Inspiration
With the pandemic still going strong, I noticed the workload placed on medical staff in hospitals we had decided to work on a ML medical classification model in which a Doctor or nurse can take the MRI scans of a patient and easily input it and get a result. The reasoning behind this is because now even non medical staff could "diagnose" a brain tumor with just the input of an image.
What it does
What it does is it prompts the user to input and MRI image of the patient and submit it, once the user submits it takes the user to another page with the accuracy of its guess and also the main classification in which we put into two categories, yes, and no.
How we built it
We used python and flask for the main building and training of the model. To train the model we went on Kaggle and found a dataset of MRI images containing brain tumors and one without. We also used python for the API to connect to our local host and used Simple Html and css for a decent looking submission page.
Challenges we ran into
Some challenges we ran into was of course the API, for a beginner this was a pain to figure out but once we got it to work we were all extremely relieved. Another problem we faced was the image resizing some images that we got from google to test gave a resizing error but we managed to move past it.
Accomplishments that we're proud of
Since Kaiyu is a 3rd year student in statistics, and Liu is a 2rd year in CS, with me being a freshman in CS, we were definitely not the best of the best and even has some language barriers but we researched and planned efficiently so we were able to get the product out on time. We were both proud of just coming out to the hackathon and being able to meet everyone and present this project.
What we learned
We learned that sleeping on the floor without a blanket is decently comfortable but extremely cold, but in all seriousness this whole experience was new to us, making a machine learning model, creating and implementing the API, image resizing, model training, and connecting the model to a website with a user input was super cool.
What's next for Brain tumor detection
We want to start branching out into different diseases in the brain, things like predicting Alzheimer's, Parkinson's, dementia and more classification types than yes or no.
Log in or sign up for Devpost to join the conversation.