EUVIMEDEUVIMED entdecken
Crossref

DPI-MoCo: Deep Prior Image Constrained Motion Compensation Reconstruction for 4D CBCT

Dianlin Hu, ChenCheng Zhang, Xuanjia Fei, Yi Yao, Yan Xi, Jin Liu, Yikun Zhang, Gouenou Coatrieux, Jean Louis Coatrieux, Yang Chen

IEEE Transactions on Medical Imaging · 2025-03 · 44 (3) · 1243-1256

DOI: 10.1109/tmi.2024.3483451 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Medical Imaging, 2025-03. Verfasst von Dianlin Hu und 9 weiteren Autor:innen. Hauptthemen laut MEDLINE: Cone-Beam Computed Tomography, Deep Learning, Four-Dimensional Computed Tomography, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 55 Literaturangaben. Gefördert von: National Key Research and Development Program of China, State Key Project of Research and Development Plan, National Natural Science Foundation of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

4D cone-beam computed tomography (CBCT) plays a critical role in adaptive radiation therapy for lung cancer. However, extremely sparse sampling projection data will cause severe streak artifacts in 4D CBCT images. Existing deep learning (DL) methods heavily rely on large labeled training datasets which are difficult to obtain in practical scenarios. Restricted by this dilemma, DL models often struggle with simultaneously retaining dynamic motions, removing streak degradations, and recovering fine details. To address the above challenging problem, we introduce a Deep Prior Image Constrained Motion Compensation framework (DPI-MoCo) that decouples the 4D CBCT reconstruction into two sub-tasks including coarse image restoration and structural detail fine-tuning. In the first stage, the proposed DPI-MoCo combines the prior image guidance, generative adversarial network, and contrastive learning to globally suppress the artifacts while maintaining the respiratory movements. After that, to further enhance the local anatomical structures, the motion estimation and compensation technique is adopted. Notably, our framework is performed without the need for paired datasets, ensuring practicality in clinical cases. In the Monte Carlo simulation dataset, the DPI-MoCo achieves competitive quantitative performance compared to the state-of-the-art (SOTA) methods. Furthermore, we test DPI-MoCo in clinical lung cancer datasets, and experiments validate that DPI-MoCo not only restores small anatomical structures and lesions but also preserves motion information.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Cone-Beam Computed TomographyDeep LearningFour-Dimensional Computed TomographyImage Processing, Computer-AssistedHumansLung NeoplasmsLungArtifactsPhantoms, ImagingAlgorithms

Förderung

  • National Key Research and Development Program of China
  • State Key Project of Research and Development Plan
  • National Natural Science Foundation of China
  • Key Research and Development Programs in Jiangsu Province of China
  • Fundamental Research Funds for the Central Universities

Zitieren

APA

Hu, D., Zhang, C., Fei, X., Yao, Y., Xi, Y., Liu, J., Zhang, Y., Coatrieux, G., Coatrieux, J. L., & Chen, Y. (2025). DPI-MoCo: Deep Prior Image Constrained Motion Compensation Reconstruction for 4D CBCT. IEEE Transactions on Medical Imaging, 44(3), 1243-1256. https://doi.org/10.1109/tmi.2024.3483451

Vancouver

Hu D, Zhang C, Fei X, Yao Y, Xi Y, Liu J, et al. DPI-MoCo: Deep Prior Image Constrained Motion Compensation Reconstruction for 4D CBCT. IEEE Transactions on Medical Imaging. 2025;44(3):1243-1256. doi:10.1109/tmi.2024.3483451

BibTeX
@article{Hu2025,
  author = {Hu, Dianlin and Zhang, ChenCheng and Fei, Xuanjia and Yao, Yi and Xi, Yan and Liu, Jin and Zhang, Yikun and Coatrieux, Gouenou and Coatrieux, Jean Louis and Chen, Yang},
  title = {DPI-MoCo: Deep Prior Image Constrained Motion Compensation Reconstruction for 4D CBCT},
  journal = {IEEE Transactions on Medical Imaging},
  year = {2025},
  volume = {44},
  number = {3},
  pages = {1243--1256},
  doi = {10.1109/tmi.2024.3483451}
}
Kostenlos starten →