EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Cross-Domain Weakly Supervised Diabetic Retinopathy Lesion Identification Method Based on Multiple Instance Learning and Domain Adaptation

Renyu Li, Yunchao Gu, Xinliang Wang, Junjun Pan iD

Bioengineering · 20.09.2023 · 10 (9) · 1100

DOI: 10.3390/bioengineering10091100 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioengineering, online erschienen am 20.09.2023. Verfasst von Renyu Li und 3 weiteren Autor:innen. Beteiligte Einrichtung: State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, Beijing 100191, China. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Technological Innovation 2030- "New Generation Artificial Intelligence" Major Project, Technological Innovation 2030—“New Generation Artificial Intelligence” Major Project, CAMS Innovation Fund for Medical Sciences.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate identification of lesions and their use across different medical institutions are the foundation and key to the clinical application of automatic diabetic retinopathy (DR) detection. Existing detection or segmentation methods can achieve acceptable results in DR lesion identification, but they strongly rely on a large number of fine-grained annotations that are not easily accessible and suffer severe performance degradation in the cross-domain application. In this paper, we propose a cross-domain weakly supervised DR lesion identification method using only easily accessible coarse-grained lesion attribute labels. We first propose the novel lesion-patch multiple instance learning method (LpMIL), which leverages the lesion attribute label for patch-level supervision to complete weakly supervised lesion identification. Then, we design a semantic constraint adaptation method (LpSCA) that improves the lesion identification performance of our model in different domains with semantic constraint loss. Finally, we perform secondary annotation on the open-source dataset EyePACS, to obtain the largest fine-grained annotated dataset EyePACS-pixel, and validate the performance of our model on it. Extensive experimental results on the public dataset FGADR and our EyePACS-pixel demonstrate that compared with the existing detection and segmentation methods, the proposed method can identify lesions accurately and comprehensively, and obtain competitive results using only coarse-grained annotations.

Schlagwörter

Stichwörter der Autor:innen

cross-domaindiabetic retinopathy identificationmultiple instance learningweakly supervised learning

Autor:innen und Institutionen

  • State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, Beijing 100191, China.
  • Hangzhou Innovation Institute, Beihang University, Hangzhou 310051, China.
  • Research Unit of Virtual Body and Virtual Surgery Technologies, Chinese Academy of Medical Sciences, 2019RU004, Beijing 100191, China.

ORCID-Profile: Junjun Pan

Förderung

  • Technological Innovation 2030- "New Generation Artificial Intelligence" Major Project
  • Technological Innovation 2030—“New Generation Artificial Intelligence” Major Project
  • CAMS Innovation Fund for Medical Sciences

Zitieren

APA

Li, R., Gu, Y., Wang, X., & Pan, J. (2023). A Cross-Domain Weakly Supervised Diabetic Retinopathy Lesion Identification Method Based on Multiple Instance Learning and Domain Adaptation. Bioengineering, 10(9), 1100. https://doi.org/10.3390/bioengineering10091100

Vancouver

Li R, Gu Y, Wang X, Pan J. A Cross-Domain Weakly Supervised Diabetic Retinopathy Lesion Identification Method Based on Multiple Instance Learning and Domain Adaptation. Bioengineering. 2023;10(9):1100. doi:10.3390/bioengineering10091100

BibTeX
@article{Li2023,
  author = {Li, Renyu and Gu, Yunchao and Wang, Xinliang and Pan, Junjun},
  title = {A Cross-Domain Weakly Supervised Diabetic Retinopathy Lesion Identification Method Based on Multiple Instance Learning and Domain Adaptation},
  journal = {Bioengineering},
  year = {2023},
  volume = {10},
  number = {9},
  pages = {1100},
  doi = {10.3390/bioengineering10091100}
}
Kostenlos starten →