120 kVp Single-Energy CT calibration for radiation treatment planning by cross-calibration with Dual-Energy CT using...

In radiation treatment planning, CT-based tissue inhomogeneity correction relies on converting Hounsfield Units (HU) to relative electron density (RED). Single-Energy CT (SECT) calibration using tissue-equivalent materials (TEMs) suffers from HU-to-RED composition degeneracy and limited tissue realism. Dual-Energy CT (DECT) better resolves these ambiguities, but restricted clinical availability keeps SECT as the current standard.

To introduce a novel SECT calibration methodology that leverages real biological tissues within a phantom, combined with DECT datasets, to address the limitations of conventional SECT calibration approaches.

A conventional SECT scanner without vendor-provided DECT capabilities was used. DECT images were obtained by repeating scans at lower (80 kVp) and higher (140 kVp) energies, and calibrated with two different sets of certified synthetic TEMs. In the same session, (120 kVp) SECT and DECT images of a biologic phantom-a heterogeneous bovine specimen with adipose, muscle and bone tissues-were acquired. From TEM-calibrated DECT data, a RED map of the phantom was generated and subsequently used to cross-calibrate SECT, obtaining a final HU-to-RED calibration curve. Finally, a stoichiometric SECT calibration was performed for comparison.

The proposed methodology and the resulting HU-to-RED calibration curve, derived from the biological phantom dataset, demonstrated reduced dependence on the TEMs dataset used for DECT calibration, with variations within 0.8%, compared to conventional stoichiometric SECT calibration. Discrepancies with this stoichiometric curve remained within ± 2% for soft tissues, while larger differences, up to 9%, were observed in bone regions.

The proposed methodology is novel in employing biological phantoms and consecutive, multi-energy CT acquisitions, ensuring both biological representativeness and practicality. Since multi-energy CT imaging can be performed on any conventional scanner, the approach is easily transferable across centers. It may serve not only for developing new clinical calibration curves but also as a robust tool for validating existing ones.