Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jun 5, 2026

Visualizing Visual Adaptation
04:43

Visualizing Visual Adaptation

Published on: April 24, 2017

RFD: A Reducing Feature Discrepancy method for unsupervised cross-modality SAM adaptation.

Ji Xia1, Zhehan Shen1, Wei Xia1

  • 1Department of Radiology, Ruijin Hospital, College of Health Science and Technology, Shanghai Jiao Tong University School of Medicine, Shanghai, China.

Computerized Medical Imaging and Graphics : the Official Journal of the Computerized Medical Imaging Society
|June 3, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Multimodal Bidirectional Direct Preference Optimization and Instruction Fine-Tuning for Medical Image Understanding and Generation.

IEEE journal of biomedical and health informatics·2026
Same authorSame journal

WDBDM: Wavelet-based dual-branch diffusion model for low-dose CT and PET denoising.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
Same author

Stellate ganglion block attenuates gut barrier injury in sleep-deprived rats in a gut microbiota-dependent manner.

Scientific reports·2026
Same author

Ultraviolet-driven self-repair in chimeric d(GAUU) outcompetes damage formation.

Chemical communications (Cambridge, England)·2026
Same author

Early Screening for Cervical Squamous Cell Carcinoma and Precancerous Lesions Based on Surface-Enhanced Raman Spectroscopy Microarray Chips.

International journal of nanomedicine·2026
Same author

Peripheral blood IFN-γ-producing T-cell subsets and soluble IL-2 receptor as independent prognostic biomarkers in NSCLC treated with immune checkpoint inhibitor-based therapy.

Discover oncology·2026
Same journal

ScribSAM: A robust scribble-supervised framework for spatiotemporal segmentation of breast lesions in ultrasound videos.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
Same journal

Anatomically and biochemically guided deep image prior for sodium MRI denoising.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
Same journal

Segment Anything Model for medical image segmentation: A review.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
Same journal

HiCAF-Net: A Hierarchical Cross-Attention Fusion framework for cross-cancer subtype classification using histopathological and genomic data.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
Same journal

ADLS: Alignment of discrete latent spaces for unsupervised cross-modality medical image translation.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Society·2026
See all related articles
This summary is machine-generated.

This study introduces a Reducing Feature Discrepancy (RFD) method to enhance the Segment Anything Model (SAM) for unsupervised cross-modality medical image segmentation. RFD significantly improves segmentation accuracy by reducing feature discrepancies, outperforming existing methods.

Area of Science:

  • Medical Image Analysis
  • Computer Vision
  • Artificial Intelligence

Background:

  • Vision foundation models like Segment Anything Model (SAM) excel in supervised medical image segmentation but struggle with unsupervised cross-modality tasks.
  • Accurate masks are crucial for SAM, and performance degrades significantly without them in unsupervised settings.
  • Reducing the need for manual masking and improving unsupervised performance are key challenges.

Purpose of the Study:

  • To propose a novel method, Reducing Feature Discrepancy (RFD), to improve SAM's performance in unsupervised cross-modality medical image segmentation.
  • To reduce the cost associated with manual masking while maintaining high segmentation accuracy.
  • To address the performance degradation of SAM in unsupervised cross-modality scenarios.

Main Methods:

Keywords:
Contrastive learningMedical image segmentationOptimal transportSegment Anything ModelUnsupervised domain adaptation

More Related Videos

Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities
07:13

Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities

Published on: October 27, 2023

Related Experiment Videos

Last Updated: Jun 5, 2026

Visualizing Visual Adaptation
04:43

Visualizing Visual Adaptation

Published on: April 24, 2017

Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities
07:13

Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities

Published on: October 27, 2023

  • Integrating medical adapters into the SAM encoder to enhance feature capture.
  • Employing Structural Prototype-based Contrastive Learning (SPCL) with structural information distance for improved feature discrimination.
  • Utilizing Inexact Prototype-based Pixel Transport (IPPT) for dynamic prototype selection and allocation in online clustering.
  • Implementing Reweighting Unbalanced Feature Adaptation (RUFA) to promote inter-class discrimination and intra-class compactness using SAM latent space features.

Main Results:

  • The proposed RFD method, incorporating RUFA and SPCL, significantly improves unsupervised cross-modality medical image segmentation performance.
  • Experimental results show substantial gains over state-of-the-art unsupervised domain adaptation (UDA) and SAM-based methods on four public datasets.
  • The method demonstrates effectiveness and generalization capabilities across different medical imaging modalities.

Conclusions:

  • The RFD method effectively enhances SAM for unsupervised cross-modality medical image segmentation by minimizing feature discrepancies.
  • The integration of SPCL and RUFA offers a robust approach to improving feature discrimination and adaptation.
  • This work provides a valuable solution for accurate and efficient medical image segmentation without extensive manual annotation.