CelluScope: An app that has the potential to diagnose malaria, sickle cell disease, and more with extremely inexpensive equipment
Last weekend I attended my first hackathon - HackMIT. I took a redeye flight from LAX to BOS and arrived at MIT very sleepy with only 2 hours of sleep and a small duffle bag.
During check-in and breakfast, I found three other people for my team - Poseidon, David, and ArisKnight. We decided to create an app that would utilize a small microscope that Poseidon brought with him. The microscope cost less than $50 and once attached to the iPhone camera, could magnify images up to 200x-300x. After researching what we could see at that scale, we decided to focus on red blood cells. One of the sponsors at the hackathon was Clarifai. They had just finished a new API and released it for the attendees of HackMIT. With their new API, you could now customize training for Clarifai. Now, we could feed any input to Clarifai and train it to recognize different things. We trained it using images from Google Images of malaria infected red blood cells, sickle cell anemia affected red blood cells, and healthy red blood cells. Soon enough, it was able to do so with surprising accuracy. Even though we were able to only train Clarifai with less than 20 images for each classification, it was still diagnosing test pictures accurately.
We created CelluScope. CelluScope was an iOS app built with Objective.C able to identify malaria infected red blood cells, sickle cell anemia, and healthy red blood cells. The idea is that you can use this app without being an expert on what you're taking pictures of. Although we chose to focus on blood cells, you could use this app to discover what microorganisms are lurking in your neighborhood pond. We think this app would be perfect for doctors working in the field, perhaps at a clinic. Not only do they not need to carry a bulky microscope, but they can identify samples on the spot, save the pictures, and have the app help them identify samples.
The greatest challenge in creating CelluScope was creating a workaround for the fact that Clarifai's new API had not been optimized for iOS. We did this using Parse and javascript.
Out of over 250+ teams and 1000 hackers, our app made it to Top 10 and as a result, we got to demo our app at the closing ceremony! :)
Our app being used with the attachable microscope
Here is a backlight, magnet and a slide with a plant stem sample.
Here is what that plant stem cell looks like using our app and microscope.
We bled for this project. This is the blood of one of our teammates underneath the app's microscope.