Comparative Study of the Different Types of Electroconductive Textile Integrated ECG Sensors

1. Methodology

For this investigation, stainless steel electrodes and silver + copper platted polyester textile electrodes were produced by ____ processing a size of ___by _____. The performance of the fabricated textile ECG electrodes was analyzed using the extracted ECG signal feature variations from signal processing techniques, and the results were compared with those of commercially available silver-silver chloride electrodes. Figure 1 shows the proposed method.

1.1. ECG Signal Acquisition, Preprocessing and Denoising

The ECG signals were collected from five subjects using textile electrodes and conventional disposable silver-silver chloride electrodes simultaneously in the lead II conventional sitting position. The ECG signals were recorded from everyone at rest, during treadmill exercise, or during rest using washed textile electrodes. The duration of each signal is 1 minute, and the sampling frequency is 1000 Hz.

Figure 2. Images from the data recording procedures.

Figure 2. Images from the data recording procedures.

Respiratory muscle movement, electrode motion artifacts, EMG signals, power lines, and electromagnetic interference noise are expected to accompany the raw obtained signal. Due to its advantage in preserving the most significant features, the Savitzky Golay filter is selected for smoothing and baseline wander noise filtering [11,12]. The recorded signals were annotated and then soothed using a Savitzky Golay filter to remove irregularities and low-frequency noises. Wavelet-based denoising is the best and most powerful technique for filtering ECG signals [12,13,14]. The performances of different denoising techniques evaluated using the signal-to-noise ratio (SNR) and wavelet multiresolution analysis (WMRA) with the discrete mayor (dmey) mother wavelet were better. To separate the low-frequency components less than 1 Hz that are assumed to be baseline wandering noise, decomposition level 9 is the best. The low-frequency component hoses are outside the ECG frequency range, and the higher-frequency components contribute insignificant energy to the signal excluded during reconstruction, as shown in equation 1.

$$Denoisedsignal=Level9+Level8+Level7+Level6+Level5\dots \dots \dots$$

(1)

1.2. Feature Extraction and Analysis

In this study, Pan Tompkins QRS detection, peak detection and interval calculation algorithms were used to extract ECG features. Pan-Tompkins is the most common QRS detection method and consists of a bandpass filter, derivative filter squaring, thresholding and moving window integration algorithms sequentially [1,15,16]. Using local maximum and minimum concepts, the amplitude and location of the P, Q, R, S, and T waves were obtained. Intervals, isoelectric lines, and segments were calculated using peak locations and threshold values. Finally, the obtained ECG features from different subjects were averaged for each electrode category for better manipulation of the data and for the variations observed.

2. Results

2.1. Data Acquisition, Preprocessing and Denoising Results

Figure 3.2. Functional electrode and the outputs of smoothing and denoising processes.

Figure 3.2. Functional electrode and the outputs of smoothing and denoising processes.

Figure. ECG signals recorded by functional electrodes (a), polyester textile electrodes (b) and stainless-steel textile electrodes (c) during treadmill exercise.

Figure. ECG signals recorded by functional electrodes (a), polyester textile electrodes (b) and stainless-steel textile electrodes (c) during treadmill exercise.

Figure. ECG signal recorded by a functional electrode under static conditions.

Figure. ECG signal recorded by a functional electrode under static conditions.

Figure. ECG signals recorded by a Polyester textile electrode before washing (a) and after repeated washing (b) under static conditions.

Figure. ECG signals recorded by a Polyester textile electrode before washing (a) and after repeated washing (b) under static conditions.

Figure. ECG signals recorded by a Polyester textile electrode before washing (a) and after repeated washing (b) under static conditions.

Figure. ECG signals recorded by a Polyester textile electrode before washing (a) and after repeated washing (b) under static conditions.

2.2. Feature Extraction and Analysis Results

The R, P, and T peaks were extracted from all the collected signals. The averaged extracted features in each electrode category are shown in Table 3.1.

Table 3.1. The averaged ECG feature values of the selected electrodes under different conditions.

ECG feature values at static condition
Feature	Functional Electrode	Polyester Electrode	Stainless steel Electrode
R peak	0.6623828	0.59242381	0.56699506
P peak	0.08922754	0.0848366	0.08950509
T peak	0.182598	0.13889651	0.1462053
ECG feature values at dynamic condition
R peak	0.46834	0.42212	0.541078
P peak	0.031635	-0.00447	0.098881
T peak	0.239087	0.216838	0.167139
ECG feature values using washed textile electrodes at static condition
R peak	0.922986	0.575616	0.587001
P peak	0.06638	0.040977	0.048649
T peak	0.165789	0.083446	0.081361

Table 3.1. The averaged ECG feature values of the selected electrodes under different conditions.

ECG feature values at static condition
Feature	Functional Electrode	Polyester Electrode	Stainless steel Electrode
R peak	0.6623828	0.59242381	0.56699506
P peak	0.08922754	0.0848366	0.08950509
T peak	0.182598	0.13889651	0.1462053
ECG feature values at dynamic condition
R peak	0.46834	0.42212	0.541078
P peak	0.031635	-0.00447	0.098881
T peak	0.239087	0.216838	0.167139
ECG feature values using washed textile electrodes at static condition
R peak	0.922986	0.575616	0.587001
P peak	0.06638	0.040977	0.048649
T peak	0.165789	0.083446	0.081361

Figure 4 Extracted ECG peak values under different conditions using different electrodes

3. Discussion

For long-term monitoring, textile electrodes are comfortable [1]. To evaluate the performance of the textile-based ECG electrodes, signal processing algorithms were used. The signals were recorded using textile and conventional electrodes simultaneously and denoised, and features were extracted. The features were averaged for ease of data manipulation.

At static condition the respective P, QRS, and T amplitudes were 0.089mV, 0.66mV, and 0.18mV for Ag\AgCl electrode; 0.085mV, 0.59mV, 0.14mV for polyester textile electrode; and 0.090mV, 0.57mV, and 0.15mV for stainless-steel textile electrode. Using textile-based conductive electrodes, average peak amplitudes of 0.14, 0.96 and 0.36 mV were reported for the P, QRS and T peaks, respectively [3,17,18,19,20].

The respective dynamic condition P, QRS, and T wave amplitude values were 0.032mV, 0.47mV, and 0.24mV for Ag\AgCl electrode; -0.004mV, 0.42mV, 0.22mV for polyester textile electrode; and 0.099mV, 0.54mV, and 0.17mV for stainless-steel textile electrode [17,21].

After washing the textile (polyester and stainless-steel) electrodes the respective static condition P, QRS, and T wave amplitudes were 0.066mV, 0.92mV, and 0.17mV for Ag\AgCl electrode; 0.041mV, 0.58mV, 0.083mV for polyester textile electrode; and 0.049mV, 0.59mV, and 0.081mV for stainless-steel textile electrode. After one cycle, wash the values of 0.17, 0.83, and 0.34 mV for the P, QRS, and T peaks, respectively [3,17,18,19,20].

4. Conclusion

To evaluate the performance of the textile electrodes, ECG signals were recorded using those electrodes and the functional electrodes. Then, the signals were denoised, and vital features were extracted. The changes observed in the extracted features. Finally, the textile electrodes are acceptable for recording ECG signals.