Breast Cancer Predictor

• 

  Cover

• 

  Malignant Tissue Scan

• 

  Benign Tissue Scan

Inspiration

Breast cancer is a major health issue in my community in Tamale, Ghana, where access to diagnostic tools is limited. I have seen firsthand how this lack of resources impacts the lives of women, with many cases going undetected until it's too late. This inspired me to create a solution that could make early detection more accessible and affordable.

What it does

The Breast Cancer Predictor is a web app that uses an AI model trained on 500 ultrasound images of breast tissue from a dataset on Kaggle. It can classify the images as either malignant or benign with a single click, to assist medical professionals in the early detection of breast cancer. This app aims to contribute to early cancer detection, particularly in areas with limited access to specialized diagnostic equipment.

How we built it

The project started with data preprocessing, where we gathered a dataset of 500 ultrasound images, each labeled as either malignant or benign. Using Python and TensorFlow, we trained a convolutional neural network (CNN) model to accurately classify the images. We then integrated the model into a web application using Streamlit, providing an easy-to-use interface where users can upload an ultrasound image and receive an immediate diagnosis.

Challenges we ran into

One of the main challenges was dealing with the limited size of our dataset. With only 500 images, ensuring the model’s accuracy and avoiding overfitting required careful tuning and validation. Additionally, integrating the model into a web app while maintaining speed and performance posed another challenge, especially considering the need for the app to function effectively in areas with limited internet connectivity.

Accomplishments that we're proud of

We’re proud to have developed an AI tool that has the potential to make a real impact on healthcare in underserved areas. Despite the challenges, we successfully built a model with a high level of accuracy, and the web app's interface is both intuitive and efficient. Most importantly, we are proud to have created something that could contribute to early cancer detection, potentially saving lives by facilitating timely intervention.

What we learned

This project taught us a lot about the complexities of building AI models with limited data, especially in the healthcare domain, where accuracy is critical. We also gained valuable experience in developing web applications that need to be both powerful and user-friendly, and in understanding the real-world impact of technology on global health challenges.

What's next for Breast Cancer Predictor

Our next steps include expanding the dataset to improve the model's accuracy further, and conducting field tests to evaluate the app's performance in real-world settings. We also plan to add features that allow for more detailed analysis and to explore partnerships with healthcare providers to bring this tool into clinics and hospitals in underserved regions. Ultimately, our goal is to continue refining and expanding the app to make breast cancer detection more accessible and to contribute significantly to early cancer detection techniques worldwide.

Built With

• python
• streamlit
• tensorflow