Performance analysis of pretrained convolutional neural network models for ophthalmological disease classification
2022

Analyzing AI Models for Eye Disease Detection

Sample size: 5000 publication Evidence: moderate

Author Information

Author(s): Emir Busra, Colak Ertugrul

Primary Institution: Izmir Katip Celebi University, Turkey; Eskisehir Osmangazi University, Turkey

Hypothesis

This study aimed to evaluate the classification performance of pretrained convolutional neural network models for ophthalmological disease classification.

Conclusion

The pretrained convolutional neural network architectures can effectively identify ophthalmological diseases from fundus images, with ResNet50 performing particularly well for several conditions.

Supporting Evidence

  • ResNet50 achieved an accuracy of 97.1% and was the best model for classifying cataract.
  • VGG16 and Inceptionv3 also performed well but with lower accuracy compared to ResNet50.
  • The study utilized data augmentation techniques to balance the dataset for better model training.

Takeaway

The study used computer programs to help doctors find eye diseases by looking at pictures of the inside of the eye, and some programs were better than others.

Methodology

The study used a dataset of 10,000 fundus images from 5,000 patients, divided into training, validation, and testing sets, to evaluate three CNN architectures: VGG16, Inceptionv3, and ResNet50.

Potential Biases

The dataset was collected from a single ethnicity, which may limit the generalizability of the findings.

Limitations

The dataset may not adequately represent all ethnicities and lacks sufficient images for some rare diseases, which could affect model performance.

Participant Demographics

The dataset included fundus images from 5,000 patients, but specific demographic details were not provided.

Digital Object Identifier (DOI)

10.5935/0004-2749.2022-0124

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication