Related Experiment Videos
Automated detection algorithm for arteriolar narrowing on fundus images
Yuji Hatanaka1, Toshiaki Nakagawa, Akira Aoyama
1Department of Electric Control Engineering, Gifu National College of Technology, Kamimakuwa 2236-2, Motosu 501-0495, Japan (phone: 81-58-320-1384; fax: 81-58-320-1384;
Insights
A new computer-aided diagnosis system (CAD) automates the detection of arteriolar narrowing and focal arteriolar narrowing in fundus images, aiding ophthalmologists in diagnosing hypertensive changes.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer-Aided Diagnosis
Background:
- Hypertensive changes in fundus images are typically identified by ophthalmologists observing arteriolar narrowing.
- Automated detection of these changes can improve diagnostic efficiency.
Purpose of the Study:
- To develop an automated method for detecting arteriolar narrowing and focal arteriolar narrowing in fundus images.
- To assist ophthalmologists in diagnosing ocular diseases through automated vessel analysis.
Main Methods:
- Blood vessel candidates detected using density analysis and centerline tracking.
- Vessel structure established using vector-based direction comparison and intersection analysis.
- Arteriolar narrowing detected via artery-to-vein diameter ratio (A/V ratio); focal narrowing measured by artery diameter.
Main Results:
- Sensitivity of 76% and specificity of 91% for detecting arteriolar narrowing in 100 images.
- Sensitivity of 75% with 2.9 false positives per image for focal arteriolar narrowing in 70 images.
- Ongoing development aims to reduce false positives.
Conclusions:
- The developed automated system shows promising results for detecting abnormal vessels in fundus images.
- This technology has the potential to support ophthalmologists in clinical diagnosis.
- Further refinement is planned to enhance accuracy and reduce false positives.
Abstract:
We have developed a computer-aided diagnosis system (CAD) to detect abnormalities in fundus images. In Japan, ophthalmologists usually detect hypertensive changes by identifying arteriolar narrowing and focal arteriolar narrowing. The purpose of this study is to develop an automated method for detecting arteriolar narrowing and focal arteriolar narrowing on fundus images. The blood vessel candidates were detected by the density analysis method. In blood vessel tracking, a local detection function was used to determine the centerline of the blood vessel. A direction comparison function using three vectors was designed to optimally estimate the next possible location of a blood vessel. After the connectivity of vessel segments was adjusted based on the recognized intersections, the true tree-like structure of the blood vessels was established. The blood vessels were recognized as arteries or veins by hue of HSV color space and their diameters. The arteriolar narrowing was detected by the ratio of diameters (artery vs. vein; A/V ratio). Focal arteriolar narrowing was detected by measuring the diameter of an artery. By applying this method to 100 fundus images, the detection sensitivity for arteriolar narrowing was found to be 76% when the specificity was 91%. Furthermore, by applying this method to 70 other different fundus images, the detection sensitivity for the focal arteriolar narrowing was 75% with 2.9 false positives per image. The number of some false positives is planned to be reduced during the next stage of development. Such an automated detection of abnormal vessels could help ophthalmologists in diagnosing ocular diseases.