Deep learning for ultra-widefield imaging: a scoping review.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Medical Imaging

Background:

• Ultra-widefield (UWF) imaging provides a broad view of the retina.
• Deep learning (DL) algorithms are increasingly applied in medical image analysis.
• The integration of DL with UWF imaging holds potential for advancing eye care.

Purpose of the Study:

• To conduct a scoping review of deep learning applications in ultra-widefield imaging.
• To identify and summarize the uses of DL in UWF imaging for disease detection, image synthesis, quality assessment, and feature segmentation.
• To provide an overview of the current landscape and future directions of DL in UWF ophthalmology.

Main Methods:

• A comprehensive literature search was conducted across PubMed, Embase, Cochrane Library, and Google Scholar up to August 31st, 2021.
• Inclusion criteria focused on studies utilizing both deep learning and ultra-widefield imaging.
• Exclusion criteria targeted non-English, non-peer-reviewed, or unavailable full-text articles, and studies not employing deep learning.

Main Results:

• Thirty-six studies met the inclusion criteria.
• The majority of studies (23) focused on ophthalmic disease detection and classification.
• Other applications included segmentation (5), generative image synthesis (3), image quality assessment (3), and systemic disease detection (2).

Conclusions:

• Deep learning demonstrates significant efficacy in diagnosing and detecting ophthalmic diseases such as diabetic retinopathy, retinal detachment, and glaucoma using UWF imaging.
• DL is also utilized for generating synthetic ophthalmic images.
• This review highlights the current utility and future potential of DL in conjunction with UWF imaging for ophthalmic applications.