Neural Network architecture for breast cancer detection and classification

Abstract

Early detection of cancer or any disease is among the main keys to its cure. One of the state of the art methods in cancer detection is machine learning, namely ANNs (Artificial Neural Networks). ANNs have proved to be efficient due to their ability to learn and generalize from data. This paper proposes a low-complexity architecture of an ANN that classifies breast cancer as either Benign or Malignant through pattern recognition. It focuses on finding the optimal activation function that minimizes the classification error with less number of blocks. This results in reduction of the complexity of the implementation with CMOS technology.