An Updated Near-Field Image Reconstruction Algorithm for Compton Telescopes

The MeV range is one of the least explored energy ranges in astrophysics, so the Compton Spectrometer and Imager (COSI) has been selected by NASA to study gamma rays in this part of the electromagnetic spectrum. The 3D imaging capabilities of COSI were largely unexplored but clearly visible in near-field source calibration data. It was theorized that the Compton imaging analysis should permit the identification of hot spots in the detector and spacecraft that activate due to hadron interactions (e.g., cosmic-ray protons and alphas, atmospheric neutrons). These hotspots could be modeled to improve the background estimation for the instrument. We developed a hybrid binned/list mode imaging technique to identify background events in the surrounding material of COSI. The imaging algorithm was created using List Mode Maximum-Likelihood Expectation-Maximization (a method based on the Richardson-Lucy deconvolution algorithm) with two steps: The first step (calculating the expectation) is a forward projection from the image into the data space, and the second step (maximizing the expectation and correcting the old image) is a back-projection from data space back into the image space. This imaging technique was applied to simulated data as well as real data from COSI balloon flights and resulted in an improvement of the localization of near field sources in three dimensions. The imaging algorithm developed will not only optimize the science return of COSI but could impact future MeV missions and other areas of study (e.g., nuclear medicine, emergency response/contamination remediation, and nuclear security/safeguards).