: comparison of full-ring and limited-angle Compton imaging

147 Introduction: Realization of Compton imaging as collimator-less SPECT has been desired for a long time. Last year, we developed the whole gamma imaging (WGI) prototype, which was a novel combination of PET and Compton imaging to measure various types of gamma rays with one system, and we succeeded to show an animal Compton image comparable with a PET image. The Compton imaging is a technique to image single-gamma rays by limiting the radiation source position on a surface of a cone. A pair of a scatterer detector and an absorber detector is used. All previous developments, which suffered imaging quality inferior to conventional PET and SPECT, were based on a limited-angle geometry. On the other hand, to the best of our knowledge, the WGI prototype is the world’s first realization of a full-ring Compton imaging system. The purpose of this study is to factorize the success of WGI in terms of the projection view angle used in image reconstruction. Methods: We measured a cylindrical phantom and a healthy white mouse by using the WGI prototype with 89Zr, which emits both 909-keV single-gamma rays and positrons with the time difference in the order of seconds. Compton events were extracted from coincidence list-mode events by applying energy windows of 50-350 keV for the scatterer, 550-850 keV for the absorber, and 800-1000 keV for the total energy. To simulate limited-angle geometry from the full-ring data, we extracted list-mode events obtained by the detectors comprising the limited-angle geometry. We simulated up to geometry equivalent to a single view Compton imaging system. The size of the detector for the limited angle geometry forming an arc was parameterized by the central angle of the arc. We explored from 30° to 360°, which corresponds one detector block of the scatterer ring and the full-ring geometry, respectively. For all cases, image reconstruction was done by the list-mode ordered subset expectation maximization method. Normalization and random correction were applied but attenuation and scatter corrections were not applied because the sizes of the phantom and the mouse were small, and these effects were negligible. Results: Reconstructed images for the limited-angle geometry showed strong streaking artifacts at the region far from the detector, especially when the detector coverage was small. Only regions surrounded by the limited angle geometry with wide angle showed similar image quality as the full-ring geometry. We found that maximum intensity projection (MIP) images projected in the direction from the detector toward imaging subjects did not change significantly when the partial ring had wide angle, even though 3D images showed strong streaking artifacts toward the direction away from the detector, which was parallel to the MIP. We can use the partial-ring Compton imaging only for applications where the 2D MIP image can provide sufficient information. In the case equivalent to a single view system, in which the geometry was two parallel detector planes, image quality was deteriorated even for the MIP image from the detector. Conclusions: It was shown that the detector coverage of the field-of-view is essential for complete 3D Compton imaging, although further consideration will be required to assess the tolerance for removing some detectors for specific applications.