ECG-Gating to Aid Attenuation Map Alignment in Cardiac SPECT using Data Consistency Conditions

In Single Photon Emission Computed Tomography (SPECT), accurate attenuation correction requires alignment of emission data with an attenuation map. Data consistency conditions (DCCs) depend on attenuation and can thus be exploited to provide an alignment method. DCCs have been used previously for attenuation correction in PET and SPECT. DCCs exist for parallel-hole geometry but not for pinhole SPECT collimation, however, interpolation of pinhole projections into parallel-hole projections allows application of parallel-hole DCCs. The assumptions of the DCC approach require the activity to be contained within a contiguous convex region of constant attenuation. This is violated in cardiac imaging due to extra-cardiac activity in structures like the liver and the heterogeneity of the attenuation in thorax. We hypothesize that electrocardiogram (ECG)-gating can be used to separate dynamic activity in the heart from static background signal and thus allow DCC-based attenuation map alignment. The approach was evaluated using computer simulated acquisitions of a pinhole cardiac SPECT camera. Two different ECG-gated activity distributions were used. 1) Activity only in the myocardium and 2) activity in both the heart and background structures. ECG-gated projections were fit pixelwise to a sine function to create a sine-amplitude projection. Using exponential DCCs, an attenuation map was translated through a range of spatial positions, and, at each location, the sine-amplitude exponential projection data were evaluated. The attenuation map was aligned where the eDCCs were most consistent, that is, where the relative difference between the eDCC-transformed projections was at a minimum. With myocardium-only activity, using a single gated projection gave a registration error of 0.28 mm, and the sine fit amplitude of ECG-gated projections produced an error of 0.14 mm. With extra-cardiac activity present, the registration error with a single projection was 13.14 mm but reduced to 1.01 mm for the sine amplitude. The sine fit method is a promising approach to correct for the violation of the eDCC-assumptions caused by extra-cardiac activity and thus allow an eDCC-based registration of the attenuation and emission datasets.