Room: Exhibit Hall | Forum 6
Purpose: To validate planning CT to CBCT deformable image registration (DIR)-based dose calculation for prostate cancer online adaptive radiotherapy (ART).
Methods: To establish gold standard for DIR validation, replan CT (rpCT) was transformed to rescaled CBCT in pixel-level by means of histogram-matching method (HM) and slice-specific method (SS). HM/SS noise map was generated using deformed CBCT and HM/SS rescaled CBCT subtraction. Deassigning visible deformation errors with a phantom based gaussian noise to correct HM/SS noise map, hence HM/SS virtual CBCT can be acquired by the addition of HM/SS rescaled CBCT and HM/SS corrected noise map. Taking the dose calculated on rpCT as reference, the resulting dose based on virtual CT derived from pCT to virtual CBCT DIR are compared to reference. To verify the reliability of virtual CBCT result from pixel-level process, independent t test was done for the matching rate of dose distribution calculated on different set of virtual CT. Besides, pair t test was done for the dice similarity coefficient (DSC) of commercial-recommendation and updated DIR workflow-based organ at risks (OARs). To test the impact of different DIR workflow, dose distrubution calculated on different DIR workflow based virtual CT was compared.
Results: Contouring of HM/SS virtual CBCT are similar to rpCT. Compared to dose distribution calculated on rpCTï¼Œthe gamma index of dose distribution calculated on HM/SS virtual CT are 98.08Â±0.40/97.95Â±0.35 (P>0.05) for 1mm/1% and 99.53Â±0.08/99.55Â±0.10 (P>0.05) for 2mm/2%. The DSC of commercial-recommendation and updated DIR workflow propagated OARs are 0.78-0.95 and 0.85-0.96. No statistical difference was found between dose distribution calculated on different DIR workflow based virtual CT.
Conclusion: virtual CBCT is a promising approach for DIR validation, and offer reliable verification for DIR-based dose calculation for prostate cancer ART. Moreover, updated DIR workflow has the potential to achieve better geometric factor while dose distribution not substantially change.