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

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

Multiple contexts and frequencies aggregation network for deepfake detection.

PloS one·2026

Same author

Application of a Micropatterned Cocultured Hepatocyte System To Predict Preclinical and Human-Specific Drug Metabolism.

Drug metabolism and disposition: the biological fate of chemicals·2015

Same author

Misshapen/NIK-related kinase (MINK1) is involved in platelet function, hemostasis, and thrombus formation.

Blood·2015

Same author

Solid-State Thin-Film Supercapacitors with Ultrafast Charge/Discharge Based on N-Doped-Carbon-Tubes/Au-Nanoparticles-Doped-MnO2 Nanocomposites.

Nano letters·2015

Same author

Genetics, Receptor Binding, Replication, and Mammalian Transmission of H4 Avian Influenza Viruses Isolated from Live Poultry Markets in China.

Journal of virology·2015

Same author

Comparative Study of Degradation Behavior of Bioresorbable Cardiovascular Scaffolds.

Cardiovascular engineering and technology·2015

Search research articles

Related Experiment Video

Updated: Jun 3, 2025

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Published on: May 24, 2022

Adapting SAM2 Model from Natural Images for Tooth Segmentation in Dental Panoramic X-Ray Images.

Zifeng Li¹, Wenzhong Tang¹, Shijun Gao¹

¹School of Aeronautic Science and Engineering, Beihang University, 37 Xueyuan Road, Haidian District, Beijing 100191, China.

Entropy (Basel, Switzerland)

|January 8, 2025

Summary

This study introduces an efficient deep learning method for dental X-ray segmentation using a fine-tuned SAM2 model and knowledge distillation. The proposed LightUNet model achieves high accuracy with significantly reduced parameters and inference time, enabling edge device deployment.

Keywords:

SAM2 X-ray deep learning knowledge distillation segmentation small sample dataset tooth

More Related Videos

Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures

Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures

Published on: August 5, 2021

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Published on: February 23, 2024

Related Experiment Videos

Last Updated: Jun 3, 2025

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Guided Endodontics: Three-Dimensional Planning and Template-Aided Preparation of Endodontic Access Cavities

Published on: May 24, 2022

Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures

Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures

Published on: August 5, 2021

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization

Published on: February 23, 2024

Area of Science:

Medical Imaging
Computer Vision
Artificial Intelligence

Background:

Dental panoramic X-ray imaging is cost-effective and low-dose but faces challenges in accurate tooth segmentation due to image quality and limited datasets.
Traditional deep learning models struggle with overfitting and generalization on dental X-ray data, and high-precision models demand substantial computational resources.
Accurate tooth segmentation is vital for dental diagnostics, lesion analysis, and treatment planning.

Purpose of the Study:

To develop an accurate and efficient tooth segmentation method for dental X-ray images.
To address the limitations of traditional deep learning models, including overfitting and high computational costs.
To enable the deployment of advanced segmentation models on resource-constrained edge devices.

Main Methods:

Fine-tuning the pre-trained Segment Anything Model 2 (SAM2) with adapter modules for dental images.
Incorporating ScConv modules and gated attention mechanisms to enhance semantic understanding and multi-scale feature extraction.
Employing knowledge distillation to train a smaller, efficient model (LightUNet) from the fine-tuned SAM2 teacher model.

Main Results:

The proposed method significantly outperforms the traditional UNet model in segmentation accuracy metrics like IoU on the UFBA-UESC dataset.
The LightUNet model demonstrates improved robustness, especially with limited sample datasets.
LightUNet achieves comparable performance to UNet with only 1.6% of its parameters and 24.0% of its inference time.

Conclusions:

The developed tooth segmentation method effectively improves accuracy and robustness for dental X-ray analysis.
The LightUNet model offers a computationally efficient solution, suitable for deployment on edge devices in clinical settings.
This approach enhances the practical application of deep learning in dental diagnostics, particularly with limited data and resources.