Click here to


Are you sure ?

Yes, do it No, cancel

Topogram-Guided and Principal-Component-Regularized (TGPCR) Image Reconstruction with Amplitude Binning for 4DCT Under Irregular Breathing

H Gao1*, F Yin2 , (1) Duke University Medical Center, Durham, NC, (2) Duke University Medical Center, Durham, NC


(Tuesday, 7/31/2018) 1:15 PM - 1:45 PM

Room: Exhibit Hall | Forum 5

Purpose: Irregular breathing patterns with changes in amplitude, frequency and baseline are common for patients going through 4DCT scans. Phase binning is often the routine clinical choice, since amplitude binning tends to generate incomplete 4DCT images due to insufficient projection data binned to each phase caused by irregular breathing. However phase binning is subject to severe motion artifacts due to inaccurate binning. This work proposes a new reconstruction method with slow-speed topogram acquisition that potentially enables high-quality 4DCT image reconstruction based on amplitude binning.

Methods: The proposed Topogram-Guided and Principal-Component-Regularized (TGPCR) image reconstruction method utilizes the slow-speed topogram to learn the irregular breathing pattern for improving 4DCT image reconstruction. Specifically, TGPCR utilizes high-temporal-resolution amplitude binning to models high-temporal-resolution 4DCT images as a decomposed product of spatial principal components (SPC) and temporal motion coefficients (TMC), acquires the slow-speed topogram for extracting TMC, reconstructs SPC with the prior TMC using high-temporal-resolution then generate ten-phase 4DCT images using SPC and TMC. Here the key hypothesis of TGPCR is that TMC for 4DCT images can be reasonably approximated by TMC learned from the slow-speed topogram.

Results: The 4DCT simulation studies based on XCAT phantoms using irregular patient breathing patterns are performed to validate TGPCR. 4D total-variation (4DTV) regularized iterative image reconstruction with phase binning had significant reconstructed motion artifact, while 4DTV regularized reconstruction with amplitude binning had incomplete images with missing spatial information.

Conclusion: The proposed TGPCR method is a new 4DCT method that acquires a slow-speed topogram for motion learning and then reconstructs components with accurately binned data via high-temporal-resolution amplitude binning. TGPCR potentially enables high-quality 4DCT image reconstruction for patients with regular breathing.


Not Applicable / None Entered.



Contact Email