Related Experiment Video
Updated: Apr 15, 2026
07:23
Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography
Published on: March 26, 2020
9.2K
Adaptive thresholding technique for retinal vessel segmentation based on GLCM-energy information
Temitope Mapayi1, Serestina Viriri1, Jules-Raymond Tapamo2
1School of Mathematics, Statistics & Computer Science, University of KwaZulu-Natal, Durban 4000, South Africa.
Computational and Mathematical Methods in Medicine
|March 25, 2015
Summary
This study introduces a new local adaptive thresholding method for retinal vessel segmentation using GLCM-energy information. The technique offers a robust, time-efficient solution with high accuracy and sensitivity on standard datasets.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Retinal vessel segmentation is crucial for diagnosing various eye conditions.
- Existing methods often lack robustness and efficiency.
- A need exists for improved automated segmentation techniques.
Purpose of the Study:
- To develop and evaluate a novel local adaptive thresholding technique for retinal vessel segmentation.
- To enhance the robustness and time efficiency of retinal image analysis.
- To compare the proposed method against existing techniques.
Main Methods:
- A local adaptive thresholding technique was developed utilizing Gray Level Co-occurrence Matrix (GLCM) energy information.
- Thresholds were computed based on GLCM-energy for segmentation.
- The method was tested using grayscale intensity and the Green Channel of retinal images.
Main Results:
- The proposed technique achieved high performance on the DRIVE and STARE databases.
- Maximum average accuracy rates of 0.9511 (DRIVE) and 0.9510 (STARE) were recorded.
- Maximum average sensitivity rates of 0.7650 (DRIVE) and 0.7641 (STARE) were achieved, demonstrating superior performance over previous methods.
Conclusions:
- The developed local adaptive thresholding technique is a robust and time-efficient method for retinal vessel segmentation.
- The technique demonstrates high average accuracy and sensitivity, comparable to very good specificity.
- This method shows significant potential for clinical applications in automated retinal image analysis.
