Cognosante NIH Challenge

Inspiration

This seemed like an interesting challenge to solve that involved machine learning.

What it does

We analyze multi-layer ome-tiff medical images and use machine learning to analyze those images. We can then use information from that analysis to provide visualizations. In theory, this would all be done upon user request by uploading a file.

How we built it

We built a react app that is hosted in Cloudfront. It communicates with a python api. The api uses data stored in an S3 bucket that is generated using machine learning.

Challenges we ran into

The two main problems we ran into were domain knowledge and being able to automate the use of SageMaker to apply an algorithm to the input data.

Accomplishments that we're proud of

We switched to a python api stack even though we usually use javascript

What we learned

We learned a lot about the problem domain

What's next for Cognosante NIH Challenge

Built With

• cell-profiler
• cloudfront
• cognito
• javascript
• k-means
• lambda
• leaflet.js
• pca
• python
• sagemaker