(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: Patients who undergo fractionated radiotherapy (RT) are subjected to day to day anatomical changes, the changes might affect the gross tumour volume (GTV) delineated in the initial planning CT (pCT). Only in case of excessive weight loss, the patient is prescribed to undergo a repeat CT and re-planning procedure to incorporate the observed changes. This results in adding additional cost to the treatment and undocumented imaging dose to the patient. Thus, we have developed adaptive HU mapping and deformable image registration (DIR) technique which uses programmable neural network to create a new synthetic CT using pre-treatment verification CBCT
Methods: Prior to perform DIR, the initial pCT was registered with an extended CBCT (eCBCT) i.e, a protocol developed in our previous study for extending the field-of-view of conventional CBCT by acquiring the image through iterative couch shift. After rigid image registration, the Eclipse demons DIR algorithm was used to deform the pCT with respect to eCBCT. The images are then exported to MatLab system for performing adaptive HU mapping. Subsequently, the deformed CT (dCT) and eCBCT are optimally matched using MatLab mean-square rigid registration algorithm. later, the HU values in tissue and bone regions are segmented for dynamic HU correction and composite mapping of eCBCT and dCT using customised neural network, resulting in an optimal synthetic CT incorporating the anatomical changes and HU values of dCT.
Results: The synthetic CTs were compared with the repeat CT (rCT) of patients who completed RT. The standard comparison metrics viz, centre of mass shift (CMS) and volumetric dice-similarity coefficient (DSC) were analysed between the registered images. identical re-planning procedure was used for dose comparison and gamma analysis.
Conclusion: The observed deviation was well within the limits, hence the dynamic HU corrected synthetic CT can be used as an alternative for repeat CT.
Funding Support, Disclosures, and Conflict of Interest: This research work has been funded by Atomic Energy Regulatory Board (AERB), Govt of India (Grant No: AERB/CSRB/Proj.No.65/02/2016).
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email