Inspiration
We had never worked in depth with AI before, and wanted to try out a computer vision related project. After brainstorming ideas that could have practical applications, we landed on an MRI image classification project, designed to aid hospitals in viewing and detecting abnormalities in their patients' scans.
What it does
This project takes in MRI scans of the brain, knee, or lungs, and then detects the presence of brain tumors, arthritis, and illness in the lungs.
How we built it
We built a full stack application to tackle this problem. The frontend was built using React, and the backend uses tensorflow and keras for the computer vision implementation, as well as Flask for handling requests. We used multiple online datasets to train our models, which are listed below. https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia https://www.kaggle.com/datasets/farjanakabirsamanta/osteoarthritis-prediction https://www.kaggle.com/datasets/navoneel/brain-mri-images-for-brain-tumor-detection
Challenges we ran into
As most of us were unfamiliar with the tech stack used to create this process, our largest challenges revolved around learning the new modules and frameworks used to make this project succeed.
Accomplishments that we're proud of
We are proud of learning new frameworks, especially tensorflow, which none of us had used before. After rigorously adjusting our models, we are very proud to have created accurate representations of the MRI scans passed through them.
What we learned
We learned how to use an unfamiliar tech stack, as well as computer vision concepts using tensorflow and opencv.
What's next for you?
We would like to continue expanding the project to use even more accurate models, as well as process a larger amount of diseases.
Log in or sign up for Devpost to join the conversation.