Room: Davidson Ballroom B
Purpose: This study aims to determine the optimal energies of mono-energetic images (MEIs) derived from Dual Energy (DE) CT to improve discrimination between tumors and adjacent organs at risk (OAR) and detection of CT textural changes during RT delivery.
Methods: MEIs from 40 to 190 keV in 5 keV increments were generated from DECT acquired for a phantom and 15 patients with pancreatic, liver, and breast cancers using both simultaneous (Drive, Siemens) and sequential (Definition, Siemens) protocols. CT histogram textures in tumors and OARs were calculated from MEIs. For tumor delineation, energy dependencies of texture features were analyzed to identify an optimal MEI energy associated with the maximum differences between the textures of tumors and OARs. For treatment response assessment, energy dependent relationships between the features and RT dose were compared to identify textures with the largest change during RT.
Results: Feature values of mean, kurtosis and skewness in the soft-tissue inserts of the phantom demonstrate the largest energy dependencies as confirmed in patients. The optimal MEI energy has the largest textural differentiation between tumor and OAR and was patient specific. For a pancreas patient, the difference in kurtosis between the tumor and the surrounding tissue was maximal at 45 keV, 4 times larger than that from the conventional CT. For a breast patient the difference in kurtosis was maximal at 110 keV 5 times larger than that from the conventional CT. All patients studied had the maximum difference in mean CTN at 40 keV. The radiation-induced change in kurtosis during RT was 4 times larger for 40 keV MEI than that with the conventional CT for a breast patient.
Conclusion: There are patient specific energy dependencies of CT textures from DECT MEI. Identifying the optimal energies to amplify differences between tumors and OARs improves delineation and CT-based response assessments.
Dual-energy Imaging, Dose Response, Texture Analysis
IM/TH- Image Analysis (Single modality or Multi-modality): Imaging biomarkers and radiomics