Author: Brian Burger, Carri K. Glide-Hurst, John W. Hayes, Carson Hoffman, Minglei Kang, Jessica R. Miller, Jordan M. Slagowski, Yuhao Yan 👨🔬
Affiliation: Department of Human Oncology, University of Wisconsin-Madison, Departments of Human Oncology and Medical Physics, University of Wisconsin-Madison, Leo Cancer Care 🌍
Purpose: The combination of upright CT and patient positioner for particle therapy present tremendous potential for high precision IGRT and online adaptive radiation therapy (ART), where reliable CT performance is essential. We performed the first comprehensive characterization of upright CT reliability via a longitudinal phantom study.
Methods: A prototype upright CT with 85 cm diameter bore and 6 degree-of-freedom patient positioner was used to scan a Multi-Energy CT phantom (MECT) with tissue-mimicking inserts (lung, soft tissue, adipose, bone) at 120 kVp/250 mAs. Scanning occurred over three separate sessions following consensus guidelines of scanning bone inserts individually for two phantom sizes (body/head) to derive site-specific HU look-up tables (HLUTs). To test CT number stability, a phantom configuration with four bone inserts distributed across the body phantom was scanned over 3 sessions (n = 5 repeat scans/session). Stopping-power ratio (SPR) was computed following the Bethe equation with mean excitation energies from ICRU-49. HLUT was generated by piecewise linear regression between SPR and HU. To evaluate dose calculation sensitivity, HLUTs derived from both configurations were implemented to lung and spine proton plans on a thorax phantom.
Results: Excellent intra- and inter-session repeatability was observed (standard deviation <1.2 HU and <1.3 HU, respectively). Size-dependent beam hardening led to differences between body and head HLUTs with ΔSPR up to 6.5% within [-200, 200] HU and 4.4% within [200, 1500] HU, suggesting the need for additional beam hardening correction and size-specific HLUT. Comparing HLUTs derived from two different configurations of the body phantom, minimal discrepancies were observed (ΔSPR<3% within [-200, 1500] HU). Dose differences were negligible between the two HLUTs with target ΔD95<0.4% and OAR ΔD2<0.7% (ipsilateral lung and cord).
Conclusion: Our upright CT is reliable for proton dose calculation. With confirmation of long-term stability, online ART in the upright position can be realized.