Purpose/Objective(s):
To quantify CT number (Hounsfield Unit, HU) variability in Radixact ClearRT across clinically relevant acquisition parameters and to evaluate its dosimetric impact on adaptive radiotherapy (ART) dose recalculation in RayStation and Precision. This work aims to strengthen clinical confidence and workflow robustness in daily ART by identifying actionable sources of variability and readily implementable standardization strategies.
Materials/Methods:
CT number stability in daily verification imaging is essential for safe adaptive radiotherapy (1). ClearRT images were acquired using a TomoPhantom-HE (“cheese phantom”) with eight tissue-equivalent inserts (lung, bone, water). Anatomy protocols followed vendor-recommended configurations (2,3) while systematically varying Body Size (mAs), field of view (FOV), and acquisition Mode (couch speed, gantry period, pitch). Five regions of interest per insert were contoured to extract mean HU values and standard deviations. One-way and factorial ANOVA assessed statistical significance, and partial eta-squared (?p²) quantified the contribution of each parameter and interaction to HU variability. Dosimetric impact was assessed through dose recalculation in an anthropomorphic phantom with lung, bone, and soft-tissue equivalents using RayStation® and Precision®, with absolute dose validation using an A1SL ionization chamber.
Results:
Factorial ANOVA identified the Body Size–Mode interaction as the dominant contributor to HU variability (~70% of total variance), followed by Body Size (~50%), Mode (~40%), and FOV (~20%), all exceeding the 10% clinical relevance threshold (p<0.05). For the Thorax Large protocol (120 kVp, 160 mA, 440 mm FOV), mean HU values were -518.5±2.0 (lung), 1772.3±21.5 (bone), and -23.9±1.9 (water). One-way ANOVA showed significant HU differences among Anatomy protocols (Head, Thorax, Pelvis, Whole Body; p<0.001), except for water in the Head protocol (p=0.1). Dose recalculation demonstrated consistent behavior in RayStation® and Precision®, with agreement within 2% (lung), 1% (soft tissue), and 3% (bone), supporting stable clinical performance in ART workflows.
Conclusion:
CT number variability in ClearRT is primarily driven by acquisition parameter interactions, particularly Body Size and Mode, consistent with prior tomo QA observations (1). Although RayStation® and Precision® demonstrated comparable and clinically acceptable dosimetric performance, unstandardized acquisition settings introduce avoidable uncertainty in ART dose recalculation. Immediate standardization of CT density calibration curves across Anatomy, Body Size, FOV, and Mode, following Accuray guidance (2,3), represents a low-burden, high-impact action to enhance clinical confidence and strengthen daily adaptive radiotherapy workflow robustness.
References (1) Langen KM et al. Med Phys. 2010;37:4817–4853. (2) Accuray. PEG v3.0.x; 2021. (3) Accuray. ClearRT™ Helical kVCT Imaging for Radixact®; 2022.
