A color fundus image quality assessment method based on ensemble learning

By employing an ensemble learning approach that combines ResNet-18 and U-Net models, the quality of color fundus images is comprehensively evaluated. This addresses the issue of low accuracy in existing technologies, enabling real-time and efficient image quality assessment. It is suitable for ophthalmic diagnosis, reduces repetitive imaging processes, and improves diagnostic efficiency.

CN118968266BActive Publication Date: 2026-06-02EAST CHINA UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
EAST CHINA UNIV OF TECH
Filing Date
2024-07-15
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing methods for assessing the quality of color fundus images cannot effectively balance traditional image features and pathological structural features, resulting in low accuracy on large-scale datasets and failing to meet the requirements for real-time and efficient quality assessment.

Method used

An ensemble learning-based approach, combining the ResNet-18 and U-Net models, is used to classify and segment color fundus images. Image quality is comprehensively evaluated through ensemble learning, taking into account both traditional image features and pathological structural features.

Benefits of technology

It enables real-time and efficient quality evaluation of color fundus images, allowing for on-site determination of image suitability for ophthalmic diagnosis, reducing repetitive imaging processes, saving time, alleviating the burden on examiners and patients, and improving diagnostic efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968266B_ABST
    Figure CN118968266B_ABST
Patent Text Reader

Abstract

The present application relates to the field of electronic information technology, and in particular to a color fundus image quality evaluation method based on ensemble learning, which comprises data set division, and the data set comprises a color fundus IQA data set and a segmentation data set, 80% of the data of the data set is randomly selected as a training set, and the remaining 20% of the data is a validation set. ResNet-18 model and U-Net model are trained separately without interference. 50% (proportion adjusted according to data volume) of the images in the color fundus IQA data set training set are manually labeled with 5 features, and the 5 features are contrast, focus, illumination, shadow and reflection. The CF image quality evaluation method based on ensemble learning provided in the present application meets the actual needs of an ophthalmic clinic and can comprehensively evaluate the CF image taken on site to determine whether the image is suitable for ophthalmologist diagnosis.
Need to check novelty before this filing date? Find Prior Art