Motion compensated spatio-temporal filtering of cardiac gated SPECT images

Here we describe a post-reconstruction motion-compensated spatio-temporal processing algorithm for cardiac gated SPECT using estimated myocardium motion. In SPECT systems, radioactive dose limitations result in low photon count and therefore, reconstructed images are corrupted by noise. This is even more so in gated SPECT where the data is divided in a number of time frames. Existing noise reduction methods include spatio-temporal filtering, however, temporal filtering usually does not account for the heart motion thus introducing an artifact known as motion blur. This paper presents a motion-compensated temporal filtering method along with a motion estimation algorithm. Motion estimation is performed using a deformable mesh model, fitted to the left ventricle, which can track the myocardium displacement between the time frames. Temporal filtering is then applied utilizing the estimated myocardium displacement. Additionally, both motion estimation and temporal filtering account for the partial volume effect which is manifested by an image brightening effect when the heart wall thickens. Thickening is an important clinical indicator and as such has to be preserved by temporal filtering. The presented quantitative assessments demonstrate that the proposed method provide better images while preserving the brightening effect in comparison to methods used in clinical practice.