Thesis (Ph. D.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a high sensitivity (71- 100%) for detecting breast cancers. DCE-MRI is approximately twice as sensitive as mammography for patients who are genetically predisposed to breast cancer, who have an elevated lifetime risk (up to 85%) of contracting the disease, and who require rigorous screening. However, current methods of DCE-MRI often have difficulty distinguishing malignant from benign tumours, resulting in low positive predictive values (on average, PPV = 45%). The research in this thesis is motivated by the need for improvement in the ability of MRI to differentiate breast lesions. It is believed that the differentiation of malignant from benign lesions can be improved by acquiring images of high spatial resolution for visualizing morphological features of tumours and those of high temporal resolution for characterizing contrast-kinetic curves. However, simultaneously achieving high spatial and temporal resolution is limited by the inherent trade-off between speed and quality in MRI, requiring one to choose a particular balance of spatial and temporal resolution. In this thesis, novel techniques are presented that reduce the need to choose a spatiotemporal resolution before the scan. Flexible methods are presented that allow images to be retrospectively reconstructed with different balances of spatial and temporal resolution from the same dataset. Flexibility is achieved through radial sampling of k-space data, with 3D radial directions based on Multidimensional Golden Means (MGM) and Halton sequences. Radial sampling also allows constrained image reconstruction techniques such as Compressed Sensing and Prior-Image Constrained Compressed Sensing to be incorporated for reducing undersampling artifacts in high-temporalresolution images. This thesis demonstrates in 2D how such reconstruction algorithms are influenced by acquisition schemes and shows how reconstruction algorithms work in synergy with flexible radial sampling to provide improvements in the quality of breast MR images. The flexibility to choose any spatiotemporal resolution combined with better image quality in fast images could potentially improve the characterization of breast lesions screened by MRI.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a high sensitivity (71- 100%) for detecting breast cancers. DCE-MRI is approximately twice as sensitive as mammography for patients who are genetically predisposed to breast cancer, who have an elevated lifetime risk (up to 85%) of contracting the disease, and who require rigorous screening. However, current methods of DCE-MRI often have difficulty distinguishing malignant from benign tumours, resulting in low positive predictive values (on average, PPV = 45%). The research in this thesis is motivated by the need for improvement in the ability of MRI to differentiate breast lesions. It is believed that the differentiation of malignant from benign lesions can be improved by acquiring images of high spatial resolution for visualizing morphological features of tumours and those of high temporal resolution for characterizing contrast-kinetic curves. However, simultaneously achieving high spatial and temporal resolution is limited by the inherent trade-off between speed and quality in MRI, requiring one to choose a particular balance of spatial and temporal resolution. In this thesis, novel techniques are presented that reduce the need to choose a spatiotemporal resolution before the scan. Flexible methods are presented that allow images to be retrospectively reconstructed with different balances of spatial and temporal resolution from the same dataset. Flexibility is achieved through radial sampling of k-space data, with 3D radial directions based on Multidimensional Golden Means (MGM) and Halton sequences. Radial sampling also allows constrained image reconstruction techniques such as Compressed Sensing and Prior-Image Constrained Compressed Sensing to be incorporated for reducing undersampling artifacts in high-temporalresolution images. This thesis demonstrates in 2D how such reconstruction algorithms are influenced by acquisition schemes and shows how reconstruction algorithms work in synergy with flexible radial sampling to provide improvements in the quality of breast MR images. The flexibility to choose any spatiotemporal resolution combined with better image quality in fast images could potentially improve the characterization of breast lesions screened by MRI.

The purpose of our study was to determine the incidence, causes, and reversibility of leukoencephalopathies demonstrating confluent areas of restricted diffusion on magnetic resonant imaging (DWI+LE). We hypothesized DWI+LE would have a low incidence, and be primarily caused by toxic exposures. We performed a logic sentence based search of the Yale-New Haven MRI database to select for reports indicating restricted diffusion within the cerebral white matter. We examined patients neuroimaging studies and medical record. We identified a total of 35 cases of DWI+LE, which resulted in an overall incidence of 0.2% over the five-year period queried. The medical conditions associated with DWI+LE were as follows: toxic exposure (7), hypoxia with concurrent trauma (7), hypoxia with concurrent toxic exposure (4), hypoxia with concurrent metabolic derangements (4), seizure with concurrent metabolic derangements (2), metabolic derangements (2), antiepileptic therapy (2), hypoxia (1), trauma (1), and unknown (5). The most favorable outcomes were seen in patients with intrathecal methotrexate toxicity, while patients with hypoxia without a lucid interval fared worst. We concluded that DWI+ LE are rare, their etiology diverse, and its reversibility dependant upon the type and severity of the insult.

Thesis (D. Eng.)--University of Queensland, 2002. / Includes bibliographical references.

Thesis (Ph. D.)--University of Washington, 2002. / Vita. Includes bibliographical references (leaves 101-111).

Thesis (M.S.)--Biomedical Engineering, Georgia Institute of Technology, 2003. / John Oshinski, Committee Member; Paul Benkeser, Committee Member; Xiaoping Hu, Committee Chair. Includes bibliographical references (leaves 45-47).

Thesis (Ph.D.)--Virginia Commonwealth University, 2009. / Prepared for: Dept. of Radiology. Title from title-page of electronic thesis. Bibliography: leaves 117-124.

Thesis (M.S.)--University of Florida, 2002. / Title from title page of source document. Includes vita. Includes bibliographical references.

Magnetic Resonance Imaging (MRI) has been applied in diagnosing Cervical Spondylosis Myelopathy (CSM) clinically. However, morphometric and signal change of MRI have not shown consistent relations with neurological function or outcome after surgical intervention. Diffusion Tensor Imaging (DTI) is an advanced MRI technology, which uses the principle of anisotropic water diffusion property. Recent studies indicated that DTI could be used as diagnostic tools for Cervical Spondylosis Myelopathy (CSM). The study aims to establish a Region of Interest (ROI)-based database. 65 healthy Chinese subjects were recruited for functional MR scanning. The effects on age and gender would also be investigated. Whole cord FA values decreased from upper cord level to lower cord level. White matter FA and AD values are significant higher than grey matter. White matter RD values are significant lower than grey matter. MD values of whole cord, white matter and grey matter are similar. There are no significant differences (P>0.05) of DTI metrics between males and females. There are significant differences (P<0.05) of DTI metrics in cervical spinal cord white matter in advancing age. / published_or_final_version / Orthopaedics and Traumatology / Master / Master of Medical Sciences