Experimental evaluation of rigid patient motion detection and correction in SPECT with exponential data consistency conditions

Rigid patient motion may cause artifacts in the reconstructed SPECT images.This study evaluated a fully data-driven method exploiting exponential data consistency conditions (eDCCs) -equations describing redundancies in SPECT measurements-to estimate and correct for rigid patient motion on experimental data.We minimized the pairwise inconsistency of projections with a noise-aware metric implementing eDCCs.The method led to a significant improvement in image quality.For instance, for a 20 mm motion in the middle of the acquisition in the axial or vertical directions, the activity recovery coefficient (ARC) for the biggest sphere (37 mm diameter) of the NEMA IEC phantom was improved from 0.41 (non-corrected projections) to 0.58 (motion-corrected projections), i.e. the same ARC as the one obtained with reference motion-free projections.