Related Experiment Video
Updated: Apr 30, 2026
Quantification of Vascular Parameters in Whole Mount Retinas of Mice with Non-Proliferative and Proliferative Retinopathies
Published on: March 12, 2022
Vessel Segmentation in Retinal Images Using Multi-scale Line Operator and K-Means Clustering
Vahid Mohammadi Saffarzadeh1, Alireza Osareh1, Bita Shadgar1
1Department of Computer Engineering, Shahid Chamran University of Ahvaz, Khuzestan, Iran.
Insights
This study presents a new method for detecting retinal blood vessels, even with lesions. The algorithm effectively identifies vessels in normal and abnormal fundus images, achieving high accuracy.
Area of Science:
- Ophthalmology
- Medical Image Analysis
- Computer Vision
Background:
- Accurate detection of retinal blood vessels is crucial for diagnosing various eye conditions.
- Lesions in retinal images, such as bright and dark spots, significantly complicate vessel segmentation.
- Existing methods often struggle with the presence of these artifacts, impacting diagnostic reliability.
Purpose of the Study:
- To develop and evaluate a robust method for retinal blood vessel detection in both normal and abnormal fundus images.
- To address the challenges posed by bright and dark lesions during vessel segmentation.
- To improve the accuracy and reliability of automated retinal image analysis.
Main Methods:
- A novel algorithm employing K-means segmentation in a perceptive color space to mitigate the influence of bright lesions.
- Utilization of a multi-scale line operator designed to detect linear vessel structures while distinguishing them from dark lesions.
- Validation of the proposed method on the publicly available STARE and DRIVE retinal image databases.
Main Results:
- The algorithm demonstrated high performance in vessel localization accuracy.
- Achieved a localization accuracy of 0.9483 on the STARE dataset.
- Achieved a localization accuracy of 0.9387 on the DRIVE dataset.
Conclusions:
- The proposed method effectively detects retinal blood vessels in the presence of challenging lesions.
- The algorithm shows promising results for automated analysis of retinal fundus images.
- This technique can enhance the diagnostic capabilities in ophthalmology by providing reliable vessel segmentation.
Abstract:
Detecting blood vessels is a vital task in retinal image analysis. The task is more challenging with the presence of bright and dark lesions in retinal images. Here, a method is proposed to detect vessels in both normal and abnormal retinal fundus images based on their linear features. First, the negative impact of bright lesions is reduced by using K-means segmentation in a perceptive space. Then, a multi-scale line operator is utilized to detect vessels while ignoring some of the dark lesions, which have intensity structures different from the line-shaped vessels in the retina. The proposed algorithm is tested on two publicly available STARE and DRIVE databases. The performance of the method is measured by calculating the area under the receiver operating characteristic curve and the segmentation accuracy. The proposed method achieves 0.9483 and 0.9387 localization accuracy against STARE and DRIVE respectively.
