Deep Learning Applications in Ophthalmology

Ehsan Rahimy


Curr Opin Ophthalmol. 2018;29(3):254-260. 

In This Article


Compared with retinal diseases, there have been limited, but expanding, applications of deep learning models within the subspecialty of glaucoma. Given the multifactorial cause of glaucoma, groups have been interested in using deep learning to analyze various inputs, including optic disc photographs, visual fields, as well as OCT of the nerve and peripapillary retina.

In one study, Chen et al.[33] developed a deep learning method for detection of glaucoma based on funduscopic images of the optic disc using two different datasets (ORIGA and SCES) containing glaucoma cases. They reported AUC values for each dataset of 0.831 (ORIGA) and 0.887 (SCES), which they found better than previously reported models.

Asaoka et al.[34] compared a deep learning method [feed-forward neural network (FNN)] with other machine learning methods to differentiate visual fields of preperimetric open-angle glaucoma (OAG) patients (defined as eyes with a glaucomatous optic disc or fundus appearance, or both, and an apparently normal visual field) from those of healthy eyes. In total, 171 preperimetric glaucoma 30–2 visual fields from 51 OAG patients were analyzed with 108 30–2 visual fields from 87 healthy patients. The investigators reported an AUC of 0.926 with the deep learning algorithm, which was significantly greater than other machine learning methods employed.

Muhammad et al.[35] utilized a hybrid deep learning method combined with a single wide-field OCT protocol to distinguish eyes previously classified as either healthy suspects (n = 47) or mild glaucoma (n = 57) based on retinal nerve fiber layer thickness measurements. They reported an accuracy that ranged from 63.7 to 93.1%, depending on the input map. Overall, their findings outperformed standard OCT and visual field clinical metrics in distinguishing eyes that were healthy from those with early glaucoma.