(Sunday, 7/29/2018) 2:05 PM - 3:00 PM
Room: Room 202
Purpose: To determine the impact of the base 4D reconstruction on 4D respiratory motion models and the performance of a previously developed motion-compensated four-dimensional cone beam CT (4D-CBCT) reconstruction workflow.
Methods: Previous work demonstrated the ability of motion compensation (MoCo) to improve image CBCT quality. Here, this study assesses the impact of the underlying respiratory motion model. Three distinct respiratory motion models were estimated from groupwise deformable image registration (DIR) of 1) a FDK-based 4D-CBCT reconstruction, 2) a PICCS-based 4D-CBCT reconstruction, and 3) a 4D planning CT image taken at time of simulation (prior). The resulting 4D transformations were used to deform projection data during the backprojection operation of a subsequent FDK reconstruction to create a motion-compensated image. Tissue interface sharpness (TIS) was defined as the slope of a sigmoid curve fit to the lung-diaphragm boundary or to the tissue-airway boundary of the bronchus intermedius when no diaphragm was discernable. Image quality was assessed for 19 subjects by evaluating the mitigation of view-aliasing artifacts, tissue interface sharpness recovery, and noise reduction, and for a 6-subject subset with implanted fiducial markers by evaluating enhancement of fiducial contrast-to-noise ratio (CNR).
Results: FDK-based MoCo was found to significantly improve the fiducial marker CNR (105±55%, p=0.005) compared to FDK reconstruction alone (no MoCo). PICCS- and prior-based MoCo showed no significant improvement (respectively: 55±56%, p=0.123 and 42±62%, p=0.360). All evaluated methods demonstrated similar abilities to recover TIS and reduce image noise and were consistent with previously reported method results. PICCS-based MoCo demonstrated marked reduction in view aliasing artifacts relative to FDK- and prior-based MoCo.
Conclusion: Motion models based on day-of-treatment CBCT imaging have the potential to exceed the performance of prior CT-based motion models when applied to MoCo reconstruction. MoCo reconstruction has the potential to improve the contrast and sharpness of structures in 4D-CBCT images.
Funding Support, Disclosures, and Conflict of Interest: This study was supported by a research grant from the NIH under Award No. R01CA166119. Hugo and Weiss receive research support from Varian Medical Systems. Christensen has licensed technology to Vida Diagnostics. Weiss receives royalties from UpToDate and has two additional NIH grants.
Keywords
Cone-beam CT, Reconstruction, Registration
Taxonomy
Contact Email
