Secure steganography, compression and diagnoses of electrocardiograms in wireless body sensor networks

Ibaida, A 2014, Secure steganography, compression and diagnoses of electrocardiograms in wireless body sensor networks, Doctor of Philosophy (PhD), Computer Science and Information Technology, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Ibaida.pdf Thesis application/pdf 2.12MB
Title Secure steganography, compression and diagnoses of electrocardiograms in wireless body sensor networks
Author(s) Ibaida, A
Year 2014
Abstract Submission of this completed form results in your thesis/project being lodged online at the RMIT Research Repository. Further information about the RMIT Research Repository is available at Please complete abstract and keywords below for cataloguing and indexing your thesis/project. Abstract (Minimum 200 words, maximum 500 words) The usage of e-health applications is increasing in the modern era. Remote cardiac patients monitoring application is an important example of these e-health applications. Diagnosing cardiac disease in time is of crucial importance to save many patients lives. More than 3.5 million Australians suffer from long-term cardiac diseases. Therefore, in an ideal situation, a continuous cardiac monitoring system should be provided for this large number of patients. However, health-care providers lack the technology required to achieve this objective. Cloud services can be utilized to fill the technology gap for health-care providers. However, three main problems prevent health-care providers from using cloud services. Privacy, performance and accuracy of diagnoses. In this thesis we are addressing these three problems. To provide strong privacy protection services, two steganography techniques are proposed. Both techniques could achieve promising results in terms of security and distortion measurement. The differences between original and resultant watermarked ECG signals were less then 1%.

Accordingly, the resultant ECG signal can be still used for diagnoses purposes, and only authorized persons who have the required security information, can extract the hidden secret data in the ECG signal. Consequently, to solve the performance problem of storing huge amount of data concerning ECG into the cloud, two types of compression techniques are introduced: Fractal based lossy compression technique and Gaussian based lossless compression technique. This thesis proves that, fractal models can be efficiently used in ECG lossy compression. Moreover, the proposed fractal technique is a multi-processing ready technique that is suitable to be implemented inside a cloud to make use of its multi processing capability. A high compression ratio could be achieved with low distortion effects. The Gaussian lossless compression technique is proposed to provide a high compression ratio. Moreover, because the compressed files are stored in the cloud, its services should be able to provide automatic diagnosis capability. Therefore, cloud services should be able to diagnose compressed ECG files without undergoing a decompression stage to reduce additional processing overhead. Accordingly, the proposed Gaussian compression provides the ability to diagnose the resultant compressed file. Subsequently, to make use of this homomorphic feature of the proposed Gaussian compression algorithm, in this thesis we have introduced a new diagnoses technique that can be used to detect life-threatening cardiac diseases such as Ventricular Tachycardia and Ventricular Fibrillation. The proposed technique is applied directly to the compressed ECG files without going through the decompression stage. The proposed technique could achieve high accuracy results near to 100% for detecting Ventricular Arrhythmia and 96% for detecting Left Bundle Branch Block. Finally, we believe that in this thesis, the first steps towards encouraging health-care providers to use cloud services have been taken. However, this journey is still long.

Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Computer Science and Information Technology
Keyword(s) ECG
Version Filter Type
Access Statistics: 398 Abstract Views, 756 File Downloads  -  Detailed Statistics
Created: Fri, 01 Aug 2014, 11:19:27 EST by Denise Paciocco
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us