EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Few-Shot and Zero-Shot Learning for MRI Brain Tumor Classification Using CLIP and Vision Transformers

Abir Das iD, Saurabh Singh iD

Sensors · 2025

DOI: 10.3390/s25237341 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, 2025. Verfasst von Abir Das und 1 weiteren Autor:innen. Beteiligte Einrichtung: JW Kim College of Future Studies (JCFS), Woosong University, Daejeon 34606, Republic of Korea. Hauptthemen laut MEDLINE: Magnetic Resonance Imaging, Brain Neoplasms. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Woosong University.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate classification of brain tumors from MRI scans remains challenging due to limited annotated data. This study compares data-efficient paradigms—few-shot learning (FSL) and zero-shot learning (ZSL)—for tumor diagnosis using deep learning and vision–language models. A Prototypical Network (ProtoNet) with CNN, ResNet-18, and vision transformer backbones was evaluated under 1000 randomly sampled five-shot, four-way episodes (mean ± SD). The ResNet-18 ProtoNet achieved 85% ± 8% accuracy (F1 = 0.85), surpassing a fine-tuned ResNet-50 baseline (42% ± 12%) and the CLIP (ZSL) model (30% ± 10%). A visual-only ZSL baseline without text guidance achieved 54% ± 11%. These results highlight that metric-based FSL offers 43% absolute improvement over standard fine-tuning and establishes a robust benchmark for data-efficient MRI classification under severe label constraints.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Magnetic Resonance ImagingBrain NeoplasmsHumansDeep LearningNeural Networks, ComputerMachine LearningAlgorithmsImage Processing, Computer-Assisted

Stichwörter der Autor:innen

CNNMRIbrain tumorsfew-shot learningprototypical networkszero-shot learning

Autor:innen und Institutionen

  • JW Kim College of Future Studies (JCFS), Woosong University, Daejeon 34606, Republic of Korea.
  • Endicott College, Woosong University, Daejeon 34606, Republic of Korea.
  • AI and Big Data Department, Endicott College, Woosong University, Daejeon 34606, Republic of Korea.

ORCID-Profile: Abir Das, Saurabh Singh

Förderung

  • Woosong University

Zitieren

APA

Das, A., & Singh, S. (2025). Few-Shot and Zero-Shot Learning for MRI Brain Tumor Classification Using CLIP and Vision Transformers. Sensors. https://doi.org/10.3390/s25237341

Vancouver

Das A, Singh S. Few-Shot and Zero-Shot Learning for MRI Brain Tumor Classification Using CLIP and Vision Transformers. Sensors. 2025. doi:10.3390/s25237341

BibTeX
@article{Das2025,
  author = {Das, Abir and Singh, Saurabh},
  title = {Few-Shot and Zero-Shot Learning for MRI Brain Tumor Classification Using CLIP and Vision Transformers},
  journal = {Sensors},
  year = {2025},
  doi = {10.3390/s25237341}
}
Kostenlos starten →