Certain investigations on medical image privacy preserving techniques and analysis of neural network based image classification models in cloud framework

Abstract

The evolution of medical technologies like Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and others have created massive amounts of data and is high-dimensional and variable-rich. A large amount of data is being generated in the healthcare because of these improvements in medical technology. Hence, medical image data and their dimensionalities are rapidly expanding. Because of these expansions in medical data, manually managing the file system is becoming increasingly challenging. Therefore, the management of medical data has become a major concern for healthcare practitioners. In the medical field, cloud computing is commonly utilised for storing, computing, and exchanging patient medical records. The hospital merely needs to collect patient information from files and upload the data to the cloud for storage via cloud computing. Cloud computing provides users with flexible and scalable computer resources from remote places, which they can access based on their needs. Hence, the storage of long-term medical image data in cloud is effective in addressing a variety of medical issues. Private Cloud is a computing platform where hospitals with multiple branches can store, access, compute and exchange their patients records as a single business entity. Establishing this kind of cloud framework has multiple advantages like security, accessibility, tracking and easy diagnosis besides preventing the patients carrying their records with them. Every individual patient s data is stored in the Cloud that can be accessible across their business units that are widespread across the World. Medical data in the cloud platform must be protected while being stored so that the cloud does not learn anything about it. Therefore, it is crucial to keep the sensitive contents of medical images safe during the reconstruction process. Private cloud has the greater advantage of deploying the security level of the data and accessibility based on the organisational requirement. In this research, the initial work focused on set