SSVT: Self-Supervised Vision Transformer for Eye Disease Diagnosis Based on Fundus Images.

Machine learning-based fundus image diagnosis technologies trigger worldwideinterest owing to their benefits such as reducing medical resource power andproviding objective evaluation results. However, current methods are commonlybased on supervised methods, bringing in a heavy workload to biomedical staffand hence suffering in expanding effective databases. To address this issue, inthis article, we established a label-free method, name 'SSVT',which canautomatically analyze un-labeled fundus images and generate high evaluationaccuracy of 97.0two datasets collected by Beijing Tongren Hospital. The promising resultsshowcased the effectiveness of the proposed unsupervised learning method, andthe strong application potential in biomedical resource shortage regions toimprove global eye health.