Clinical application of dual-energy computed tomography improves stopping-power prediction


Clinical application of dual-energy computed tomography improves stopping-power prediction

Wohlfahrt, P.; Möhler, C.; Greilich, S.; Richter, C.

Purpose/Objective:

Assessment of accuracy and robustness of treatment planning on dual-energy CT (DECT) in a multi-step validation and clinical implementation scheme (Fig.1,2).

Material/Methods:

To ensure reliable translation of DECT into routine clinical application, scan settings and stopping-power prediction methods were optimized and validated using 13 different animal tissues and an anthropomorphic ground-truth phantom. The clinical relevance of DECT-based stopping-power prediction was evaluated on dual-spiral DECT scans of 102 brain-, 25 prostate- and 3 lung-tumor patients treated with protons. DECT-derived voxelwise correlations of CT number and stopping-power ratio (SPR) were used for adapting the clinically applied CT-number-to-SPR conversion (HLUT) and quantifying intra- and inter-patient variability.

Results:

The accuracy of DECT-based stopping-power prediction was within 0.3% stopping-power and 1mm range uncertainty of validation measurements. Clinically relevant mean range shifts (±1SD) of 1.2(±0.7)% for brain-, 1.7(±0.5)% for prostate- and 2.2(±1.2)% for lung-tumor patients were obtained between dose calculations using HLUT or DECT-derived SPR. These deviations were significantly reduced (p<<0.001, two-sample t-test) below 0.3% by HLUT refinement based on patient-specific DECT information. Still, the remaining large intra-patient soft tissue diversity of approx. 6% (95% CI) and age-dependent inter-patient bone variability of 5% cannot be considered by any HLUT-based range prediction.

Conclusion:

Additional tissue information provided by DECT allows for accurate stopping-power prediction and incorporation of patients’ tissue diversity in treatment planning. These advantages can contribute to reduce CT-related range uncertainty and have been gradually translated in our clinical routine: (1) DECT-derived pseudo-monoenergetic CT dataset with generic HLUT, (2) DECT-based HLUT adaptation and soon (3) patient-specific DECT-basted stopping-power prediction.

Keywords: proton therapy; range uncertainty; dual-energy CT

  • Lecture (Conference)
    57th Annual Particle Therapy Co-Operative Group (PTCOG) Meeting, 24.-26.05.2018, Cincinnati, Ohio, USA

Permalink: https://www.hzdr.de/publications/Publ-26625
Publ.-Id: 26625