Author: Afrouz Ataei, Andrew R. Godley, Mi-Ae Park, Mahbubur Rahman, Liqiang Ren, Chenyang Shen, Gary Xu 👨🔬
Affiliation: Department of Radiology, UT Southwestern Medical Center, UT Southwestern Medical Center, Rush University, Imaging Services, UT Southwestern Medical Center 🌍
Purpose: Even though many patients undergo diagnostic CT scans prior to treatment, simulation CT scans are commonly required for treatment planning. This study evaluates whether prior diagnostic CT scans can be directly utilized for treatment planning by comparing CT number (HU) to relative electron density (ED) and physical density calibration curves between diagnostic and simulation CT scanners using the same calibration phantom.
Methods: An ED phantom (Model 467, Gammex), containing 16 slots and 13 unique materials, was scanned using a simulation CT scanner and six diagnostic CT scanners from four vendors (Vendors A, B, C, and D). Vendor B had two scanners of the same model, while Vendor C had two different models. All scans were performed at 120 kV, with other scan and reconstruction parameters matched to clinical protocols. Calibration curves of ED vs. HU and physical density vs. HU were plotted for comparison. To quantify inter-scanner variations, relative HU changes with respect to ED and physical density were calculated for each material and compared across all scanners.
Results: Most calibration curves were visually aligned with the reference curve from the simulation CT scanner, except for the scanner from Vendor D. Relative HU variations compared to the simulation CT scanner were as follows: for materials with relative ED below 0.8 (lung-equivalent), scanners agreed within 7.3%; for ED between 0.8 and 1.2 (soft-tissue-equivalent), scanners agreed within 2.5%; and for ED above 1.2 (bone-equivalent), scanners agreed within 4.1%.
Conclusion: Relative HU variations across diagnostic CT scanners showed a maximum absolute difference of 7.3% (ED<0.8), 2.5% (0.8