Related Experiment Video
Updated: May 8, 2026
12:28
Quantification of Vascular Parameters in Whole Mount Retinas of Mice with Non-Proliferative and Proliferative Retinopathies
Published on: March 12, 2022
Accurate estimation of retinal vessel width using bagged decision trees and an extended multiresolution Hermite model
Carmen Alina Lupaşcu1, Domenico Tegolo, Emanuele Trucco
1VAMPIRE Project, Dipartimento di Matematica e Informatica, Università degli Studi di Palermo, Via Archirafi 34, 90123 Palermo, Italy.
Medical Image Analysis
|September 5, 2013
Summary
This study introduces a new algorithm for accurately measuring retinal vessel width in fundus images. The method demonstrates superior accuracy and stability compared to existing techniques.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Accurate measurement of retinal vessel width is crucial for diagnosing and monitoring various eye conditions, including diabetic retinopathy.
- Existing methods for retinal vessel width estimation often lack accuracy and stability.
Purpose of the Study:
- To develop and validate a novel algorithm for precise retinal vessel width estimation.
- To compare the performance of the new algorithm against established methods using public and clinical datasets.
Main Methods:
- A parametric surface model of vessel cross-sectional intensities was employed.
- Ensembles of bagged decision trees were utilized for local width estimation.
- The algorithm was tested on the REVIEW database and a custom dataset from diabetic retinopathy screening.
Main Results:
- The proposed algorithm achieved significantly higher accuracy than leading methods on both test datasets.
- The method demonstrated 100% stability, providing meaningful measurements for all tested images.
- Performance was found to be highly dependent on training data selection and testing conditions.
Conclusions:
- The novel algorithm offers a robust and accurate solution for retinal vessel width estimation.
- The findings highlight the importance of data selection for optimizing algorithm performance.
- This method holds potential for improved diagnosis and management of retinal diseases.
