Program Information
Improving Brain Soft-Tissue Contrast with Iterative Reconstruction of Volume-Of-Interest Cone-Beam CT
C Huynh1*, M Ruschin2 , S Hashemi2 , Y Lee2 , J Mainprize2 , H Nordstrom3 , M Eriksson3 , V Grouza1 , A Sahgal2 , J Grafe1 , W Song2 , (1) Ryerson University, Toronto, Ontario, (2) Sunnybrook Health Sciences Centre, Toronto, Ontario, (3) Elekta Instrument, AB, Stockholm
Presentations
SU-F-201-3 (Sunday, July 30, 2017) 2:05 PM - 3:00 PM Room: 201
Purpose: To improve the visibility of low contrast tumors in CBCT-guided brain radiosurgery, using scatter corrected, volume-of-interest (VOI) iterative reconstruction.
Methods: The built-in bow-tie filter on the Icon CBCT reduces scatter originating from the sides of the patient. Through modelling the Icon CBCT system in Monte Carlo, a VOI collimator was added to further reduce the scatter-to-primary ratio (SPR). A virtual head phantom with low contrast (ΔHU = 13.9) spheres was used to simulate synthetic, polyenergetic, scatter and noise contaminated projections. Using a VOI collimator that forms an 8x8cm² field size at isocentre, high-mAs projections were simulated. These collimated projections were combined with low-mAs, full-field (22x22cm²) projections to form VOI composite projections. VOI composite projections contain a difference in signal at the boundary of the projected VOI, which can be used to interpolate a scatter correction for the full-field projections. Assuming a perfect scatter correction by removing the known simulated scatter distribution, iterative reconstruction was performed using a small total-variation regularization weight inside the VOI and a larger regularization weight for the noisy image outside the VOI. For comparison, composite VOI images were reconstructed using the standard FDK (Feldkamp-Davis-Kress) algorithm.
Results: For the 18cm mean diameter head phantom used, the mean SPR produced in the VOI region of the projections was 0.91 when only the bow-tie filter was used. When the VOI collimator was used with the bow-tie filter, the mean SPR in the same region was reduced to 0.32. FDK reconstruction of the composite projections yielded a tumor CNR of 0.74. VOI iterative reconstruction was able to reveal the tumor with a CNR of 2.56.
Conclusion: Preliminary results show that the iterative reconstruction of VOI composite projections can improve the soft-tissue visibility of image-guided CBCT. Future work will be done to determine the optimal VOI size and beam intensities.
Funding Support, Disclosures, and Conflict of Interest: Research was funded by FedDev and Elekta.
Contact Email: