Deep Learning for Size and Microscope Feature Extraction and Classification in Oral Cancer: Enhanced Convolution Neural Network
Background and Aim: Over-fitting issue has been the reason behind deep learning technology not being successfully implemented in oral cancer images classification. The aims of this research were reducing overfitting for accurately producing the required dimension reduction feature map through Deep Learning algorithm using Convolutional Neural Network. Methodology: The proposed system consists of Enhanced Convolutional Neural Network that uses an autoencoder technique to increase the efficiency of the feature extraction process and compresses information. In this technique, unpooling and deconvolution is done to generate the input data to minimize the difference between input and output data. Moreover, it extracts characteristic features from the input data set to regenerate input data from those features by learning a network to reduce overfitting. Results: Different accuracy and processing time value is achieved while using different sample image group of Confocal Laser Endomicroscopy (CLE) images. The results showed that the proposed solution is better than the current system. Moreover, the proposed system has improved the classification accuracy by 5~ 5.5% on average and reduced the average processing time by 20 ~ 30 milliseconds. Conclusion: The proposed system focuses on the accurate classification of oral cancer cells of different anatomical locations from the CLE images. Finally, this study enhances the accuracy and processing time using the autoencoder method that solves the overfitting problem.