Program Information
Calculation of Two-Material Tissue Parameterizations for Dual-Energy Imaging
R Jennings1*, (1) FDA Center for Devices & Radiological, Silver Spring, MD
Presentations
SU-C-209-4 (Sunday, July 31, 2016) 1:00 PM - 1:55 PM Room: 209
Purpose:: The characterization of the energy dependences of tissue attenuation coefficients is fundamental in dual-energy imaging. It has been posited that a two-parameter model is both necessary and sufficient to describe these dependences for tissues of medical interest in the diagnostic energy range. In one approach the two parameters are the weighting factors for the energy-dependent linear attenuation coefficients of two distinct materials. This is commonly known as the basis material decomposition method. The purpose of this work is to describe and evaluate a method for determining these weighting factors for user-selected basis materials and tissues of interest.
Methods:: The method essentially says that the linear attenuation coefficient of the tissue of interest is equal, to a good approximation, to the weighted sum of the coefficients of two other distinct materials. The weights can be obtained from the resulting equation, evaluated at two different energies. Not all pairs of energies produce weights that yield accurate results. The method for finding a quasi-optimal pair of weights is described.
Results:: The method has been applied to a variety of combinations of tissue of interest and materials chosen to represent it. In most instances, the accuracy of the match is of the order of a few tenths of one percent over a large energy range. Even for combinations that might be expected to give poor results, water represented by PMMA and aluminum for example, the maximum deviations are of the order of one percent.
Conclusion: The method produces characterizations that are at least as good as, and usually better than, the accuracy of x-ray measurements obtainable with reasonable patient doses and currently available hardware. The results also show that many different materials are reasonable choices for basis materials. The software used for this study is available on GitHub.
Contact Email: