Cone-beam CT (CBCT) is being increasingly used in image-guided radiation therapy to ensure precise dose delivery as well as in the interventional radiology to provide 3D images during surgeries. However, the image quality and accuracy of reconstruction obtained on CBCT systems is limited by significant scatter, with CT errors up to 350 Hounsfield units. Most recently, dual-energy CT (DECT) permits improved differentiation of materials by decomposing x-ray attenuation onto two basic functions. DECT using two consecutive scans is impractical on open-gantry CBCT since the long imaging time (10s-60s) makes patient motion a significant problem. Current single-scan DECT methods require advanced x-ray tube and detector designs, which are not available on CBCT systems. We propose, for the first time, a single-scan DECT method for CBCT imaging that does not require any changes to x-ray tube, detector, or sub-system controls. The method uses our primary modulation approach, with which we have already shown excellent scatter correction by inserting a modulator sheet between the x-ray source and the object to modulate primary signals without shifting the frequency spectrum of scatter. Without increase of data measurement, the primary modulation method effectively estimates and removes scatter based on spectral analysis, and different research groups have demonstrated its success. In this program, we add a new capability of single-scan DECT into the primary modulation approach. X-rays inside and outside of the modulator shadows have different energy spectra. The projection data are separated accordingly and used for reconstruction at two different x-ray mean energies. To improve CT accuracy for each reconstruction that uses reduced projection data, the image gradient is calculated on the standard CT reconstruction with scatter corrected by primary modulation. An iterative algorithm using gradient weighting is then developed for DECT reconstruction from modulated projections that preserves small structures. To combat the ill conditioning of material decomposition, we propose a second iterative algorithm based on the principle of best linear unbiased estimation. Our preliminary phantom results show successful material decomposition with a difference of <5% between the proposed single-scan DECT method and the conventional two-scan DECT method. The overall goal of this project is to demonstrate the feasibility of single- scan DECT using primary modulation, to establish and optimize our method, and to investigate the method's performance on simultaneous scatter correction and material decomposition using phantom experiments on clinical CBCT systems. Our research will may eventually lead to new CBCT-based clinical procedures in both radiation therapy and interventional radiology and a disruptive technology widely applicable on different CT systems.
This project develops a dual energy imaging method for cone-beam computed tomography (CBCT) on a single scan, with simultaneous reduction of imaging errors from the fundamental physical processes, x-ray scatter. If proven effective, our method not only makes small and low-contrast structures more visible in a CBCT image by eliminating the shading artifacts, but also enables new clinical application by differentiating materials that appear identical on conventional CT images.
