ai & eye
glaucoma severity from oct
a published fusion model for glaucoma severity. one stream reads the oct scan and the other reads the numerical measurements from the same exam, and the combined model outputs normal, mild to moderate, or severe.
most oct models classify from the image alone and report a binary result. we add the metrics clinicians already use, and grade severity in three classes instead of two.
sections
1 of 4
why oct
glaucoma damages the optic nerve irreversibly, so detection has to happen early.
it is the second leading cause of blindness worldwide after cataract, and an estimated 76 million people had it as of 2020. the standard workup combines optic nerve examination, visual field testing, and intraocular pressure.
optical coherence tomography adds high-resolution structure of the optic disc and retinal nerve fibre layer, often before a patient notices any change. we build on scans and measurements that already exist in the clinic rather than requiring a new imaging protocol.
arrow keys move between sections.
the paper covers the architecture and the clinical reasoning in full.
ramesh, chinasigari, kadadi, et al. tnoa journal of ophthalmic science and research, 2024.