Program Information
Comprehensive Evaluation of An Adaptive Deformation-Recovery and Intensity-Correction (ADRIC) CBCT Reconstruction Technique
Y Zhang1*, J Ma2 , P Iyengar1 , Y Zhong1 , S Niu1 , J Wang1 , (1) UT Southwestern Medical Ctr at Dallas, Dallas, TX, (2) Southern Medical University, Guangzhou, Guangdong
Presentations
TU-H-605-2 (Tuesday, August 1, 2017) 4:30 PM - 6:00 PM Room: 605
Purpose: Sequential same-patient CT/CBCT images share similar information, but may present inter-image deformation and non-deformation intensity changes. We developed an adaptive deformation-recovery and intensity-correction(ADRIC) technique to reconstruct new CBCTs from prior high-quality images, by correcting inter-image deformation and intensity changes.
Methods: ADRIC solves and separates inter-image deformation and non-deformation intensity changes with two alternating steps: deformation-recovery and intensity-correction. The deformation-recovery step solves the deformation from the prior image to the to-be-reconstructed CBCT image by 2D-3D deformation, which iteratively updates the deformation field to match the simulated projections from the deformed volume to acquired projections. Based on the deformation-recovered image, the intensity-correction step further corrects non-deformation intensity changes using the algebraic reconstruction technique. A modified total-variation regularization scheme is employed to smooth the reconstructed intensity corrections and remove small ones inappropriately applied towards residual deformation errors. The resulting intensity-corrected image is iteratively fed back to the deformation-recovery step as an adaptively-updated prior image, for further updates until final convergence.The efficacy of ADRIC was evaluated using simulated images from the extended-cardiac-torso(XCAT) digital phantom, and experimentally-acquired lung phantom data. Different levels of deformation and intensity changes were introduced to evaluate its robustness.
Results: For the XCAT study, through increasing the levels of deformation and intensity changes, the corresponding relative differences between prior and ‘gold-standard’ new images increase from 6.4% to 40.2%. After ADRIC reconstruction by 40 projections, relative differences between reconstructed and ‘gold-standard’ new images reduce to a range of 5.8%-9.7%. For the lung phantom study, the intensity errors of the reconstructed balloon targets range from 4.1% to 9.1%, as compared to the range of 8.9%-63.8% of the prior balloon target before reconstruction.
Conclusion: ADRIC accurately reconstructs new CBCTs from prior images under varying deformation and intensity change levels. It can substantially reduce the imaging dose by acquiring sparse-view projections for reconstruction.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by grants from the American Cancer Society (RSG-13-326-01-CCE), from the US National Institutes of Health (R01 EB020366), and from the Cancer Prevention and Research Institute of Texas (RP130109).
Contact Email: