EUVIMEDEUVIMED entdecken
Open Access

Enhanced model iteration algorithm with graph neural network for diffuse optical tomography

Huangjian Yi iD, Ruigang Yang, Yishuo Wang, Yihan Wang iD, Hongbo Guo, Xu Cao iD, Shouping Zhu iD, Xiaowei He iD

Biomedical Optics Express · 01.03.2024 · 15 (3) · 1910

DOI: 10.1364/boe.509775 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Biomedical Optics Express, online erschienen am 27.02.2024. Verfasst von Huangjian Yi und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Information Sciences and Technology, Northwest University, Xi'an, Shaanxi 710069, China. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: National Natural Science Foundation of China, Youth Innovation Team of Shaanxi Provincial Department of Education, National Key Research and Development Program of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Diffuse optical tomography (DOT) employs near-infrared light to reveal the optical parameters of biological tissues. Due to the strong scattering of photons in tissues and the limited surface measurements, DOT reconstruction is severely ill-posed. The Levenberg-Marquardt (LM) is a popular iteration method for DOT, however, it is computationally expensive and its reconstruction accuracy needs improvement. In this study, we propose a neural model based iteration algorithm which combines the graph neural network with Levenberg-Marquardt (GNNLM), which utilizes a graph data structure to represent the finite element mesh. In order to verify the performance of the graph neural network, two GNN variants, namely graph convolutional neural network (GCN) and graph attention neural network (GAT) were employed in the experiments. The results showed that GCNLM performs best in the simulation experiments within the training data distribution. However, GATLM exhibits superior performance in the simulation experiments outside the training data distribution and real experiments with breast-like phantoms. It demonstrated that the GATLM trained with simulation data can generalize well to situations outside the training data distribution without transfer training. This offers the possibility to provide more accurate absorption coefficient distributions in clinical practice.

Autor:innen und Institutionen

  • School of Information Sciences and Technology, Northwest University, Xi'an, Shaanxi 710069, China.
  • The Xi'an Key Laboratory of Radiomics and Intelligent Perception, No. 1 Xuefu Avenue, 710127 Xi'an, Shaanxi, China.
  • School of Life Science and Technology, Xidian University, Xi'an, Shaanxi 710026, China.
  • caoxu327@gmail.com.
  • spzhu@xidian.edu.cn.

ORCID-Profile: Huangjian Yi, Yihan Wang, Xu Cao, Shouping Zhu, Xiaowei He

Förderung

  • National Natural Science Foundation of China
  • Youth Innovation Team of Shaanxi Provincial Department of Education
  • National Key Research and Development Program of China
  • Natural Science Basic Research Program of Shaanxi Province
  • Research Fund for Young Star of Science and Technology in Shaanxi Province

Zitieren

APA

Yi, H., Yang, R., Wang, Y., Wang, Y., Guo, H., Cao, X., Zhu, S., & He, X. (2024). Enhanced model iteration algorithm with graph neural network for diffuse optical tomography. Biomedical Optics Express, 15(3), 1910. https://doi.org/10.1364/boe.509775

Vancouver

Yi H, Yang R, Wang Y, Wang Y, Guo H, Cao X, et al. Enhanced model iteration algorithm with graph neural network for diffuse optical tomography. Biomedical Optics Express. 2024;15(3):1910. doi:10.1364/boe.509775

BibTeX
@article{Yi2024,
  author = {Yi, Huangjian and Yang, Ruigang and Wang, Yishuo and Wang, Yihan and Guo, Hongbo and Cao, Xu and Zhu, Shouping and He, Xiaowei},
  title = {Enhanced model iteration algorithm with graph neural network for diffuse optical tomography},
  journal = {Biomedical Optics Express},
  year = {2024},
  volume = {15},
  number = {3},
  pages = {1910},
  doi = {10.1364/boe.509775}
}
Kostenlos starten →