Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

Purpose/Objective:

Comprehensive assessment of range uncertainties in particle treatment planning for a dual-energy CT (DECT) based direct stopping-power prediction (DirectSPR) suitable for clinical implementation.

Material/methods:

The DirectSPR approach, thoroughly validated in prior work and jointly implemented with Siemens Healthineers in a prototype, is characterised by a patient-size dependent model calibration and patient-individual consideration of tissue variability. Uncertainties of this DECT-based approach were quantified regarding image acquisition, modelling and miscellaneous sources (Fig.1) and propagated to the overall range uncertainty via the GUM guideline (Guide to the expression of uncertainty in measurement). Model calibration and validation was based on a multitude of CT scans for phantoms with varying geometries. The resulting overall uncertainty was determined for different clinically relevant tumour entities, separated into an absolute term (for five treatment sites) and a term relative to the particle range (for head, lung and pelvic region).

Results:

The relative range uncertainty (1.5𝜎) was 1.7%, 2.0% and 2.0% for the head, lung and pelvic region, respectively. The absolute term was between 2.5mm (brain) and 3.5mm (head&neck, pancreas). In comparison to the safety margin currently applied in treatment planning based on single-energy CT (3.5%+2mm), the overall range accuracy is increased for beam paths with a water-equivalent thickness above 30mm (70mm) in the head (body) region (Fig.2).

Conclusion:

The uncertainty in particle range calculation is reduced by patient-individual DECT-based stopping-power prediction. The obtained range uncertainties are directly applicable to the currently ongoing clinical implementation of DirectSPR for routine treatment planning at our institution and will result in a dose reduction in normal tissue.