Genetic K-Means clustering algorithm for achieving security in medical image processing over cloud

Abstract

In healthcare domain, there is persistent pressure to improve clinical outcomes while lowering costs. In this respect, healthcare organizations can leverage cloud computing resources to avoid building an expensive in-house data center. More specifically, this new trend offers the opportunity to rent the use of imaging tools in order to process medical records. Additionally, cloud billing is based on a pay-per-use model to achieve cost savings. However, security and privacy concerns are the main disadvantages of cloud-based applications, especially when it comes to managing patients’ data. The commonly used techniques for protecting data are homomorphic algorithms, Service-Oriented Architecture (SOA) and Secret Share Scheme (SSS). These traditional approaches have some limitations that provide a boundary to its use in practice. Precisely, the implementation of these security measures in cloud environment does not have the ability to maintain a good balance between security and efficiency. From this perspective, we propose a hybrid method combining a genetic algorithm (GA) and K-Means clustering technique to meet privacy and performance requirements. This approach relies on distributed data processing (DDP) to process health records over multiple systems. Consequently, the proposal is designed to help protect clients’ data against accidental disclosure as well as accelerating the computations.