Robust PPG Peak Detection Using Dilated Convolutional Neural Networks

• DOI: 10.36227/techrxiv.16529310
• Publication Date: 2021-09-01T22:24:27Z
• Authors: Kianoosh Kazemi; Juho Laitala; Iman Azimi; Pasi Liljeberg; Amir M. Rahmani
• Keywords: Photoplethysmogram; SIGNAL (programming language)

<div>Accurate peak determination from noise-corrupted photoplethysmogram (PPG) signal is the basis for further analysis of physiological quantities such as heart rate and variability. In past decades, many methods have been proposed to provide reliable detection. These detection include rule-based algorithms, adaptive thresholds, processing techniques. However, they are designed noise-free PPG signals insufficient with low signal-to-noise ratio (SNR). This paper focuses on enhancing noise-resiliency proposes a robust algorithm noise motion artifact corrupted signals. Our based Convolutional Neural Networks (CNN) dilated convolutions. Using convolutions provides large receptive field, making our CNN model at time series processing. this study, we use dataset collected wearable devices in health monitoring under free-living conditions. addition, data generator developed producing noisy used training network. The method performance compared against other state-of-the-art tested SNRs ranging 0 45 dB. obtains better accuracy all SNRs, existing threshold transform-based methods. shows an overall precision, recall, F1-score 80%, 80% SNR ranges. these figures below 78%, 77%, respectively. proves be accurate detecting peaks even presence noise.</div>