Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Sep 26, 2025

Positron Emission Tomography-based Dose Painting Radiation Therapy in a Glioblastoma Rat Model using the Small Animal Radiation Research Platform

Positron Emission Tomography-based Dose Painting Radiation Therapy in a Glioblastoma Rat Model using the Small Animal Radiation Research Platform

Published on: March 24, 2022

SAM: Self-Supervised Learning of Pixel-Wise Anatomical Embeddings in Radiological Images.

Ke Yan, Jinzheng Cai, Dakai Jin

IEEE Transactions on Medical Imaging

|April 20, 2022

Summary

This summary is machine-generated.

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

From Slice to Sequence: Autoregressive Tracking Transformer for Consistent 3D Lymph Node Detection in CT Scans.

IEEE transactions on medical imaging·2026

Same author

Clinical Knowledge-Guided PET/CT Lesion Segmentation with Interpretable Fusion of Metabolic and Structural Cues.

IEEE transactions on medical imaging·2026

Same author

Preoperative Prediction of Esophageal Cancer Survival in CT via Tumor and Lymph Node Context and Geometry Modeling.

IEEE transactions on medical imaging·2026

Same author

Clinical Validation of a Deep Learning-Based 2D Ultrasound Steatosis Algorithm: Cutoff Transferability, Scanner Generalizability, and Comparison with FibroScan.

Diagnostics (Basel, Switzerland)·2026

Same author

Pretreatment CT Identification of Extranodal Extension in Laryngeal and Hypopharyngeal Cancers Using Deep Learning.

Radiology·2026

Same author

DistAL: A Domain-Shift Active Learning Framework With Transferable Feature Learning for Lesion Detection.

IEEE transactions on medical imaging·2025

Self-supervised Anatomical Embedding (SAM) enables precise anatomical structure localization in medical images without extensive labeled data. This approach significantly improves landmark detection and lesion matching across diverse imaging modalities.

Area of Science:

Medical Image Analysis
Artificial Intelligence
Computer Vision

Background:

Accurate localization of anatomical structures in radiological images is crucial for medical diagnosis and treatment.
Current methods like landmark detection and semantic segmentation require large labeled datasets, limiting their universal applicability.
A universal approach learning intrinsic anatomical structures from unlabeled data is needed.

Purpose of the Study:

To introduce Self-supervised Anatomical Embedding (SAM), a novel method for generating semantic pixel embeddings to describe anatomical locations.
To enable reliable localization of anatomical structures across varying medical images using unlabeled data.
To demonstrate SAM's effectiveness in various medical image analysis tasks.

Main Methods:

More Related Videos

Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography

Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography

Published on: March 26, 2020

Bioprinting of Hydrogel Tumor Slices as a 3D Model for Mantle Cell Lymphoma

Bioprinting of Hydrogel Tumor Slices as a 3D Model for Mantle Cell Lymphoma

Published on: September 12, 2025

Related Experiment Videos

Last Updated: Sep 26, 2025

Positron Emission Tomography-based Dose Painting Radiation Therapy in a Glioblastoma Rat Model using the Small Animal Radiation Research Platform

Positron Emission Tomography-based Dose Painting Radiation Therapy in a Glioblastoma Rat Model using the Small Animal Radiation Research Platform

Published on: March 24, 2022

Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography

Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography

Published on: March 26, 2020

Bioprinting of Hydrogel Tumor Slices as a 3D Model for Mantle Cell Lymphoma

Bioprinting of Hydrogel Tumor Slices as a 3D Model for Mantle Cell Lymphoma

Published on: September 12, 2025

Developed a pixel-level contrastive learning framework for generating semantic embeddings.
Implemented a coarse-to-fine strategy to encode both global and local anatomical information.
Utilized negative sample selection strategies to enhance embedding discriminability for nearest neighbor searching.

Main Results:

SAM achieved superior performance in landmark detection on a chest CT dataset compared to registration algorithms, with a 0.23-second inference time.
On X-ray datasets, SAM with one labeled image outperformed supervised methods trained on 50 labeled images.
Achieved 91% accuracy in whole-body follow-up lesion matching in CT scans.

Conclusions:

SAM offers a universal and efficient approach for anatomical structure localization in medical imaging.
The method significantly reduces the need for large labeled datasets, making it highly adaptable.
SAM shows potential for improving image registration and initializing convolutional neural network (CNN) weights.