Encrypted login | home

Program Information

Development and Evaluation of a Biomechanical Modeling-Assisted CBCT Reconstruction Technique (Bio-Recon)


Y Zhang

Y Zhang*, J Nasehi Tehrani , J Wang , UT Southwestern Medical Center, Dallas, TX

Presentations

MO-AB-BRA-9 (Monday, August 1, 2016) 7:30 AM - 9:30 AM Room: Ballroom A


Purpose:

To develop a Bio-recon technique by incorporating the biomechanical properties of anatomical structures into the deformation-based CBCT reconstruction process.

Methods:

Bio-recon reconstructs the CBCT by deforming a prior high-quality CT/CBCT using a deformation-vector-field (DVF). The DVF is solved through two alternating steps: 2D-3D deformation and finite-element-analysis based biomechanical modeling.

2D-3D deformation optimizes the DVF through an ‘intensity-driven’ approach, which updates the DVF to minimize intensity mismatches between the acquired projections and the simulated projections from the deformed CBCT. In contrast, biomechanical modeling optimizes the DVF through a ‘biomechanical-feature-driven’ approach, which updates the DVF based on the biophysical properties of anatomical structures. In general, Bio-recon extracts the 2D-3D deformation-optimized DVF at high-contrast structure boundaries, and uses it as the boundary condition to drive biomechanical modeling to optimize the overall DVF, especially at low-contrast regions. The optimized DVF is fed back into the 2D-3D deformation for further optimization, which forms an iterative loop.

The efficacy of Bio-recon was evaluated on 11 lung patient cases, each with a prior CT and a new CT. Cone-beam projections were generated from the new CTs to reconstruct CBCTs, which were compared with the original new CTs for evaluation. 872 anatomical landmarks were also manually identified by a clinician on both the prior and new CTs to track the lung motion, which was used to evaluate the DVF accuracy.

Results:

Using 10 projections for reconstruction, the average (± s.d.) relative errors of reconstructed CBCTs by the clinical FDK technique, the 2D-3D deformation-only technique and Bio-recon were 46.5±5.9%, 12.0±2.3% and 10.4±1.3%, respectively. The average residual errors of DVF-tracked landmark motion by the 2D-3D deformation-only technique and Bio-recon were 5.6±4.3mm and 3.1±2.4mm, respectively.

Conclusion:

Bio-recon improved accuracy for both the reconstructed CBCT and the DVF. The accurate DVF can benefit multiple clinical practices, such as image-guided adaptive radiotherapy.


Funding Support, Disclosures, and Conflict of Interest: We acknowledge funding support 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: