Colors of Cancer

Inspiration

The inspiration for Colors of Cancer came from the realization of bias in skin cancer detection against people of color (POC). Studies have shown that POC is often misdiagnosed or diagnosed at later stages of skin cancer due to the limited representation of their skin types in the available datasets used for training computer vision models. This inspired the team to create an application that can help improve the early detection of skin cancer for people of all skin types.

What it does

Colors of Cancer is an application that connects with a camera, uses computer vision, and detects both benign and cancerous skin diseases. The user takes a picture of their skin and the application analyzes the image using an ML model trained on a diverse dataset of skin types. The application then provides a prediction on whether the lesion is benign or cancerous and recommends the next steps, including seeking professional medical advice.

How we built it

To build Colors of Cancer, the team first developed an ML model in Google Colab using TensorFlow and Keras. The model was trained on a dataset of skin images of people with various skin types.

Made onFlutter, a mobile app development framework, to make it accessible on Android devices.

Challenges we ran into

The team faced several challenges during the development of Colors of Cancer. The first challenge was downloading the necessary software, which took longer than expected due. Finding the right dataset was also challenging and connecting the application's API to the ML model is something we're currently working on

Accomplishments that we're proud of

The team is proud of creating a computer vision model that can detect both benign and cancerous skin diseases. This was a challenging task that required extensive research and testing.