Novel encryption and compression based heart disease diagnosis with deep learning using electrocardiography signals

Abstract

Electrocardiogram (ECG) monitoring models are commonly employed for diagnosing heart diseases. In last decades, cardiovascular disease become common and it results in increased mortality rate. ECG signals can be widely used in telecardiology healthcare programs and they need to be compressed before being transmitted or stored. Since the ECG data requires a massive amount of storage area owing to the continuous heart rate logs, compression techniques are applied to the ECG data prior to transmission to the healthcare centre for examination. At the same time, securing the patient s secret details are also a crucial process in several health care models. So, a novel compression technique find useful prior to transmitting it to the telemedicine center to monitor and analyze the data. Earlier works have developed models to accomplish high quality reconstructed signals with high computation complexity. In addition, the protection of ECG signals poses a challenging issue, which can be resolved by the encryption techniques. The available Encryption-Then-Compression (ETC) models for multimedia data does not properly maintain the tradeoff between compression performance and signal quality. The entire research work is organized into set of different objectives as listed below. To design an Encryption and Compression with Deep Learning-based Heart Disease Diagnosis using Electrocardiography (ECG) Signals. newlineTo develop a Multi-Objective Optimization-based Encryption then Compression with Deep Belief Network Model for ECG Signal Classification To design and implement Optimal Homomorphic Encryption Then Vector Quantization based Compression Scheme for ECG Signal Classification In the First Objective, a new ETC with a diagnosis model for ECG data, named ETC-ECG model. The proposed model involves four major processes namely preprocessing, encryption, compression, and classification. Once the ECG data about the patient is gathered, Discrete Wavelet Transform (DWT) with Thresholding mechanism is used for noise removal. In addition, the chaotic map based encryption technique is applied to encrypt the data. Moreover, the Burrows Wheeler Transform (BWT) approach is employed for the compression of the encrypted data. Finally, a Deep Neural Network (DNN) model is applied to diagnose the decrypted data to identify the existence of heart disease. In the Second Objective, a new Encryption Then Compression (ETC) scheme with an Optimal Deep Belief Network (ODBN) based ECG signal classification model, called ETC-ODBN for heart disease diagnosis newline