Ocular Disease Prediction

• 

  abnormality test case

• 

  age related macular test case

• 

  cataract test case

• 

  diabetic test case

• 

  glaucoma test case

• 

  myopia test case

Inspiration

Since deep learning in image processing is getting more and more important, I would like to develop an application that could assist medical staff in distinguishing ocular disease or even just give a reference for an initial judgment.

What it does

It uses deep learning (VGG19) to process images of retinal image data and distinguish the disease it has.

How we built it

Since building up a general model to make the prediction has a relatively low accuracy, I separate different diseases and build models separately, and then combine them to make an optimized ensemble model.

Challenges we ran into

The time for building models is long, and adjustment for pursuing a higher accuracy costs a lot of time. Therefore, I am not available to embed the final model since it's not well-built yet.

Accomplishments that we're proud of

I successfully made a full-stack building process on my own, which was my first time.

What we learned

I've learned how to build models by using VGG19, and how to build frontend by using HTML, CSS, and JS.

What's Next for Ocular Disease Prediction

I'm going to build the final model, and due to the dataset, some of the disease's distinguishing still have space for improvement.

Built With

• css
• flask
• html
• javascript
• python
• tensorflow