Redundancy information induced edge-preserving prior for perfusion CT image reconstruction

With the widespread use of perfusion X-ray computed tomography (PCT) imaging technique in clinic, the associated radiation dose has arisen a significant concern to patients because of the multiple sequential scans. Reducing the x-ray exposure as low as possible has been of the major endeavors in PCT fields. During the PCT imaging, the CT intensity (HU) of region of interest (ROI) is changing after the contrast injected into patient. However, the most anatomical structure information of the images from different scan times does not change significantly. In other words, huge similar redundant information can be found among the time-series PCT images. With this observation, in this paper, by exploiting the redundancy information between the pre-contrast image and the enhanced image, we propose a new the edge-preserving prior (prNLMP) for high quality image reconstruction in case of low-mAs scan protocol. Evaluations with the simulated low-dose clinical brain PCT datasets clearly demonstrate that the presented method achieves higher accuracy of image reconstruction with lower image noise level, and obtains the noticeable kinetic enhanced details of the image and the associated hemodynamic parameter maps.