GPU-based List-mode Direct Parametric Reconstruction for Dynamic Cardiac SPECT

Noninvasive quantification of myocardial blood flow (MBF) using Single-Photon Emission Computed Tomography (SPECT) is an important clinical tool for the diagnosis and characterization of coronary artery disease. Current indirect MBF estimation approach using SPECT suffers from inaccurate quantification due to substantial image noise. Existing direct parametric reconstruction methods for conventional parallel-hole rotational scanners estimate kinetic parameters only for specific ROIs. In this paper we developed an innovative voxel-by-voxel based direct parametric image reconstruction algorithm to substantially reduce noise in MBF quantification using dynamic SPECT/CT and allow for patient radiation dose reduction. GPU-based parallel computing was used to achieve >2000-fold acceleration. The proposed method was evaluated in both simulation and in-vivo studies. Compared with the indirect method, the direct method achieved substantially lower image noise, particularly at high iterations and with low count levels.