sandyrides / BCR-ML

A CNN model that detects breast cancer with 85% accuracy, and classifies mammograms to different types of abnormalities.

Breast-Cancer-Recognition-ML

• Sequential Neural Network model that classifies mammograms as normal (no breast cancer) or having a specific type of breast cancer.

• The model was implemented using TensorFlow and Keras on top of Python.

Implementation Notes:

1. Processing:
   • Reading PGM files
   • Reading CSV files
   • Image pre-processing
2. Model Architecture:

• Sequential Layers:
  • 4 Convolutions
  • 3 Max Pooling
  • Dropout
  • 3 Dense layers
• Layer specifications:
  • ReLU / Softmax Activations
  • Crossentropy loss
  • RMSProp Optimizer

3. Making predictions:
   • Displaying mammograms
   • Labeling them