Arrhythmia Detection and Classification on Cardiac Sensed Signals Using Deep Learning Techniques

Introduction

Cardiologists use electrocardiography (ECG) as a core diagnostic tool because it provides invaluable information about the electrical activity of the heart. The ECG records the heart's rhythmic patterns and offers crucial data for determining cardiac health and making a variety of diagnoses. Analysis of ECG plays a critical role in the assessment of cardiovascular health in contemporary medical diagnosis. Through the use of an ECG, the electrical activity of the heart may be observed, and this information can be used to diagnose, treat, and prevent numerous cardiac diseases early on. Arrhythmia is a type of disease where the abnormal rate and rhythm of ECG signals are seen. Arrhythmia detection and classification are particularly important because of their potential to signal life-threatening circumstances and direct necessary actions. The significance of arrhythmia detection through ECG analysis is aligned with key contexts like Early Diagnosis and Risk Stratification, Guiding Treatment Decisions, Monitoring Treatment, Preventing Sudden Cardiac Death, Personalized Medicine, and Long-term Monitoring. Medical treatment in a critical situation needs immediate response from medical experts and guidance for saving lives. As the availability of medical experts in the population is also very low, in this situation computer-aided diagnosis system can help society. CADS is a combine system that may include hardware and software systems at the end. In the advanced software era, machine learning and deep learning help systems to be intelligent enough to make a decision as per previous learnings. Deep learning can support the medical field in early diagnosis, detection recommendations, etc. in this field.

This section contains information about heart disease and its attributes through which heart disease can be diagnosed from ECG signals. ECG signals sensed through an electrical conduction system have a morphology using which the status of the heart can be observed. Automation in the area of the healthcare sector inspires machine learning / deep-learning models to work for the identification of heart-related diseases.

Cardiac Conduction

In the central heating system analogy, the pump, pipes, and radiators are useless without being connected to a power source. To function, the pump requires electricity. Electricity is also used by the human heart, which has a comparable need for a power supply. Thankfully, we don't need to plug ourselves into the wall because the heart can generate its electrical impulses and regulate the path the impulses go using a specialized conduction pathway. The cardiac conduction system has 5 elements shown in Figure 1.

• •

  Sino-atrial (SA) node: Natural Peacemaker

• •

  Atrio-ventricular (AV) node: This periodically discharges electrical stimulation; the frequency is determined by the body's need.

• •

  Bundle of His:

• •

  Left and right bundle branches:

• •

  Purkinje fibers:

Figure 1.

Cardiac conduction system

There are three heart stages in a single heartbeat

• •

  Atrial depolarization

• •

  Ventricular depolarization

• •

  Atrial and ventricular repolarization.

The electrical system is attached to the human body at the time of ECG capturing, as shown in Figure 2. ECG signal plotted on the graph has two types of information i) measured time interval of heart conduction system ii) measured the strength of the electrical wave. Time interval helps to find whether the heart activity is regular or irregular, it is slow or fast and strength gives an idea about the workload on the heart through contraction of the Arial muscles. The structure of the normal ECG signal and the meaningful segment is shown in Figure 3. Three important segments exist in ECG signal structure atrial depolarization (P- wave), ventral depolarization (QRS complex wave), and repolarization (T-wave). Any kind of disorder generated from heart conduction affects the ECG, known as Arrhythmia. Different types of same are mentioned below.

Figure 2.

Heart electric conduction and signal leads

Figure 3.

ECG waveform (three segments)