Hosted by: Wei Xu

The X-ray computed tomography (CT) has been widely utilized as a nondestructive diagnostic means to visualize internal structures of human body. However, high radiation exposure in X-ray CT has been an important issue as it will increase the risk of cancer. Unfortunately, CT data acquired at low radiation doses adversely affects the quality if the reconstructions, impeding their readability. Therefore, current research is focusing on developing a SAFE (Safe, Accurate, Flexible, and Efficient) X-ray CT system in three aspects to handle reduced X-ray dose levels without compromising image quality. The first part of this research is in developing time efficient forward- and back-projection operators that are the most time-consuming parts in 3D CT reconstruction algorithms. The proposed method that is called Lookup Table-based Ray Integration (LTRI) method encapsulates 2D/3D CT geometric properties in two lookup tables to replace complex arithmetic operations with one or two memory fetching operations and it results in 4 times faster than the state-of-the-art method under modern graphics processing unit (GPU) architecture. With the novel CT projectors, iterative coordinate descent (ICD) based statistical iterative CT reconstruction algorithm is accelerated using GPU by seeking for parallelizable voxels in axial direction. we can achieve more than x90 times speed-up with cone-beam CT clinical dataset compared to the standard single voxel update scheme in ICD. The second part of this research is in utilizing the external knowledge that already exists in the domain of reconstructed high-quality CT scans to restore low-quality CT scans suffering from harsh noise and streak artifacts. We can incorporate this knowledge by creating a database (Big Data) of high-quality CT scans, either of the same patient or a diverse corpus of different patients, to assist in the restoration process since after all this is what radiologists do when they examine these low-quality CT images