Kinetic Model for Motion Compensation in Computed Tomography

Model based iterative reconstruction (MBIR) algorithms have recently been applied to computed tomography and demonstrated superior image quality. MBIR algorithms also have the potential to incorporate sophisticated models of data acquisition to address artifacts. In this paper, we introduce kinetic models to the MBIR framework, which allow voxel value to change as a function of time. Conventional reconstruction algorithms assume the voxels are constant over time. This assumptions is not true when patient motion is present, which, in turn, results in motion artifacts. Our approach to address this problem is to include such phenomena directly in the models of the cost function. We introduces a kinetic model for each voxel which is parameterized by a set of kinetic parameters. We then reconstruct the image by estimating these parameters in a MBIR framework. Results on phantom study and clinical data show that the proposed method can significantly reduce motion artifacts in the reconstruction.