Purpose: DECT-based algorithms were validated for extracting composition, density and relative stopping-power (RSP) using both fresh animal tissues and novel tissue-substitutes with known, realistic composition.
Methods: Two animal tissues and four gelatinous tissue-substitutes have been consecutively scanned at a CT scanner with 80 kV and 140 kV protocols. A published algorithm based on a comprehensive set of human tissues was then used to perform voxel-wise tissue assignation of the DECT scans to obtain an expected elemental composition. To obtain a human tissue base that better reflects the structure of bony tissues, this list has been extended beyond the published selection by additional 25 intermediate compositions and densities of spongiosa bone tissues representing various bone sites of the ICRP computational phantom. Maps of expected RSP, mass- and electron-densities were also calculated. Ground truth RSPs were measured using a multi-layer ionization chamber. The elemental compositions of the fresh samples and constituents of tissue-substitutes have been independently measured using chemical combustion analysis and prompt-gamma spectroscopy.
Results: The mean absolute mass weight differences of DECT to prompt-gamma based compositions of soft tissues were 3.8% and 8.7% for carbon and oxygen, respectively. The chemical combustion analysis yielded larger discrepancies to DECT with mean absolute differences of 8.3% and 9.8%. The concentrations of calcium and phosphorus of the two bone samples could be predicted within on average 0.9% and 1.0% of the chemical analysis. The largest discrepancies occurred for the more inhomogeneous samples like brain and tissue-substitutes which differed from tabulated human tissues, such as adipose. The RMSE of the RSP was 0.5% for soft tissues and 1.9% for bony tissues.
Conclusion: DECT based assessment of relevant elemental composition and RSP were in sufficient agreement with the ground truth of the samples to be suitable for improving dose verification workflows in the future.
Funding Support, Disclosures, and Conflict of Interest: Deutsche Forschungsgemeinschaft - Graduiertenkolleg 2274 (DFG GRK2274) National Institutes of Health, C06-CA059267 National Cancer Institute grant R01-CA229178
Dual-energy Imaging, Quantitative Imaging, Image Analysis