Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning for identifying systolic complexes in SCG traces: a cross-dataset analysis

Michele Craighero, Sarah Solbiati, Federica Mozzini, Enrico Caiani, Giacomo Boracchi

TL;DR

This work tackles ECG-free identification of the SCG systolic complex, a key step for analyzing cardiac mechanics outside clinical ECG setups. It employs a U-Net-based segmentation model trained on three heterogeneous datasets and evaluated under single- and multi-channel input regimes, including cross-dataset scenarios and real-world data collected during sleep. The study demonstrates a clear domain shift between controlled and real-world SCG signals and shows that personalization or fine-tuning, as well as multi-source training, improve generalization. The results underscore the practical viability of ECG-free SCG analysis while highlighting the need for adaptation to real-world conditions and sensor fusion to enhance robustness.

Abstract

The seismocardiographic signal is a promising alternative to the traditional ECG in the analysis of the cardiac activity. In particular, the systolic complex is known to be the most informative part of the seismocardiogram, thus requiring further analysis. State-of-art solutions to detect the systolic complex are based on Deep Learning models, which have been proven effective in pioneering studies. However, these solutions have only been tested in a controlled scenario considering only clean signals acquired from users maintained still in supine position. On top of that, all these studies consider data coming from a single dataset, ignoring the benefits and challenges related to a cross-dataset scenario. In this work, a cross-dataset experimental analysis was performed considering also data from a real-world scenario. Our findings prove the effectiveness of a deep learning solution, while showing the importance of a personalization step to contrast the domain shift, namely a change in data distribution between training and testing data. Finally, we demonstrate the benefits of a multi-channels approach, leveraging the information extracted from both accelerometers and gyroscopes data.

Deep Learning for identifying systolic complexes in SCG traces: a cross-dataset analysis

TL;DR

This work tackles ECG-free identification of the SCG systolic complex, a key step for analyzing cardiac mechanics outside clinical ECG setups. It employs a U-Net-based segmentation model trained on three heterogeneous datasets and evaluated under single- and multi-channel input regimes, including cross-dataset scenarios and real-world data collected during sleep. The study demonstrates a clear domain shift between controlled and real-world SCG signals and shows that personalization or fine-tuning, as well as multi-source training, improve generalization. The results underscore the practical viability of ECG-free SCG analysis while highlighting the need for adaptation to real-world conditions and sensor fusion to enhance robustness.

Abstract

The seismocardiographic signal is a promising alternative to the traditional ECG in the analysis of the cardiac activity. In particular, the systolic complex is known to be the most informative part of the seismocardiogram, thus requiring further analysis. State-of-art solutions to detect the systolic complex are based on Deep Learning models, which have been proven effective in pioneering studies. However, these solutions have only been tested in a controlled scenario considering only clean signals acquired from users maintained still in supine position. On top of that, all these studies consider data coming from a single dataset, ignoring the benefits and challenges related to a cross-dataset scenario. In this work, a cross-dataset experimental analysis was performed considering also data from a real-world scenario. Our findings prove the effectiveness of a deep learning solution, while showing the importance of a personalization step to contrast the domain shift, namely a change in data distribution between training and testing data. Finally, we demonstrate the benefits of a multi-channels approach, leveraging the information extracted from both accelerometers and gyroscopes data.
Paper Structure (10 sections, 1 equation, 2 figures, 4 tables)

This paper contains 10 sections, 1 equation, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Datasets
  4. SCG Systolic Identification algorithm
  5. Results and Discussion
  6. Single-dataset experiments
  7. Personalization
  8. Cross-dataset experiments
  9. Fine-tuning and Model Personalization
  10. Conclusions

Figures (2)

  • Figure 1: AO fiducial points and systolic complexes in a clean portion of SCG signal. Note that the bounding boxes are centered in the AO points.
  • Figure 2: Example of Input (SCG signal), Output (probability prediction) and Ground Truth (Systolic complex) for our DL model. The SCG portion reported here is from BioPoli dataset, thus recorded in uncontrolled conditions.