| Author: | Zhang, Qichang |
| Title: | A deep learning-based model for human non-invasive vital sign signal monitoring with optical fiber sensor |
| Advisors: | Yu, Changyuan (EEE) |
| Degree: | M.Sc. |
| Year: | 2023 |
| Department: | Department of Electrical and Electronic Engineering |
| Pages: | vi, 42 pages : color illustrations |
| Language: | English |
| Abstract: | This study introduces a non-contact monitoring system employing a micro-sensor and deep learning methodologies for precise vital sign measurements. Existing detection techniques in healthcare facilities often necessitate direct contact, which is expensive and restricts patient movement. The suggested system mitigates these constraints through a non-invasive strategy. The system integrates a Ballistocardiogram (BCG) monitor, a Mach-Zehnder interferometer (MZI), a phase shifter, and a proportional-integral-derivative (PID) controller. The BCG signal is transformed into an electrical signal, processed via a low-pass filter and PID controller. The study employs deep learning techniques to enhance vital sign measurement precision, specifically Long Short-Term Memory (LSTM) networks. LSTM, a recurrent neural network (RNN) variant, effectively captures long-term dependencies in time-series data. The proposed model merges LSTM with Empirical Mode Decomposition (EMD), an adaptive signal processing technique, to extract valuable features from raw signals. To tackle mode mixing in EMD, an enhanced algorithm, Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), is utilized. The model's performance is assessed using a BCG signal dataset from 20 volunteers and compared with reference models such as Support Vector Regression (SVR), Extreme Learning Machine (ELM), Backpropagation Neural Network (BPNN), and traditional EMD and EEMD algorithms. Experimental results indicate that the proposed model surpasses reference models in curve fitting to the ground truth and heart rate variability (HRV) measurement accuracy. The non-contact monitoring system offers a viable alternative to traditional contact-based methods. Integrating micro-bend fiber sensors, the MZI-based BCG monitor, and deep learning techniques improves vital sign measurement accuracy, reduces resource costs, and minimizes patient discomfort. The proposed model has considerable potential for various medical diagnostic applications and interdisciplinary collaboration with the electroencephalography (EEG) and human ergonomics fields. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 8287.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 968.99 kB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13880