Date of Award
Spring 1-1-2013
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Biomedical Engineering
First Advisor
Gilat-Schmidt, Taly
Second Advisor
Ropella, Kristina
Third Advisor
Clough, Anne
Abstract
The purpose of this study was to quantify the effectiveness of techniques intended to reduce dose to the breast during CT coronary angiography (CTCA) scans with respect to task-based image quality, and to evaluate the effectiveness of optimal energy weighting in improving contrast-to-noise ratio (CNR), and thus the potential for reducing breast dose, during energy-resolved dedicated breast CT.
A database quantifying organ dose for several radiosensitive organs irradiated during CTCA, including the breast, was generated using Monte Carlo simulations. This database facilitates estimation of organ-specific dose deposited during CTCA protocols using arbitrary x-ray spectra or tube-current modulation schemes without the need to run Monte Carlo simulations. The database was used to estimate breast dose for simulated CT images acquired for a reference protocol and five protocols intended to reduce breast dose. For each protocol, the performance of two tasks (detection of signals with unknown locations) was compared over a range of breast dose levels using a task-based, signal-detectability metric: the estimator of the area under the exponential free-response relative operating characteristic curve, AFE. For large-diameter/medium-contrast signals, when maintaining equivalent AFE, the 80 kV partial, 80 kV, 120 kV partial, and 120 kV tube-current modulated protocols reduced breast dose by 85%, 81%, 18%, and 6%, respectively, while the shielded protocol increased breast dose by 68%. Results for the small-diameter/high-contrast signal followed similar trends, but with smaller magnitude of the percent changes in dose. The 80 kV protocols demonstrated the greatest reduction to breast dose, however, the subsequent increase in noise may be clinically unacceptable. Tube output for these protocols can be adjusted to achieve more desirable noise levels with lesser dose reduction.
The improvement in CNR of optimally projection-based and image-based weighted images relative to photon-counting was investigated for six different energy bin combinations using a bench-top energy-resolving CT system with a cadmium zinc telluride (CZT) detector. The non-ideal spectral response reduced the CNR for the projection-based weighted images, while image-based weighting improved CNR for five out of the six investigated bin combinations, despite this non-ideal response, indicating potential for image-based weighting to reduce breast dose during dedicated breast CT.