Maximum-Likelihood Estimation of Detector Response for PET Image Reconstruction

An accurate system model is essential for reconstructing high-resolution images. While the system response of a positron emission tomography (PET) system can be measured directly, the process is often difficult to perform and time-consuming. Here we propose a maximum likelihood method for estimating the detector response from projections of a point source at a set of radial locations in the field of view. The detector response functions for all radial bins are estimated simultaneously. No interpolation is required. Compared to the method of direct measurement, the proposed approach requires less measurements and results in sparse system matrix because a factored system matrix is used, which reduces the reconstruction time. We conducted computer simulation to demonstrate the feasibility of the proposed method