HackStack

Project Info

BrainTumor-Segmentation_Model

Devpost

This project did not submit a demo video on Devpost.

Inspiration

As a group, we came together with an idea that we wanted to make a product that could potentially have genuine effects in the real world. After researching various different ai model topics through databases such as Kaggle, we came across the topic of image segmentation specifically corresponding with brain tumors. This idea of autonomous image segmentation sounded like an incredibly interesting project to explore as an AI model for this would be able to minimize the labor-intensive process of instead doing manual segmentation. Besides the aspect of labor, we also believe the model we’ve created could yield other benefits from an educational and even emotional aspect which will be discussed later.

What it does

The goal of the AI model is to simply perform the process of brain tumor segmentation as accurately as possible which would then be connected to a website to help match people.

How we built it

This project was built using datasets provided by Kaggle in order to perform the process of segmentation. The reflex website was built using the reflex workshop's given information.

Challenges we ran into

Some challenges we ran into for this project were that we had trouble finding larger datasets in order to train the model efficiently. Along with that, understanding the syntax for the reflex website was also a struggle for us as this was brand new.

Accomplishments we're proud of

Accomplishments that we're incredibly proud of is that we were able to get the website up and running along with the fact that our model achieved a 0.8509 dice score which is very high for a model we built from scratch.

What we learned

From this project, we learned a great deal about the efforts that go into training an AI model, brief information about the frontend and backend of fullstack development, and even what it was like to work in a group to build all of this effectively.

What's next

I believe the next step for this model would be to increase the model's prediction by providing the model with a larger dataset. Along with this, I believe incorporating the AI aspect in to the website would also be something to work on in the future.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • PythonIn code

1 of 1 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

38 KB

Source files

28

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars