Multi-center comparison of relative stopping power estimation of an ex vivo phantom using clinical X-ray CT scanners and a proton CT prototype.

Background

AND

Purpose

Accurate estimation of relative stopping power (RSP) is essential for proton therapy (PT) treatment planning. Emerging X-ray computed tomography (xCT) technologies are enabling more advanced RSP calibration strategies, but may also introduce greater variability across centers. This study aimed to assess inter-center variability in xCT calibration and RSP estimation for proton treatment planning, using a proton CT (pCT) prototype as reference. MATERIAL AND

Methods

An ex vivo phantom was transported to five European PT centers and one research center, and scanned with both single-energy and dual-energy xCT systems. Acquisitions were performed in three phantom configurations, simulating different anatomical sizes. The reconstructed xCT images were calibrated to RSP according to each center's clinical protocol and compared against the pCT-derived RSP maps.

Results

Voxel-by-voxel comparison between xCT-based and pCT-based RSP maps showed inter-center RSP variability of about 1% in soft tissues and up to 5% in bone, with larger differences observed for larger phantom sizes. The largest discrepancies (up to 5%) relative to pCT were observed in the same configurations. Centers implementing beam-hardening corrections generally showed closer agreement with pCT, whereas the effect of size-specific HLUTs was tissue- and anatomy-dependent. The impact of these RSP variations on water-equivalent thickness in representative clinical scenarios resulted in a 2σ inter-center variability of 1.09% and 1.39% for brain and prostate cases, respectively.

Conclusions

This work presents the first inter-center comparison of xCT-based RSP estimation using an ex vivo phantom and pCT measurements as reference. The observed variability highlights the relevance of a comprehensive and standardized validation of xCT calibration protocols.