Main Session
Sep
29
PQA 05 - Physics
3193 - Electron-to-Mass Density Conversion in Photon-Counting CT for Dosimetric Calculation: Phantom and Clinical Evaluation on Direct Density Reconstruction
Presenter(s)
Tsz Ching Fok, MSc - The Hong Kong Sanatorium & Hospital, Hong Kong, Hong Kong
F. K. Yeung1, C. W. Cheung1, T. L. Chiu1, H. M. Poon1, C. L. K. Chung1, T. C. Fok1, and S. K. Yu2; 1Hong Kong Sanatorium & Hospital, Happy Valley, Hong Kong, 2Medical Physics Department, Hong Kong Sanatorium & Hospital, Happy Valley, Hong Kong
Purpose/Objective(s):
Conventional CT simulation images reconstructed from all energy bins (T3D) suffer from beam-hardening artifacts, leading to inaccurate density values and dosimetric errors. Photon-Counting CT (PCCT) with Direct Density (Rho) reconstruction utilizes entire spectral information to generate robust relative electron density (rED) maps. However, some treatment planning systems (TPS) require mass density (MD) for dose calculation. This study aimed to develop and validate a PCCT-based workflow converting Rho-derived rED to MD and to quantify its dosimetric superiority over the conventional T3D approach, particularly near bony anatomy.Materials/Methods:
A tissue characterization phantom was scanned in both head and body configurations using a PCCT simulator with identical acquisition parameters (120kVp, 200mAs, Qr40 kernel, 2mm slice thickness). T3D and Rho images were reconstructed. HU-to-MD calibration curve was derived from HU-to-rED curve and tissue-equivalent density plugs. The stability of CT number (HU), rED, and converted MD was assessed across phantom sizes for both reconstruction methods. Subsequently, two clinical treatment plans of pelvic cases were calculated on both images. Dose distributions were compared using gamma analysis with strict 1%/1mm criteria (10% threshold).Results:
The Rho reconstructions demonstrated excellent stability across phantom sizes, with maximal differences of only 13 HU and =1% in rED and MD. In stark contrast, conventional T3D images exhibited pronounced beam-hardening artifacts in the larger body phantom, leading to a substantial underestimation of bone density with differences up to 186 HU, 4.6% in rED and 4.9% in MD on cortical bone. Results were tabulated. Crucially, these phantom-based inaccuracies translated to clinically significant dosimetric errors. While overall gamma passing rates exceeded 95%, the T3D-based plans resulted in a systematic target overdose, averaging 1.6% across the target volume and reaching up to 4.0% at critical bone-tissue interfaces, corresponding to density underestimation observed in phantom study.Conclusion:
The conventional T3D method leads to beam-hardening-induced density underestimation in bone, causing clinically significant target overdoses. Our validated workflow, combining PCCT Direct Density (Rho) reconstruction with a robust rED-to-MD conversion, effectively mitigates these artifacts. This ensures superior dosimetric accuracy for TPS requiring mass density and should be considered a preferred method for PCCT-based simulation, especially for treatment sites adjacent to or within bone.| Density plug | CT number Deviation (HU) | rED Deviation | MD Deviation | |||
| T3D | Rho | T3D | Rho | T3D | Rho | |
| lung exhale | -2 | -1 | -0.5% | -0.3% | -0.4% | -0.2% |
| adipose | 3 | -7 | 0.3% | -0.8% | 0.3% | -0.8% |
| breast | 1 | -8 | 0.1% | -0.8% | 0.1% | -0.8% |
| water | -2 | -4 | -0.2% | -0.3% | -0.2% | -0.4% |
| muscle | -3 | -4 | -0.1% | -0.4% | -0.1% | -0.6% |
| bone200 | -24 | 2 | -1.6% | 0.2% | -1.6% | 0.2% |
| bone800 | -77 | -3 | -2.5% | -0.2% | -2.8% | -0.3% |
| bone1750 | -186 | -13 | -4.6% | -0.7% | -4.9% | -0.7% |