Paperdoi:10.1109/jsen.2025.3561604
Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks
IEEE Sensors Journal
TL;DR
Bolpagni et al. train lightweight UNet / LSTM-UNet / Atrous-UNet for PPG motion-artifact segmentation on Empatica E4 stress/PASS subsets (30 s @ 64 Hz). LSTM-UNet and Atrous-UNet report Dice 0.8194 on their test split; quantized Atrous-UNet/UNet run on EmotiBit and Bangle.js 2 with accuracy still above ~78% after INT8.
- Form factor: wrist
Abstract
The widespread adoption of wearable health monitoring devices has highlighted the need for robust algorithms to detect and mitigate motion artifacts in photoplethysmographic (PPG) signals. This article introduces a novel approach to motion artifact segmentation, supported by the creation of a new dataset. PPG signals from publicly available repositories were curated and enhanced with precise segmentation masks, enabling rigorous model training and evaluation. Three deep learning architectures—UNet, LSTM-UNet, and Atrous-UNet—were assessed using a comprehensive set of metrics, including dice score, intersection over union (IoU), and average Hausdorff distance (AHD), to highlight the model performance beyond standard evaluations. This multimetric approach underscored the importance of addressing diverse aspects such as boundary precision and overall robustness. Atrous-UNet emerged as the most effective model, offering a balance of high performance and computational efficiency suitable for real-time deployment. The models were validated on real-world data from open-source wearable devices, EmotiBit and Bangle.js 2 (accuracy above 80% for both), demonstrating generalizability across varying hardware and environmental conditions. Comparisons with expert and nonexpert human annotators revealed that the models significantly outperformed the two groups in reliability and detection consistency. The models were optimized using quantization techniques to enable deployment on resource-constrained devices. Although this introduced some performance losses, the models retained robust artifact detection capabilities with an accuracy of over 78%.
Verbatim from Semantic Scholar (opens in a new tab) · Abstract via Semantic Scholar; © the authors/publisher
Results on ChatPPG leaderboards
All leaderboardsNo results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.
Tasks & signals
Tasks
Signals
Devices
Keywords
Datasets used
How it works
Cite this paper
Bolpagni, M., Campanella, S., Gabrielli, S., & Palma, L. (2025). Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks. IEEE Sensors Journal. https://doi.org/10.1109/jsen.2025.3561604
Bolpagni, Marco, Sara Campanella, Silvia Gabrielli, and Lorenzo Palma. 2025. "Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks." IEEE Sensors Journal. https://doi.org/10.1109/jsen.2025.3561604.
Bolpagni, M., Campanella, S., Gabrielli, S. and Palma, L. (2025) 'Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks', IEEE Sensors Journal. doi: 10.1109/jsen.2025.3561604.
Bolpagni, Marco, et al. "Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks." IEEE Sensors Journal, 2025. https://doi.org/10.1109/jsen.2025.3561604.
M. Bolpagni, S. Campanella, S. Gabrielli, and L. Palma, "Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks," IEEE Sensors Journal, 2025, doi: 10.1109/jsen.2025.3561604.
@article{bolpagni2025segmentation,
title = {{Segmentation of Motion Artifacts in Wearable PPG Signals Using Lightweight Neural Networks}},
author = {Bolpagni, Marco and Campanella, Sara and Gabrielli, Silvia and Palma, Lorenzo},
year = {2025},
journal = {IEEE Sensors Journal},
doi = {10.1109/jsen.2025.3561604}
}Added to the ChatPPG research hub
