| Low dose computed tomography of the breast (breast CT) is a relatively new technology that has potential to improve breast cancer detection and also has potential to serve as a platform for diagnostic and therapeutic work on the breast. We have built two breast CT scanners which are being used for a number of different clinical trials and have scanned more than 200 patients to produce more than 400 high resolution 3D images. The work reported in this dissertation was to advance this technology through: the development and improvement of these breast CT scanners (in both hardware and software) so as to increase their usability and flexibility; the development of tools and algorithms to de-noise, reconstruct, segment, view, and perform statistical analysis on breast CT images; and analysis of breast CT data to determine factors such as average breast density (which is important for both radiation dosimetry and cancer detection), and mass lesion detection sensitivity (for comparing breast CT with mammography).;Improvements made to the system include: additional detector modes, the addition of a filter changer, a step and shoot scanning mode, a motor to control the vertical position of the x-ray tube, and the addition of a dedicated breast PET scanner. The breast CT system has been fully streamlined to provide automated calibrations, scans, reconstructions, and image viewing. These improvements have allowed us to focus on clinical trials in breast CT, contrast enhanced breast CT, breast PET CT, breast CT plus breast tomosynthesis, rheumatoid arthritis PET CT, and a dual energy study. Breast density analysis and ideal observer performance analysis of the breast CT image data have shown that breast CT provides relevant information about the breast to researchers as well to clinicians, and has potential to provide earlier detection of cancer in the breast. |
