Application of deep learning systems for detecting glaucomatous optic neuropathy
Glaucoma is one of the leading causes of irreversible vision loss and blindness worldwide. Early screening and diagnosis of glaucoma is made difficult due to the asymptomatic nature of the disease in its early stages and subjective approaches to diagnosis. Recent evidence has indicated that deep learning based artificial intelligence screening systems which have been applied to the detection of ophthalmic diseases can achieve excellent sensitivities and specificities, suggesting a revolutionary improvement in disease screening. The purpose of our research is to further evaluate the performance of a deep learning system for the detection of referable glaucomatous optic neuropathy in different population settings.
Using fundus and OCT images as the primary inputs, we will further validate the accuracy of a deep learning algorithm to identify referable glaucomatous optic neuropathy, which has previously been developed and validated basing on 48,116 fundus images from different clinical settings in China.
Based on the above, we are requesting data from all patients (68,151) that have had retinal imaging performed in UK Biobank. The validation of our deep learning algorithm will be based on medical information such as age, gender and diagnoses, and results from eye measurements, such as intraocular pressure (IOP), fundus photographs and OCT.
If the efficacy of this deep learning algorithm is successfully proven, this work will help to improve the detection of glaucoma and potentially other ophthalmic diseases. Other vital potential benefits include improving screening coverage and reducing healthcare costs due to earlier diagnosis.