Arrhythmia Detection from PPG Signals

Introduction: These days, sudden cardiac arrest is the leading cause of death worldwide, affecting a large number of people. Cardiac arrhythmias can be diagnosed with Electrocardiogram (ECG). No one will have ECG machines in home and it also requires the electrodes to stick at appropriate places and will take time to analyze the ECG. Meanwhile, after the COVID pandemic, most of people are using finger pulse oximeter to check their vitals. Since there is a correlation between Electrocardiogram (ECG) and plethsymogram (PPG), it can be used to detect serious arrhythmias. A wearable finger pulse oximeter uses spectrophotometry and plethysmography to determine the oxygen saturation level and pulse rate. My objective is to build a wearable finger pulse oximeter with an MAX30102 pulse oximeter sensor and Arduino microcontroller that detects major cardiac arrhythmias with the help of the signal processing tool Neurokit. With the help of this project, I hope arrhythmias can be detected early and save lives with proper treatment.

The PPG wave can be used as a tool to detecting and diagnosing cardiac arrhythmias. The beat-to-beat variation in PPG amplitude may be a clue that the patient has developed an irregular rhythm. The sudden reduction in the amplitude of the pulse oximeter waveform, combined with the typical ECG pattern, should give an important warning regarding the presence of a dangerous situation.

Materials Used and Methodology: In order to get the SPO2 and Heart rate, we need a sensor to record the plethysmogram. Photo Plethysmogram is a technique in which RED(660nm) and IR(940nm) light is illuminated on the tissue. Some of the light particles (photons) are scattered, reflected, and absorbed. A photodetector detects the light after passing through the tissue. Those photons that reach the photodetector has undergone scattering and absorption. Thereby, by measuring the change in intensities of detected light, relative change within the tissue can be determined. MAX30102 monitors heartbeat per minute and plays a crucial role in measuring blood oxygen saturation (SPO2).

Arduino UNO is a microcontroller based on ATmega328p. It has 14 digital input/output pins (of which 6 can be used as PWM outputs), 6 analog inputs, a 16 MHz ceramic resonator (CSTCE16M0V53-R0), a USB connection, a power jack, an ICSP header and a reset button. It contains everything needed to support the microcontroller. It can be simply connected to a computer with a USB cable or powered with an AC-to-DC adapter or battery to get started. Arduino programs are written in the Arduino Integrated Development Environment (IDE). Arduino IDE is a special software running on the system that allows the user to write sketches.

NeuroKit2 is used as a Signal processing tool in this project. The NeuroKit2 is an open-source, community-driven, and user-centered Python package for neurophysiological signal processing. It provides a comprehensive suite of processing routines for a variety of bodily signals (e.g., ECG, PPG, EDA, EMG, RSP) Neurokit2 analyzes the signal data Frame (2-D data structure). The data containing the signals are preprocessed and analyzed by inbuilt functions.

The MAX30102 Pulse oximeter sensor, integrated with an Arduino microcontroller, is utilized to capture PPG signals. The MAX30102 sensor raw data can be viewed on a serial monitor in Arduino IDE. With a little coding, SPO2 and Heart rate can be calculated. These data were collected with timestamps and saved as CSV files and updated every 30 seconds. The data in the CSV file were then used for analysis. The data contained within the CSV file undergoes preprocessing utilizing functions from the Neurokit2 library. The PPG signal is analyzed with the PPG analyze function. PPG_clean function aids in refining the signal by eliminating noise. In order to detect the arrhythmia, the peaks of the PPG wave have to be found. PPG_peaks function helps in finding peaks. Following peak detection, the algorithm assesses the rhythmicity of the signal, triggering an alarm if any irregularities in the rhythm are detected.

Results: Various data were tested and trained in the system. The normal PPG data were collected from normal subjects. The atrial fibrillation PPG data were collected from the MIMIC PERform AF database.

On the top of the output images, the rhythm is shown as normal or irregular rhythm. The average heart rate is also calculated for the 15 seconds. The algorithm detects the rhythm and mentions Irregular rhythm in outputs figures. As you can see the output images has 2 graphs, the blue color is the clean PPG signal data and the red color is peak detected signals.

Conclusion: The future plan is to design a very compact device similar to a pulse oximeter, with inbuilt SPO2 sensor and Bluetooth module. This Bluetooth module can communicate with Mobile phones, Laptops or even Arduino. Since NeuroKit2 is open source, we can use it to analyze and interpret the plethsymogram anytime and anywhere. This project will be useful for elderly people and even in some emergency situations.