ABSTRACT : Breaches of data can seriously affect healthcare providers' operations and reputations, as they can delay treatment or lead to financial loss. Despite efforts to put in place measures to prevent unauthorized access to patient information, breaches are still happening in the healthcare industry. Due to the increasing number of data breaches, it has become more important that organizations implement effective security measures for their Big Data projects. Classifying the data method tends to produce higher error when compared to other methods due to the large variance directions. One of the most common types of detection technology is one-class classification. This method is very competitive in detecting fake medical images due to the data's unbalanced nature. However, it can also produce higher error when compared to other methods. One of the most effective ways to improve the accuracy of one-class classification is by implementing incremental covariance-guided support vector machine (iCOSVM) especially with a real time system which is proposed in this thesis. In this part of the system, we first generate fake images using Generative Adversarial Network (GAN) framework. After, extracting features of fake images samples using CNN-VGG-16 and perform further analysis. Then, in order to classify the images, in real time system iCOSVM is introduced. Finally a comparative analysis is performed of iCOSVM and the existing relevant incremental boundary-based methods is performed. In addition, to increase productivity, an effective unsupervised clustering technique is necessary. It serves as a crucial preprocessing recommended step prior to any supervised machine learning system. Therefore, an Improved Dynamic Autoencoder (IDynAE) model is proposed in this study too as a pre-processing technique for our system. We evaluate the impact of our proposed IDynAE model which demonstrates that our