Medical Physics Seminar – Monday, February 3, 2020
Advances in Diffusion Weighted Magnetic Resonance Imaging: Quality Assurance, Biophysical Modeling, and Data Analysis Approaches
Jose M Guerrero (student of Andy Alexander)
High Pitch Helical CT Reconstruction Using Combined MBIR and Deep Learning Prior Networks
John Hayes (student of Guang-Hong Chen)
Since the introduction of spiral/helical computed tomography (CT) to the clinic in the early 1990’s, there have been software and hardware developments to enable larger volume coverages per gantry rotation. Current state-of-the-art scanners are limited to helical pitches slightly larger than 1, meaning during each rotation the source traverses an axial distance equal to the detector collimation in the cone beam direction. We are proposing a reconstruction framework to enable higher pitch scans with the goal of achieving pitch of 2, 3, or even 4 detector collimations. The highly undersampled helical dataset is iteratively reconstructed and constrained in the image domain by a series of priors learned by deep learning networks.