Related Experiment Video
Updated: Jun 12, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.7K
A comparative evaluation of deep learning approaches for ophthalmology
Glenn Linde1, Waldir Rodrigues de Souza2,3, Renoh Chalakkal4
1oDocs Eye Care Research, Dunedin, New Zealand.
Scientific Reports
|September 18, 2024
Summary
Artificial intelligence (AI) in ophthalmology leverages machine learning models for image analysis. This review guides researchers in selecting optimal AI-dataset combinations for ophthalmic imaging tasks.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Machine Learning
- Medical Imaging
Background:
- Increasing availability of ophthalmic imaging datasets and open-source machine learning code facilitates AI research.
- Advancements in AI and imaging necessitate careful selection of appropriate AI architectures for specific ophthalmic tasks.
Purpose of the Study:
- To discuss and propose optimal machine learning models and deep learning architectures for ophthalmic imaging tasks.
- To evaluate AI-dataset combinations based on accuracy, training time, deployability, interpretability, and adaptability to small datasets.
Main Methods:
- Extensive review of state-of-the-art AI methods for ophthalmology.
- Training and evaluation of various AI architectures using diverse public and private ophthalmic image datasets (retinal, OCT, 3D OCT).
Main Results:
- Comparison of different AI models and deep learning architectures across various ophthalmic imaging modalities.
- Assessment of AI performance and viability considering factors beyond accuracy, such as computational efficiency and interpretability.
Conclusions:
- The optimal AI-dataset combination is task- and data-dependent.
- This review provides insights into selecting appropriate AI tools for advancing ophthalmic research and clinical practice.
