Encrypted login | home

Program Information

Lateral Truncation Artifact Correction for 4DCBCT-Based Motion Modeling and Dose Assessment

W Cai

S Dhou, F Cifter, M Myronakis, R Berbeco, J Lewis, W Cai*, Brigham and Women's Hospital, Boston, MA and Harvard Medical School, Boston, MA


SU-E-I-3 (Sunday, July 12, 2015) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: To allow accurate motion modeling and dose assessment based on 4DCBCT by addressing the limited field of view (FOV) and lateral truncation artifacts in current clinical CBCT systems. Due to the size and geometry of onboard flat panel detects, CBCT often cannot cover the entire thorax of adult patients. We implement method to extend the images generated from 4DCBCT-based motion models and correct lateral truncation artifacts.

Methods: The method is based on deforming a reference 4DCT image containing the entire patient anatomy to the (smaller) CBCT image within the higher quality CBCT FOV. Next, the displacement vector field (DVF) derived inside the CBCT FOV is smoothly extrapolated out to the edges of the body. These extrapolated displacement vectors are used to generate a new body contour and HU values outside of the CBCT FOV. This method is applied to time-varying volumetric images (3D fluoroscopic images) generated from a 4DCBCT-based motion model at 2 Hz. Six XCAT phantoms are used to test this approach and reconstruction accuracy is investigated.

Results: The normalized root mean square error between the corrected images generated from the 4DCBCT-based motion model and the ground truth XCAT phantom at each time point is generally less than 20%. These results are comparable to results from 4DCT-based motion models. The anatomical structures outside the CBCT FOV can be reconstructed with an error comparable to that inside the FOV. The resulting noise is comparable to that of 4DCT.

Conclusions: The proposed approach can effectively correct the artifact due to lateral truncation in 4DCBCT-based motion models. The quality of the resulting images is comparable to images generated from 4DCT-based motion models. Capturing the body contour and anatomy outside the CBCT FOV makes more reasonable dose calculations possible.

Contact Email: