Improving the Accuracy in Classification of Blood Pressure from Photoplethysmography Using Continuous Wavelet Transform and Deep Learning

Abstract

Background. Continuous wavelet transform (CWT) based scalogram can be used for photoplethysmography (PPG) signal transformation to classify blood pressure (BP) with deep learning. We aimed to investigate the determinants that can improve the accuracy of BP classification based on PPG and deep learning and establish a better algorithm for the prediction. Methods. The dataset from PhysioNet was accessed to extract raw PPG signals for testing and its corresponding BPs as category labels. The BP category of normal or abnormal followed the criteria of the 2017 American College of Cardiology/American Heart Association (ACC/AHA) Hypertension Guidelines. The PPG signals were transformed into 224 ∗ 224 ∗ 3-pixel scalogram via different CWTs and segment units. All of them are fed into different convolutional neural networks (CNN) for training and validation. The receiver-operating characteristic and loss and accuracy curves were used to evaluate and compare the performance of different methods. Results. Both wavelet type and segment length could affect the accuracy, and Cgau1 wavelet and segment-300 revealed the best performance (accuracy 90%) without obvious overfitting. This method performed better than previously reported MATLAB Morse wavelet transformed scalogram on both of our proposed CNN and CNN-GoogLeNet. Conclusions. We have established a new algorithm with high accuracy to predict BP classification from PPG via matching of CWT type and segment length, which is a promising solution for rapid prediction of BP classification from real-time processing of PPG signal on a wearable device.

1D Convolutional Autoencoder-Based PPG and GSR Signals for Real-Time Emotion Classification

“…Therefore, when comparing the results excluding the model calculation time, the EEG signal required the most time at approximately 64.91s. The existing scalogram (PPG) method [47] is approximately 3.44 s faster than the power density feature extraction (EEG) method [46] but slower than the proposed method. In addition, the proposed method has twice the sample size as the scalogram feature extraction method, but requires a short time of approximately 1.27 s. The accuracy of emotion recognition and classification of the proposed method was 80.89% in the arousal domain, which was the highest among the three methods, and 81,25% in the valence domain, which was the second highest among the three methods.…”

“…In addition, time, frequency, and time-frequency domain characteristics were extracted from the data in the same way as the existing research method [27,46]. For the 1-channel PPG signal, a scalogram that could check time change according to the frequency band was used [47].…”

“…To compare the performance of previous studies [46,47], feature extraction from the EEG signals acquired from 12 channels was performed in the time domain for changes over time and the frequency domain to understand the spatial resolution of the EEG signal. Four regions were divided using the PSD characteristics: slow alpha (8-10 Hz), alpha (8-13 Hz), beta (13-29 Hz), and gamma .…”

A Data-Driven Model with Feedback Calibration Embedded Blood Pressure Estimator Using Reflective Photoplethysmography

“…Many studies adopted features, such as Sensors 2022, 22, 1873 3 of 26 temporal domain characteristics, including systolic time, diastolic time, cardiac period, and pulse width, by detecting various feature points of PPG signals for BP estimation [25,26]. Other PPG studies used frequency domain characteristics that contained valuable healthrelated information to estimate BP values based on Fast Fourier Transform and generalized transfer function [27,28]. In these studies of temporal and frequency domains, different machine learning (ML) algorithms were employed for BP estimation, such as regression algorithms [29], artificial neural networks (ANN) [30], fuzzy logic [31], and support vector machine [32].…”

Classification of Blood Pressure Levels Based on Photoplethysmogram and Electrocardiogram Signals with a Concatenated Convolutional Neural Network

“…In recent studies, several researchers used machine learning with PPG features to automatically predict or categorize blood pressure levels. Wu et al proposed continuous wavelet transforms to transform PPG signals into 224 × 224 × 3 scalogram images [ 31 ]. The scalogram images of the PPG signals are be used as input to the 2D CNN, which consists of two convolutional layers followed by pooling layers in the feature extraction layers and two hidden layers in the classification layers.…”