Glaucoma is an eye-disease where patients lose vision gradually due to increased pressure in the eye. Over 2.2 million Americans had glaucoma and globally the number is 60.5 million. It is irreversible, meaning that medical treatment can slow your vision from worsening but it cannot recover the vision that you have already lost. Hence, early detection is crucial. However, detection of glaucoma requires an ophthalmologist, and this is a big problem in low-resource settings. For example, in certain parts of rural india, ophthalmologist to population ratio is 1:100,000. The ophthalmologists simply cannot handle such a large population. Fundux seeks to bridge this gap.

Fundux is an automatic screening app for glaucoma that hopes to screen out the patients that are highly unlikely to have glaucoma. This way, only those who are likely to have glaucoma will be sent to the doctors, easing the pressure on ophthalmologists in low resource settings.

This is achieved by measuring the cup-to-disc ratio in fundus images. The cup to disc ratio is one of the most indicative signals of glaucoma. Fundux automatically calculate the cup to disc ratio by computer vision techniques that I specified at the Demo presentation on stage (some combination of edge detection and morphological dilation).

Next step is to get hold of a labeled database and transform this computer vision problem into a supervised learning problem and also create a web-based cloud platform for users to upload fundus images.

I have had this idea for a long time, and I have been mentally thinking about the algorithm constantly. I have implemented different versions before that doesn't work so well. Fundux is a new implementation of this idea that I created at Yhack with a new way of tackling the problem. It actually works quite well.

If you want to learn more about glaucoma, visit glaucoma.org for more information

Contact me at etam2@jhu.edu for any questions.

Built With

Share this project:
×

Updates