Application-Specific Discriminant Analysis of Cardiac Anomalies Using Shift-Invariant Wavelet Transform

1. Introduction

Biomedical signals represent the electrical dynamism of the human body. The medical experts study these signals to acquire vital information regarding the health challenges. Electrocardiogram (ECG) is a non-invasive electrical signal whose automated analysis has motivated the current medical diagnostic scenario for user applications to handle at ease. These applications deal with free ECG check-up camps, Intensive Care Unit (ICU), Emergencies, and mobile fast track devices. Despite having difficulties extracting accurate and reliable ECG information, the biological systems eager to shift towards automation and E-health programs. Many upcoming novel types of research are encouraged to resolve these issues, keeping in mind the appropriate ECG data acquisition, meticulous feature extraction, ace classifiers, computational cost, and dynamic memory management.

ECG pattern is a simple and useful tool for analyzing cardiac disturbances as it has set operations for a healthy heart. The depolarization-repolarization cycle of heart chambers shows the exact time evolution of the P-QRS-T pattern, as shown in Figure 1. R-peak is the most prominent feature which helps in recognizing the heartbeat count. Every ECG pattern is unique and different according to the human system (Mirvis and Goldberger, 2001). Any change in the morphological and temporal characteristics presents malfunctioning of heart, which leads to arrhythmia. An arrhythmia is an irregularity in the cardiac rhythm. It disturbs the workflow of other organs like the brain, lungs, etc. Foremost, the health check requires ECG testing and analysis. In this paper, the discriminant analysis of shift-invariant ECG features is done with aspired classifiers to make an automated proposal for specialized medical applications.