Room: Exhibit Hall | Forum 2
Purpose: To determine whether mean CT scans obtained using three methods that vary widely in reconstruction time lead to equivalent treatment plans.
Methods: 4DCT scans were obtained using an iterative reconstruction algorithm (IRA) and a mean CT dataset (Mean-IRA) was created for clinical planning for three patients. A second mean (Mean-FDK) was created using a second 4DCT reconstructed using the same raw projection data with an FDK algorithm. Our scanner also reconstructs a native mean scan (Mean-Siemens) using all raw projection images without any binning. Mean-IRA was used for the clinically-treated plan. The other two mean scans were imported into the treatment planning system and fused based on DICOM origin to the Mean-IRA scan. The clinical structure set was copied onto the other two scans. An attending GI physician reviewed the fusion quality and way the contours align with the structures. The treatment plan was then copied onto the other two scans and the dose distribution was calculated and compared to the clinical plan.
Results: The reconstruction time for the Mean-Siemens , Mean-FDK and Mean-IRA scans was roughly 1, 5 and 20 minutes respectively. For all three patients, the physician approved with the fusion and contour sets without any change. DVHs from the corresponding organs overlaid almost perfectly. For four dosimetric datapoints (3D max, DMin-PTV, DMean-PTV and DMax-PTV) evaluated, the max difference was 0.9% (average = 0.2%, standard deviation = 0.4%).
Conclusion: While there are very minimal differences between plans calculated using the three mean CT scans, the reconstruction time for the scans varies quite significantly. It is therefore acceptable to use the Mean-Siemens scan for time-sensitive tasks such as setting and tattooing isocenter where the chance of patient motion increases the longer we wait.