Transmission tomography algorithm for 3D list-mode or projection data

Existing reconstruction strategies for transmission tomography (on emission tomography systems) expect the transmission and blank scan data to be in projection format. However, no reconstruction methods have been specifically devised for use with list-mode data sets, and it can be undesirable to bin high resolution list-mode data into projections for two reasons. First, the projections may be sparse - which is inefficient for data storage and processing, and second, there may be information losses unless very fine sampling is used for the projections. Since the two list-mode data sets (from a transmission scan and a blank scan) may each contain lines of response which are not present in the other data set, direct line-by-line comparisons are disallowed. The method suggested here for list-mode data (but also applicable to projection data), based on a maximum likelihood objective, involves backprojecting the transmission data, and then using the blank data set in an image-space based reconstruction of the linear attenuation coefficient distribution. The algorithm is assessed using multiple realizations of simulated list-mode data as well as measured 3D list-mode data from the ANIPET scanner. Results suggest the technique converges very rapidly indeed, whilst offering reconstructed image quality comparable to that obtained via the paraboloidal surrogates algorithm for maximum likelihood estimation.