CAIS-Data-Science-Bowl-Demo

Summary: Takes in a numpy file containing preprocessed lung scans (each patient has 20 50x50 scan slices), and uses Matplotlib generates 2D lung scan images in real time while training a convolutional neural network in TensorFlow to predict whether or not the patient has lung cancer.

Created initially for the 2017 Data Science Bowl, and adapted into a demo for our AI club at USC (CAIS++).

Sample screenshot from demo

Dependencies

• Python 3.5 (preferably via Anaconda)
• Numpy
• Tensorflow
• Matplotlib

Instructions

1. Clone the repo locally
2. Naviagate to the repo folder
3. Run: python dsb_train_demo.py
4. Sit back and enjoy!

Acknowledgements

• Sentdex, for his CNN-based data science bowl kernel on Kaggle