EUVIMEDSkopri EUVIMED
Crossref

A semi‐supervised multi‐connection contrastive learning framework for x‐ray lung segmentation based on mutual distillation

Xiangrui Zeng, Nibras Abdulla, Baixue Liang, Ola A Alwesabi, Putra Sumari

Medical Physics · 2025

DOI: 10.1002/mp.17984 ↗

At a glance

Journal article in Medical Physics, 2025. Written by Xiangrui Zeng and 4 co-authors. Institution: School of Computer Science, Universiti Sains Malaysia, Penang, Malaysia. Major MEDLINE topics: Lung, Image Processing, Computer-Assisted, Tomography, X-Ray Computed, Deep Learning. 40 references.

AstrattSubjectsDettalji biblijografiċiCiteRelated studies

Abstract (original language)

Background

Deep learning techniques have demonstrated impressive results in medical image segmentation tasks. However, building labeled data for fully supervised models requires significant labor, which can be costly. Additionally, medical devices often prioritize higher data security, smaller size, and improved portability, which typically leads to limited computing and storage capabilities and necessitates offline model deployment.

Purpose

Given these challenges, it is meaningful to conduct in‐depth research on high‐performance tiny offline models suitable for edge deployment. This study aims to pursue a higher‐performance segmentation model while keeping the inference model small enough to be used in clinical practice where computing resources are usually limited.

Methods

This study proposes a semi‐supervised framework based on contrastive learning for developing an organ contour segmentation model using a few labels. The framework employs multiple consistency alignment and mutual distillation mechanisms, in which its backbone can be adapted based on performance or speed requirements.

Results

The framework was tested on three chest X‐ray datasets with 128×128 resolution (JSRT, Montgomery County, and Shenzhen Hospital). When only using two labeled images, the Dice scores of lung segmentation were 0.9636, 95% CI [0.9633, 0.9640], 0.9596, 95% [0.9589, 0.9604], and 0.9527, 95% [0.9508, 0.9546], respectively, and the inference model parameters obtained were only 1.15 M.

Conclusion

These results indicate that the framework performance and potential for edge deployment are leading all similar studies, proving its suitability for clinical applications. (Code Address: https://github.com/mozixr/SemiMCD/tree/main )

Subjects

MeSH terms (MEDLINE)

LungImage Processing, Computer-AssistedTomography, X-Ray ComputedDeep LearningSupervised Machine LearningHumans

Author keywords

contrastive learningdistillationlung image segmentationsemi‐supervisedx‐ray

Authors and institutions

  • School of Computer Science, Universiti Sains Malaysia, Penang, Malaysia.
  • Faculty of Computer Studies, Arab Open University, Jeddah, Saudi Arabia.

Cite

APA

Zeng, X., Abdulla, N., Liang, B., Alwesabi, O. A., & Sumari, P. (2025). A semi‐supervised multi‐connection contrastive learning framework for x‐ray lung segmentation based on mutual distillation. Medical Physics. https://doi.org/10.1002/mp.17984

Vancouver

Zeng X, Abdulla N, Liang B, Alwesabi OA, Sumari P. A semi‐supervised multi‐connection contrastive learning framework for x‐ray lung segmentation based on mutual distillation. Medical Physics. 2025. doi:10.1002/mp.17984

BibTeX
@article{Zeng2025,
  author = {Zeng, Xiangrui and Abdulla, Nibras and Liang, Baixue and Alwesabi, Ola A and Sumari, Putra},
  title = {A semi‐supervised multi‐connection contrastive learning framework for x‐ray lung segmentation based on mutual distillation},
  journal = {Medical Physics},
  year = {2025},
  doi = {10.1002/mp.17984}
}
Start for free →