EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MedZeroSeg: Zero-shot medical image segmentation via vision foundation models

Ronghui Zhang iD, Min Huang, Rui Li

PLOS One · 27.03.2026 · 21 (3) · e0344978

DOI: 10.1371/journal.pone.0344978 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 27.03.2026. Verfasst von Ronghui Zhang und 2 weiteren Autor:innen. Beteiligte Einrichtung: Concord University College Fujian Normal University, Fuzhou, China. Hauptthemen laut MEDLINE: Image Processing, Computer-Assisted, Diagnostic Imaging. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research on Multi-Instrument Automatic Reading Method Based on Character Recognition and Key Point Detection for the Fujian Natural Science Foundation General…, Education and Research Project for Young and Middle-aged Teachers by the Department of Education of Fujian Province, Project of the Fujian Provincial Department of Science and Technology on the Research of Cavity Decay OTDR Distributed Sensing Technology and Current Detection….

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A novel medical image segmentation framework, MedZeroSeg , is proposed to address key challenges in the field. Leveraging vision foundation models such as CLIP (Contrastive Language-Image Pre-training) and SAM (Segment Anything Model), it achieves zero-shot segmentation, accurately delineating previously unseen medical images without requiring additional labeled data. This significantly reduces reliance on large-scale annotated datasets. At its core, MedZeroSeg introduces a Dual-Path Feature Extraction Module that captures both fine anatomical details and global contextual information through the integration of local and global perception mechanisms, enhancing robustness against the complexity and variability inherent in medical imaging.Additionally, a Context-Enhanced Hard-Negative Contrast Loss is introduced to enhance contrastive learning by exploiting contextual cues and refining hard-negative sampling, leading to better representations and higher efficiency. The key innovation of MedZeroSeg lies in its ability to leverage generalizable knowledge from CLIP and SAM without any task-specific fine-tuning, making it highly adaptable across different medical imaging modalities. Extensive experiments on three publicly available datasets, including cardiac MRI (ACDC), multi-organ abdominal CT (Synapse), and chest X-ray (COVID-QU-Ex), demonstrate that MedZeroSeg achieves superior results in both zero-shot and weakly supervised segmentation settings, showcasing strong generalization capabilities and minimal data dependency. The framework represents a significant advancement in medical image analysis and opens up promising directions for future research in applying advanced foundation models and innovative learning strategies to healthcare applications.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Image Processing, Computer-AssistedDiagnostic ImagingHumansMagnetic Resonance ImagingAlgorithmsTomography, X-Ray Computed

Autor:innen und Institutionen

  • Concord University College Fujian Normal University, Fuzhou, China.
  • College of Electronic Information Science, Fujian Jiangxia University, Fuzhou, China.
  • Smart Home Information Collection and Processing on Internet of Things Laboratory of Digital Fujian, Fuzhou, China.
  • School of Finance and Economics, Jimei University, Xiamen, China.

ORCID-Profile: Ronghui Zhang

Förderung

  • Research on Multi-Instrument Automatic Reading Method Based on Character Recognition and Key Point Detection for the Fujian Natural Science Foundation General…
  • Education and Research Project for Young and Middle-aged Teachers by the Department of Education of Fujian Province
  • Project of the Fujian Provincial Department of Science and Technology on the Research of Cavity Decay OTDR Distributed Sensing Technology and Current Detection…
  • the Fujian Jiangxia University Research Talent Cultivation Project

Zitieren

APA

Zhang, R., Huang, M., & Li, R. (2026). MedZeroSeg: Zero-shot medical image segmentation via vision foundation models. PLOS One, 21(3), e0344978. https://doi.org/10.1371/journal.pone.0344978

Vancouver

Zhang R, Huang M, Li R. MedZeroSeg: Zero-shot medical image segmentation via vision foundation models. PLOS One. 2026;21(3):e0344978. doi:10.1371/journal.pone.0344978

BibTeX
@article{Zhang2026,
  author = {Zhang, Ronghui and Huang, Min and Li, Rui},
  title = {MedZeroSeg: Zero-shot medical image segmentation via vision foundation models},
  journal = {PLOS One},
  year = {2026},
  volume = {21},
  number = {3},
  pages = {e0344978},
  doi = {10.1371/journal.pone.0344978}
}
Kostenlos starten →