Author: Carina Gassner, Guyue Hu, Thomas Kröncke, Christopher Kurz, Guillaume Landry, Katharina Niepel, Katia Parodi, Franka Risch, Florian Schwarz 👨🔬
Affiliation: Department of Radiation Oncology, LMU University Hospital, LMU Munich, Department of Diagnostic and Interventional Radiology, University Hospital Augsburg, Department of Medical Physics, Ludwig-Maximilians-Universität (LMU) München, Department of Medical Physics, Ludwig-Maximilians-Universität München (LMU Munich) 🌍
Purpose: To evaluate the performance of a clinical photon-counting CT (PCCT) scanner in terms of proton relative stopping power (RSP) estimation using PCCT data and to compare the impact of different object sizes, different positioning of material inserts, and different scanning conditions.
Methods: Custom cylindrical plastic phantoms, housing tissue-equivalent inserts, with varying sizes (∅ 13-25cm) and different insert positions (cylinder center & given radial distance to the center) were scanned at a NAEOTOM Alpha PCCT scanner with four protocols combining different tube voltages (120 & 140kVp) and different acquisition modes (normal & high resolution). For each phantom, relative electron density, effective atomic number, and RSP were estimated relying on calibration using a pair of low-energy and high-energy virtual mono-energetic images. Results using calibration phantoms of different sizes and using data acquired at different modes were compared. RSP accuracy was evaluated by comparing to water-column measurements of the insert materials and calculating the root-mean-square errors (RMSEs) over the phantom inserts; spatial resolution by the task-based MTF using the radial edge of the phantom.
Results: The estimated RSP agree mostly with the reference values within 1%, except for lung-equivalent inserts. Comparison of results using different calibration phantoms has shown minor variations in RSP accuracy: for a given phantom, the resulting RMSEs are lowest (0.72%-0.96% at 120kVp; 0.68%-0.83% at 140kVp) when self-calibrated, and slightly higher when calibrated with other phantoms. Spatial resolution of 0.60 and 0.64lp/mm was achieved for the normal-resolution and high-resolution modes when using the same kernel, respectively.
Conclusion: This work assessed the quantitative determination of proton RSP using PCCT data obtained at different tube voltages and different acquisition modes. The comparison between a variety of phantom sizes and insert positions has indicated consistent RSP accuracy of around 1%, when excluding lung-equivalent inserts, irrespective of the size of the calibration phantom.