An Optimal Neural Network Classifier for Cervical Pap Smear Data

Abstract

Neural Networks are one of the widely used Soft Computing Techniques. Neural Networks are adaptive and learn from past examples. Neural Networks are used successfully in extensive range of applications related to different areas particularly in Medical Domain. Neural Networks mimic human brain to solve problems concern to non-linear and complex data such as clinical samples. Cervical Cancer is a silent cancer which does not disclose any pain and symptoms. But it becomes dangerous silently with in a long period of 10-15 years. Hence early diagnosis is an essential action to prevent it in early stages. In this study most commonly used Neural Networks such as Multi Layer Perceptron (MLP), Probabilistic Neural Network (PNN), Radial Basic Function (RBF) and Linear Vector Quantization (LVQ) networks are used. Dimensionally reduced Cervical Pap smear Dataset using Fuzzy Edge Detection method is considered for classification. The Four Neural Networks are compared and the best suitable network to classify the dataset is evaluated.