Purpose: To describe a computational approach to the design of spectrum-independent CT calibration materials and describe preliminary experimental verification.
Methods: : The basis material paradigm asserts that the linear attenuation coefficient of an arbitrary material can be accurately represented by a combination of the linear attenuation coefficients of two dissimilar materials. The combined thicknesses, relative to the material being studied, of the basis materials do not, in general, equal unity. Adding a third material resolves that issue. Although the desired result is that the combined relative thicknesses of the three materials be equal to 1.0, volume changes when components are mixed may occur. Replacing the thickness condition of 1.0 with a constant, call it k, close to 1.0 allows for an iterative procedure that results in the desired match. If the constant is allowed to have values significantly greater or less than 1.0, and the material being characterized is water, the result is a mixture that has a constant ratio in attenuation coefficient relative to water, i.e., its CT number is independent of energy and is determined by the value of k.
Results: Mixtures with nominal HU values greater than 0 were calculated using glycerol, water, and salt. Mixtures for negative CT numbers were calculated using isopropanol, water, and potassium nitrate. CT numbers up to 248 and down to -183 could be reached with the components studied. CT scans on a SIEMENS SOMATOM Force dual source CT scanner with samples calculated to produce HU values of 100 after correction for volume change were within 1 HU of the target value at kVs from 80 to 140.
Conclusion: The method described produces material formulations that give CT numbers that are independent of x-ray spectrum. These formulations may be useful for the calibration of CT systems, especially in facilities with multiple scanners.