skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: SU-F-I-41: Calibration-Free Material Decomposition for Dual-Energy CT

Journal Article · · Medical Physics
DOI:https://doi.org/10.1118/1.4955869· OSTI ID:22626800
 [1];  [2];  [3]; ;  [4]
  1. Huazhong University of Science & Technology, Wuhan, Hubei (China)
  2. Stanford University School of Medicine, Stanford, CA (United States)
  3. Huazhong University of Science and Technology, Wuhan, Hubei (China)
  4. Dalio Institute of Cardiovascular Imaging NewYork-Presbyterian Hospital and, New York, NY (United States)

Purpose: To eliminate tedious phantom calibration or manually region of interest (ROI) selection as required in dual-energy CT material decomposition, we establish a new projection-domain material decomposition framework with incorporation of energy spectrum. Methods: Similar to the case of dual-energy CT, the integral of the basis material image in our model is expressed as a linear combination of basis functions, which are the polynomials of high- and low-energy raw projection data. To yield the unknown coefficients of the linear combination, the proposed algorithm minimizes the quadratic error between the high- and low-energy raw projection data and the projection calculated using material images. We evaluate the algorithm with an iodine concentration numerical phantom at different dose and iodine concentration levels. The x-ray energy spectra of the high and low energy are estimated using an indirect transmission method. The derived monochromatic images are compared with the high- and low-energy CT images to demonstrate beam hardening artifacts reduction. Quantitative results were measured and compared to the true values. Results: The differences between the true density value used for simulation and that were obtained from the monochromatic images, are 1.8%, 1.3%, 2.3%, and 2.9% for the dose levels from standard dose to 1/8 dose, and are 0.4%, 0.7%, 1.5%, and 1.8% for the four iodine concentration levels from 6 mg/mL to 24 mg/mL. For all of the cases, beam hardening artifacts, especially streaks shown between dense inserts, are almost completely removed in the monochromatic images. Conclusion: The proposed algorithm provides an effective way to yield material images and artifacts-free monochromatic images at different dose levels without the need for phantom calibration or ROI selection. Furthermore, the approach also yields accurate results when the concentration of the iodine concentrate insert is very low, suggesting the algorithm is robust with respect to the low-contrast scenario.

OSTI ID:
22626800
Journal Information:
Medical Physics, Vol. 43, Issue 6; Other Information: (c) 2016 American Association of Physicists in Medicine; Country of input: International Atomic Energy Agency (IAEA); ISSN 0094-2405
Country of Publication:
United States
Language:
English