Comparative Analysis of Medical Images using Transfer Learning Based Deep Learning Models

Abstract

Deep learning is becoming more popular in practically every industry, but especially in medical imaging for better diagnostics of various deadly diseases. Deep learning is used to explain difficulties based on medical image processing as part of machine learning artificial intelligence. Most commonly used machine learning algorithm named Convolutional Neural Network (CNN) grasps a resilient position for image recognition tasks. In this paper, we compared the performance of basic CNN and three state of the art transfer-learning models namely, VGG-16, ResNet50 and GoogleNet (Inception-v3) by extracting features from pre-trained CNN architecture. Small datasets of three fatal diseases, which are brain tumor, breast cancer and skin cancer are used. The determination of this study is to discover the finest trade-off between accuracy.