Through the development of novel techniques for reconstructing artifact-free X-ray computed tomographic (CT) images, we aim to enhance the accuracy of target and critical anatomy delineation for radiotherapy treatment planning when foreign metal bodies are present in the patient, with specific emphasis on imaging in the presence of intracavitary applicators, seeds, and needles used in brachytherapy. We hypothesize that artifact-free image-reconstruction technologies will support sufficiently quantitative CT imaging, allowing us to solve the problem of non-invasive measurement of photon cross sections by CT imaging. This will allow Monte-Carlo based dose-calculation engines to account for the influence of tissue heterogeneities on dose specification for low-energy seed implants and other modalities utilizing low-energy photon fields. ? In Specific Aim 1,we propose a fundamental theoretical and experimental investigation of origin of streaking artifacts, aimed at understanding the relationship between CT detector readings and the radiological properties of the scan subject, including metal objects. We hypothesize that an important cause of artifacts is mismatch between overly simplistic models of CT detector response and the actual process of CT signal formation.
In Specific Aim 2, we propose to develop and test a novel algorithms, derived from rigorous physical models of the CT signal formation process (including noise, beam hardening, blur and scatter), for reconstructing artifact-free CT images from measured and simulated spiral transmission sinograms that accurately estimates applicator/metal object locations in the patient.
In Specific Aim 3, we propose to develop and clinically test a prototype system capable of off-line reconstruction of a three-dimensional (3D) imaging study from spiral transmission sinograms obtained from commercial CT scanners and including both current spiral and newer multi-row detector scanners. Finally, in Specific Aim 4, we will develop and test extensions of our model-based image-reconstruction algorithms for noninvasive measurement of photon attenuation coefficients and other radiological data from single- and dual-energy scans with a target accuracy of 3%. ? ?
Showing the most recent 10 out of 16 publications
