Advances in medical imaging and image reconstruction
Advances in medical imaging and image reconstruction have revolutionized health care over the past century, improving diagnostic accuracy of the medical images, and safety and comfort for patients. These advances included imaging acceleration, diagnostic imaging with non-ionizing radiation, dynamic imaging, multi-modal imaging, silent imaging, wide scanner development, (model-based) iterative image reconstruction and its algorithm acceleration. In particular, the incorporation of signal processing techniques and mathematical theory with imaging physics have made a big contribution in such achievements. ^ However, imaging techniques are not yet sufficiently incorporated with image reconstruction theories to maximize the accuracy of reconstructed images (in particular with fewer measurements), because underlying mathematical analyses are largely absent. Based on compressed sensing (CS), mean square error analysis, or several mathematical theories and signal processing techniques, this dissertation aims to shed light on maximizing the reconstruction performance from both the perspectives of medical imaging and image reconstruction. In particular, it concentrates on magnetic resonance imaging (MRI) and X-ray computed tomography (CT), since these two are the current mainstream state-of-the-art modalities. ^ In chapter 1, a new efficient CS sensitivity encoding (SENSE) parallel MRI reconstruction framework promoting joint sparsity is presented for reliable imaging time reduction and imaging flexibility. In chapter 2, a new framework called coil precoded multiple-input multiple-output SENSE MRI is presented to maximize the performance of high-field MRI. In chapter 3, the radiation dose problem of X-ray CT is dealt with new frameworks based on non-convex CS, tensor discrete Fourier slice theorem, and/or uniform sampling of projection angles at random.^
Thomas M. Talavage, Purdue University, Benjamin J. Adcock, Purdue University.
Applied mathematics|Electrical engineering|Medical imaging