Bachelor and Master Theses

Background:

Biomedical signal processing is a critical component in modern healthcare, extracting valuable insights from physiological signals like the Photoplethysmogram (PPG). PPG, a non-invasive optical technique, provides information about blood volume changes and monitors various aspects of health, including cardiovascular conditions. Its potential lies in early detection and continuous monitoring of diseases.

AI and machine learning have gained prominence in healthcare for revolutionizing disease diagnosis, predictive modeling, and risk assessment. These technologies enhance the accuracy and efficiency of cardiovascular disease diagnosis by analyzing PPG data. Real-time data processing is vital in healthcare, allowing immediate responses to changes in patient conditions. In cardiovascular health, real-time analysis enables early anomaly detection, timely medical interventions, and potentially life-saving measures, reducing healthcare costs.

Embedded systems are compact, cost-effective solutions for real-time data acquisition and processing. They serve as a vital link between the SpO2 finger clip device and the AI-based predictive modeling system, making them integral to this project.

Motivation :

The primary motivation driving this project is the development of a real-time system for signal processing algorithms deployed on embedded systems, specifically for healthcare monitoring. This endeavor seeks to create a platform that can provide immediate insights into signals, enabling timely interventions and continuous monitoring. Real-time analysis of physiological signals, such as PPG data, promises to enhance patient care and ensure informed decision-making in healthcare settings.

Project Overview:

This project involves using a PPG simulator device to develop a real-time preprocessing algorithm for PPG signals. The key steps include:

1. Noise Removal: Developing an algorithm to preprocess and remove noise from the PPG signals.
2. Real-time Implementation: Ensuring the algorithm works in real-time.
3. Feature Extraction: Extracting features from the filtered waveform using various Signal Processing algorithms, with potential for AI techniques to be implemented
4. Deployment: Deploying the resulting system on an embedded (portable) system.

Task:

1. Literature Review
2. Data Preprocessing and Feature Engineering
3. SpO2 Device Integration /Simulators
4. Real-time Signal Processing and AI Modeling

References:

• Abdullah S, Hafid A, Folke M, Lindén M and Kristoffersson A (2023) PPGFeat: a novel MATLAB toolbox for extracting PPG fiducial points. Front. Bioeng. Biotechnol. 11:1199604. doi: 10.3389/fbioe.2023.1199604
• S. Abdullah, A. Hafid, M. Lindén, M. Folke and A. Kristoffersson, "Machine Learning-Based Classification of Hypertension using CnD Features from Acceleration Photoplethysmography and Clinical Parameters," 2023 IEEE 36th International Symposium on Computer-Based Medical Systems (CBMS), L'Aquila, Italy, 2023, pp. 923-924, doi: 10.1109/CBMS58004.2023.00344.
• The 2023 wearable photoplethysmography roadmap. DOI 10.1088/1361-6579/acead2

Knowledge of Python or Matlab, Signal processing, AI, and Machine learning. A basic understanding of biosignals is beneficial.

Proficiency in English.

For further information, please contact saad.abdullah@mdu.se