Federated Learning in Ocular Imaging: Current Progress and Future Direction

Abstract

Advances in artificial intelligence deep learning (DL) have made tremendous impacts on the field of ocular imaging over the last few years. Specifically, DL has been utilised to detect and classify various ocular diseases on retinal photographs, optical coherence tomography (OCT) images, and OCT-angiography images. In order to achieve good robustness and generalisability of model performance, DL training strategies traditionally require extensive and diverse training datasets from various sites to be transferred and pooled into a “centralised location”. However, such a data transferring process could raise practical concerns related to data security and patient privacy. Federated learning (FL) is a distributed collaborative learning paradigm which enables the coordination of multiple collaborators without the need for sharing confidential data. This distributed training approach has great potential to ensure data privacy among different institutions and reduce the potential risk of data leakage from data pooling or centralisation. This review article aims to introduce the concept of FL, provide current evidence of FL in ocular imaging, and discuss potential challenges as well as future applications.

Federated Learning for Healthcare: A Comprehensive Review

“…GANs have a wide range of applications, including image analysis, and do not need to use Markov chains. When the discriminator's cost function increases, the generator's cost function drops [5].…”

“…Because of the nature of RNNs, their calculations are slow. GANs have a wi of applications, including image analysis, and do not need to use Markov chain the discriminator's cost function increases, the generator's cost function drops [5 Federated learning (FL) [6] is a relatively new method for protecting patien while training deep learning models on federated healthcare data. By avoiding for the transfer of medical data through a centralized aggregate server, this metho for decentralized training of deep learning models [7].…”

Optical Coherence Tomography and Optical Coherence Tomography Angiography in Pediatric Retinal Diseases

“…Challenges to the application of a DL system in pediatric retinal disease include the lack of standardized images, noise in images, artifacts during image acquisition and the smaller sample size in patients with OCT/OCTA in comparison with color fundus photography [ 175 ]. Although AI may be an exciting new frontier, it has inherent issues in implementation, such as sustainability, cost-effectiveness and scalability [ 178 , 179 ].…”

Federated learning for diagnosis of age-related macular degeneration

“…Investigating diabetic retinopathy leveraged FL's potential to develop more generalized models by utilizing diverse datasets without compromising data privacy ( 18 ). Lastly, a comprehensive review by Nguyen et al ( 19 ) emphasized the transformative potential of DL in ocular imaging, with FL providing an effective solution to data security concerns.…”