Related Experiment Video
Updated: Feb 24, 2026
10:39
A Label-Free Segmentation Approach for Intravital Imaging of Mammary Tumor Microenvironment
Published on: May 24, 2022
2.8K
Retina Image Vessel Segmentation Using a Hybrid CGLI Level Set Method
Guannan Chen1, Meizhu Chen2, Jichun Li1
1Key Laboratory of Optoelectronic Science and Technology for Medicine of Ministry of Education, Fujian Provincial Key Laboratory of Photonics Technology, Fujian Normal University, Fuzhou 350007, China.
Biomed Research International
|August 26, 2017
Summary
This study introduces a novel hybrid active contour model for automatic retina image segmentation. The method effectively extracts blood vessels from fundus images, improving diagnosis of eye diseases.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Retina imaging is crucial for diagnosing ophthalmologic diseases.
- Automatic extraction of retinal vessel profiles is essential for image analysis.
- Low contrast in fundus images presents a significant segmentation challenge.
Purpose of the Study:
- To propose a novel hybrid active contour model for automatic fundus image segmentation.
- To enhance the robustness and accuracy of retinal vessel extraction.
- To overcome segmentation difficulties caused by low contrast in retinal images.
Main Methods:
- A hybrid active contour model combining Selective Binary and Gaussian Filtering Regularized Level Set (SBGFRLS) and Local Binary fitting (LBF) models.
- Integration of signed pressure force function and local intensity properties for improved segmentation.
- Evaluation on public datasets: DRIVE and STARE.
Main Results:
- Achieved high accuracy on DRIVE (0.9390) and STARE (0.9409) datasets.
- Demonstrated strong performance with average sensitivity of 0.7358-0.7449 and specificity of 0.9680-0.9690.
- Showed robustness to initial conditions and effectiveness on pathology images compared to traditional methods.
Conclusions:
- The proposed hybrid active contour model provides an effective and robust method for automatic retinal image segmentation.
- This approach offers an easily implementable alternative to supervised vessel extraction methods.
- The model shows promise for improved diagnosis and analysis of ophthalmologic conditions.
