Cardiac motion in gated spect

This work presents new methods to utilize cardiac motion information in single photon emission computed tomography (SPECT). Cardiac SPECT imaging is routinely used in clinical environment to assess ischaemic heart disease, a leading cause of death in the US. Cardiac SPECT allows for functional visualization of myocardial perfusion, and can assist detection of coronary artery disease. Further, cardiac gated SPECT, provides additional information on cardiac wall motion which is used to evaluate myocardium viability. However, gated reconstructions face increased noise levels and reconstructed images require further processing to restore perfusion and wall motion visibility.;In this work, we propose methods utilizing cardiac motion information to reduce the existing artifacts in cardiac gated SPECT. We have developed an accurate cardiac motion estimation algorithm relying on deformable mesh modeling and tracking of myocardial motion while accounting for myocardial brightening, an effect that can degrade performance of classical motion estimation techniques. The proposed motion estimation algorithm is utilized in three image processing approaches. The first approach performs post-reconstruction motion-compensated temporal filtering along the motion trajectory. The presented results show that this approach can improve image quality compared to existing clinical methods. The second approach is an iterative motion-compensated reconstruction modeling the physics of the imaging system while accounting for cardiac motion. It achieves significant image quality improvement. Thirdly, we analyze image quality assessment in a task-based framework and we have developed a new method for automatic image quality assessment using cardiac motion information. To this end, we present a numerical observer for prediction of human diagnostic performance in detection of cardiac motion defects. The proposed numerical observer relies on features extracted from the estimated deformable mesh model and on a prediction model based on machine learning techniques to predict human diagnosis. This new image quality assessment method has a potentially high impact on gated cardiac SPECT since it can be used for evaluation and optimization of imaging devices and processing algorithms aiming at better cardiac motion assessment. In summary, this work we developed cardiac motion estimation methods aiming at improving image quality in SPECT and provide means to automatically assess image quality.