Dataset Information

Deep learning-based motion compensation for four-dimensional cone-beam computed tomography (4D-CBCT) reconstruction.

ABSTRACT:

Background

Motion-compensated (MoCo) reconstruction shows great promise in improving four-dimensional cone-beam computed tomography (4D-CBCT) image quality. MoCo reconstruction for a 4D-CBCT could be more accurate using motion information at the CBCT imaging time than that obtained from previous 4D-CT scans. However, such data-driven approaches are hampered by the quality of initial 4D-CBCT images used for motion modeling.

Purpose

This study aims to develop a deep-learning method to generate high-quality motion models for MoCo reconstruction to improve the quality of final 4D-CBCT images.

Methods

A 3D artifact-reduction convolutional neural network (CNN) was proposed to improve conventional phase-correlated Feldkamp-Davis-Kress (PCF) reconstructions by reducing undersampling-induced streaking artifacts while maintaining motion information. The CNN-generated artifact-mitigated 4D-CBCT images (CNN enhanced) were then used to build a motion model which was used by MoCo reconstruction (CNN+MoCo). The proposed procedure was evaluated using in-vivo patient datasets, an extended cardiac-torso (XCAT) phantom, and the public SPARE challenge datasets. The quality of reconstructed images for XCAT phantom and SPARE datasets was quantitatively assessed using root-mean-square-error (RMSE) and normalized cross-correlation (NCC).

Results

The trained CNN effectively reduced the streaking artifacts of PCF CBCT images for all datasets. More detailed structures can be recovered using the proposed CNN+MoCo reconstruction procedure. XCAT phantom experiments showed that the accuracy of estimated motion model using CNN enhanced images was greatly improved over PCF. CNN+MoCo showed lower RMSE and higher NCC compared to PCF, CNN enhanced and conventional MoCo. For the SPARE datasets, the average (± standard deviation) RMSE in mm^-1 for body region of PCF, CNN enhanced, conventional MoCo and CNN+MoCo were 0.0040 ± 0.0009, 0.0029 ± 0.0002, 0.0024 ± 0.0003 and 0.0021 ± 0.0003. Corresponding NCC were 0.84 ± 0.05, 0.91 ± 0.05, 0.91 ± 0.05 and 0.93 ± 0.04.

Conclusions

CNN-based artifact reduction can substantially reduce the artifacts in the initial 4D-CBCT images. The improved images could be used to enhance the motion modeling and ultimately improve the quality of the final 4D-CBCT images reconstructed using MoCo.

SUBMITTER: Zhang Z

PROVIDER: S-EPMC10278033 | biostudies-literature | 2023 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep learning-based motion compensation for four-dimensional cone-beam computed tomography (4D-CBCT) reconstruction.

Zhang Zhehao Z Liu Jiaming J Yang Deshan D Kamilov Ulugbek S US Hugo Geoffrey D GD

Medical physics 20221203 2

<h4>Background</h4>Motion-compensated (MoCo) reconstruction shows great promise in improving four-dimensional cone-beam computed tomography (4D-CBCT) image quality. MoCo reconstruction for a 4D-CBCT could be more accurate using motion information at the CBCT imaging time than that obtained from previous 4D-CT scans. However, such data-driven approaches are hampered by the quality of initial 4D-CBCT images used for motion modeling.<h4>Purpose</h4>This study aims to develop a deep-learning method ...[more]

PMID: 36412165

Similar Datasets

Project description:Data sufficiency are a major problem in four-dimensional cone-beam computed tomography (4D-CBCT) on linear accelerator-integrated scanners for image-guided radiotherapy. Scan times must be in the range of 4-6 min to avoid undersampling artifacts. Various image reconstruction algorithms have been proposed to accommodate undersampled data acquisitions, but these algorithms are computationally expensive, may require long reconstruction times, and may require algorithm parameters to be optimized. The authors present a novel reconstruction method, 4D volume-of-interest (4D-VOI) reconstruction which suppresses undersampling artifacts and resolves lung tumor motion for undersampled 1-min scans. The 4D-VOI reconstruction is much less computationally expensive than other 4D-CBCT algorithms.The 4D-VOI method uses respiration-correlated projection data to reconstruct a four-dimensional (4D) image inside a VOI containing the moving tumor, and uncorrelated projection data to reconstruct a three-dimensional (3D) image outside the VOI. Anatomical motion is resolved inside the VOI and blurred outside the VOI. The authors acquired a 1-min. scan of an anthropomorphic chest phantom containing a moving water-filled sphere. The authors also used previously acquired 1-min scans for two lung cancer patients who had received CBCT-guided radiation therapy. The same raw data were used to test and compare the 4D-VOI reconstruction with the standard 4D reconstruction and the McKinnon-Bates (MB) reconstruction algorithms.Both the 4D-VOI and the MB reconstructions suppress nearly all the streak artifacts compared with the standard 4D reconstruction, but the 4D-VOI has 3-8 times greater contrast-to-noise ratio than the MB reconstruction. In the dynamic chest phantom study, the 4D-VOI and the standard 4D reconstructions both resolved a moving sphere with an 18 mm displacement. The 4D-VOI reconstruction shows a motion blur of only 3 mm, whereas the MB reconstruction shows a motion blur of 13 mm. With graphics processing unit hardware used to accelerate computations, the 4D-VOI reconstruction required a 40-s reconstruction time.4D-VOI reconstruction effectively reduces undersampling artifacts and resolves lung tumor motion in 4D-CBCT. The 4D-VOI reconstruction is computationally inexpensive compared with more sophisticated iterative algorithms. Compared with these algorithms, our 4D-VOI reconstruction is an attractive alternative in 4D-CBCT for reconstructing target motion without generating numerous streak artifacts.

Project description:BackgroundTo evaluate the effects of dose to tumors and organs at risk (OARs) on inter-fractional anatomic changes.MethodsWe evaluated nine patients with cervical cancer treated with intensity-modulated radiotherapy (IMRT) (45 Gy in 25 fractions) using kV cone-beam computed tomography (CBCT) image guidance once or twice a week before treatment. For each patient, the original plan on the computed tomography (CT) image was copied to merged images, and then the fractional doses were calculated. Subsequently, deformable accumulated doses were obtained by summing the fractional absolute doses into a single dose in MIM Maestro software. The volume changes in the target and OARs were compared between the original CT and merged CBCT images, and the differences in the fractional and accumulated doses were also evaluated.ResultsSixty-nine merged CBCT images were obtained and analyzed in this study. For the target areas, the volume changes in the clinical target volume (CTV) and planning target volume (PTV) reached -18.05% and -24.11% at most, respectively. The fractional D2% of the CTV and PTV was generally higher than the original plans, and the accumulated deviations were 2.27%±0.82% (P<0.01) and 2.42%±1.28% (P<0.01), respectively. The fractional D98% of the PTV was underdosed up to 18.28% for 78% of patients, and the accumulated deviations were -2.06% to -17.29% (P<0.05). For the OARs, the bladder volume changes were the most dramatic, reducing up to 93.60%. The fractional Dmean and D2cc of the bladder were generally higher than the original plans, and there were significant differences in their accumulated values (P<0.05). There was no obvious trend of rectal volume change with -69.65% to 74.20%. The rectum Dmean and D2cc of the accumulated were not significantly different from the planned dose (P>0.05).ConclusionsFor patients with cervical cancer, the changes in bladder and rectal volume were greater than in the target volume. Although the volume changes in the bladder and rectum had no significant effect on D98% of the CTV and PTV, they had a significant effect on their own D2cc and the D2% of the CTV and PTV. More attention should be paid to the volume changes in the bladder and rectum in clinical work.

Project description:Under-sampling artifacts are a major problem in four-dimensional cone-beam computed tomography (4D-CBCT) and may compromise evaluation of target motion. Several studies have addressed scan parameter selection for minimizing under-sampling artifacts; however, the role of the target characteristics in scan parameter selection has not been investigated. In this work, the authors evaluated 4D-CBCT performance by assessing the accuracy of target motion measurements for various target sizes and motions. The results may serve as patient-specific guidelines for scan parameters selection in 4D-CBCT.The authors acquired 4D-CBCT scans of a moving phantom consisting of six water-filled sphere targets of sizes 10-37 mm, using various scan times ranging from 30 s to 3 min., setting the motion to 3-s and 6-s periods. The authors used automatic image registration to extract the target motion trajectories and evaluated these measurements for various target sizes and motions over various combinations of scan parameters including scan time, detector configuration, number of respiration phases, and reconstruction filters.The most important object parameter to 4D-CBCT performance was the period of motion. Measurements for the 6-s motion were always systematically less accurate than measurements for the 3-s motion for 34 of 36 objects of various sizes and periods of motion. The 6-s motion required a greater scan time than the 3-s motion for equivalent measurement accuracy. The second most influential parameter was the target size. For the 3-s period of motion, objects larger than 13 mm were tracked with sub-millimeter accuracy with a 1-min scan time. For the 6-s period of motion, objects larger than 22-mm were tracked with sub-millimeter accuracy with a 1.5-min scan time. For all sizes and motions, temporal blurring was observed when the number of phases was fewer than 8. Offset detector configuration provided the same performance as centered detector except for small targets (20 mm.) at short scan times (≤1 min). Finally, reconstruction filtering and number of respiratory phases did not affect performance.Scan time should be set according to target size and motion. The authors have provided figures that provide the minimum scan time needed to achieve the particular motion measurement accuracy for a particular size and motion period.

Dataset Information

Deep learning-based motion compensation for four-dimensional cone-beam computed tomography (4D-CBCT) reconstruction.

Background

Purpose

Methods

Results

Conclusions

Publications

Deep learning-based motion compensation for four-dimensional cone-beam computed tomography (4D-CBCT) reconstruction.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets