Medical Image Classification Using Deep Learning for Brain Tumors Detection: An Overview

Image analysis, including segmentation, detection and classification, has recently gained remarkable attention using deep learning. Image classification plays a crucial role in computer-assisted diagnosis (CAD), but it poses a formidable challenge in image analysis. Thus, deep learning can customize treatment plans based on patient data, genetics, and medical history. It excels at accurately detecting and classifying abnormalities in X-rays, MRIs, and CT scans, which facilitates early disease diagnosis by specialists. Therefore, this paper firstly introduces an overview of image classification techniques for diagnosing a variety of human maladies and deep learning methods, such as autoencoders, ensemble learning, and transfer learning. Furthermore, the Convolutional Neural Network (CNN) to detect brain tumors in MRI images is also investigated to illustrate the profound effect of CNN in image classification. Experimental cases are conducted using the VGG-16 model for two datasets of different sizes. The results show that large dataset can achieve an accuracy of 99% at epoch 5 without data augmentation.