EUVIMEDSkopri EUVIMED
Crossref

An unsupervised sparse‐view CT reconstruction framework using combination of iterative deep image prior and ADMM

Jiahao Chang, Shuo Xu, Jintao Fu, Zirou Jiang, Zhentao Wang, Xincheng Xiang, Peng Cong, Yuewen Sun

Medical Physics · 2025-07 · 52 (7)

DOI: 10.1002/mp.17933 ↗

At a glance

Journal article in Medical Physics, published online on 2025-07-15. Written by Jiahao Chang and 7 co-authors. Institution: Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing, China. Major MEDLINE topics: Tomography, X-Ray Computed, Image Processing, Computer-Assisted, Deep Learning, Unsupervised Machine Learning. Cited 1 times so far (Crossref); 47 references.

AstrattSubjectsDettalji biblijografiċiCiteRelated studies

Abstract (original language)

Background

Sparse views and low‐dose scanning reduce radiation exposure computed tomography (CT), but the reconstructed images exhibit severe artifacts and noise due to inadequate view sampling and diminished ray intensity. Recent advances in supervised deep learning (DL) methods have achieved remarkable success in CT reconstruction. However, their reliance on large datasets of paired high‐quality and degraded images has constrained their applicability.

Purpose

In this work, we introduce an unsupervised DL framework called ADMM‐DRP, which integrates an untrained neural network with alternating direction method of multipliers (ADMM) iterative reconstruction algorithm.

Methods

Specifically, the method employs an untrained neural network as an image generator to optimize the data inconsistency in the Radon domain. To avoid the overfitting phenomenon of traditional deep image prior (DIP)‐based methods, we further utilize the ADMM with total variation (TV) regularization continuously update the input of the neural network during the training process.

Results

Experiments on sparse‐view and low‐dose CT reconstruction tasks demonstrate that the proposed framework outperforms conventional supervised and iterative reconstruction methods in terms of metric and visual quality.

Conclusions

ADMM‐DRP reduces the algorithm's reliance on training data, achieves excellent performance in sparse‐view and low‐dose CT reconstruction, and demonstrates substantial potential for further development in medical imaging.

Subjects

MeSH terms (MEDLINE)

Tomography, X-Ray ComputedImage Processing, Computer-AssistedDeep LearningUnsupervised Machine LearningAlgorithmsHumansNeural Networks, Computer

Author keywords

computed tomography (CT)deep learningreconstruction algorithmsparse‐view CTunsupervised learning

Authors and institutions

  • Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing, China.
  • Beijing Key Laboratory of Nuclear Detection, Beijing, China.
  • Hefei Meyer Optoelectronic Technology INC, Hefei, China.

Cite

APA

Chang, J., Xu, S., Fu, J., Jiang, Z., Wang, Z., Xiang, X., Cong, P., & Sun, Y. (2025). An unsupervised sparse‐view CT reconstruction framework using combination of iterative deep image prior and ADMM. Medical Physics, 52(7). https://doi.org/10.1002/mp.17933

Vancouver

Chang J, Xu S, Fu J, Jiang Z, Wang Z, Xiang X, et al. An unsupervised sparse‐view CT reconstruction framework using combination of iterative deep image prior and ADMM. Medical Physics. 2025;52(7). doi:10.1002/mp.17933

BibTeX
@article{Chang2025,
  author = {Chang, Jiahao and Xu, Shuo and Fu, Jintao and Jiang, Zirou and Wang, Zhentao and Xiang, Xincheng and Cong, Peng and Sun, Yuewen},
  title = {An unsupervised sparse‐view CT reconstruction framework using combination of iterative deep image prior and ADMM},
  journal = {Medical Physics},
  year = {2025},
  volume = {52},
  number = {7},
  doi = {10.1002/mp.17933}
}
Start for free →