Room: AAPM ePoster Library
Purpose: We have developed realistic 4D numerical lung CT phantoms to provide comprehensive ‘ground truth’ anatomical motion data for imaging and radiotherapy applications. As such, multiple numerical 4DCTs, based on CT data from 10 lung cancer patients, have been constructed by deforming these using motion extracted from 4DMRIs acquired on 5 volunteers.
Methods: First, lung masks extracted from each 4DMRI acquisition were registered to each CT, using deformable image registration (DIR) to establish inter-subject correspondence. Next, again using DIR, motion vector fields were extracted from each breathing cycle of the multiple cycle 4DMRI data. Finally, using the predetermined CT-MR lung mesh correspondence, these motion fields were used to deform each of the patient CTs to generate multiple numerical 4DCT-MRs per patient, each representing a wide range of motion patterns.
Results: A variety of anatomical scenarios were considered, based on 10 lung cancer CTs with pronounced differences in lung volumes/shapes and tumor locations/sizes. In addition, a total of 8 deformable motion patterns, each consisting of 50-100 individual motion phases, could be extracted from the 4DMRI data, generating a total of 80 simulated 4DCT data sets (also each 50-100 individual phases) when applied to each patient CT. In total, the full phantom data set therefore consists of 5200 different anatomical and motion phase scenarios.
Conclusion: The developed 4DCT-MR based numerical phantoms provide a realistic representation of a large selection of anatomical and motion scenarios. In addition, the motion vectors used to warp the CTs can be considered as the ‘ground-truth’ motion for each phantom, providing invaluable information for testing 4D imaging and motion modeling scenarios. Furthermore, we plan to extend the method to other organs in the thorax/abdomen region and incorporate more comprehensive shape and motion models into the platform, in order to expand the variability of both motion and patient anatomy.
Funding Support, Disclosures, and Conflict of Interest: This project is funded by Krebsliga Schweiz.