Inspiration
A few months ago, covid in India was bad. It was so bad that cancer beds had to be lent out to covid patients. We had to stand in long queues to get any kind of reports and that itself caused a lot of lives to end. It made me realize how cutting down the time of diagnosis in any way can save lives. In 10th grade, I lost a friend to cancer so I decided to cater my project to cancer as a tribute to him, even though it can be used for other kinds of scans as well. I also always wonder if a new treatment could have potentially saved his life so I added an element to help pathologists find patients for their new treatment as well.
What it does
Supercomputers coupled with machine learning can achieve high accuracy rates, almost as good as the human eye. With projects such as the distributed computer labs that are bringing the power of supercomputers to your doorstep with little to no code, we can easily run Machine Learning Algorithms on supercomputers and get highly accurate predictions. There are supercomputers already doing this. [Proof of concept]
[Technical Ability] My design currently doesn't have supercomputers as a part of its design. My web application has features to make cancer diagnosis a lot faster. All you have to do is upload a scan and it gives you a prediction if a tumor [brain and breast cancer has been added for this hackathon] was detected or not. Other than that, all the cancer clinics that use the website for diagnosis enter patient information that is stored safely in a database that can be used by pathologists to find patients for new treatments. This can benefit the patients and pathologists and change the course of people's lives. Using this application, the waiting time for diagnosis can be cut down by a lot.
How we built it
- HTML - To build the basic structure of the website.
- CSS - To control the presentation, formatting, and layout.
- javascript - To control the behavior of different elements.
- Python/Machine Learning - Python is used to build models that can accurately detect if a person has the tumor or not based on the uploaded scans.
Challenges we ran into
After building the machine learning models, increasing its accuracy was a challenge. Also completing the complete project in 24 hours and doing it by myself was a little challenging. There were also a few errors in integrating the HTML, CSS, and JS smoothly in the beginning so resolving that was challenging as well.
Accomplishments that we're proud of
I enjoyed learning about how supercomputers are on the verge of revolutionizing healthcare. Soon, we can get results for any kind of test in a matter of minutes through AI. I am also proud of finishing everything on time and creating an aesthetically pleasing website.
What we learned
- Using CNN and many other ML and deep learnings models.
- UI/UX
What's next for Cancer diagnostics.
Integrate supercomputers to the web application and integrate the backend with the frontend properly. Perhaps, pitch the idea to cancer clinics and see if they would be interested in this concept.
Log in or sign up for Devpost to join the conversation.