Biomedical Engineering

PIETRO ARICÒ

The course provides the student with advanced methods for extracting features from biomedical signals, both in the time and frequency domains, with particular regards on digital filters, multivariate analysis, time-frequency and time-scale spectral methods, examples of electroencephalographic, electrocardiographic, photoplethysmographic, and skin conductance signal processing. The course also provides basic tools for biomedical signal classification for brain-computer interface applications, such as LDA, SVM and clustering elements.

• Apply numerical processing methods to biomedical signals and data;
• Use analytical tools in the time, frequency, time-frequency, and time-scale domains;
• Design and implement digital filters (FIR, IIR, Wiener) for signal analysis and filtering;
• Apply multivariate analysis methods (PCA, ICA) to extract meaningful components;
• Analyze physiological signals such as ECG, PPG, EEG, EOG, and EDA;
• Understand and use the main basic MATLAB functions to develop algorithms for the analysis, processing, and visualization of biomedical signals.

● Geometry/Mathematical Analysis
● Biomedical Data and Signal Processing 1:
o Basic concepts of data transformation and related operators (e.g., Correlation, Cross-correlation, Auto-correlation, Convolution)
o Fourier Analysis
o Non-parametric Spectral Estimation
● Basic Programming Skills and Introductory Knowledge of the MATLAB Environment

• Course Introduction – Application examples of biomedical measurement systems, with a focus on passive Brain-Computer Interface systems based on EEG signals, and live demonstration.
• Biomedical Measurement System (review) – Analog filters and sampling.
• Digital Filters
o Z-transform
o FIR filters
o IIR filters
o Optimal filters, with a particular focus on Wiener filters
• Multivariate Analysis
o Principal Component Analysis (PCA)
o Independent Component Analysis (ICA)
• Time-Frequency Spectral Methods
o Short-Time Fourier Transform (STFT)
o Spectrogram
o Wigner-Ville Distributions
o Instantaneous autocorrelation function
o Choi-Williams Distributions
• Wavelet Transform
o Continuous Wavelet Transform (CWT)
o Time-Frequency features of the Wavelet Transform
o Discrete Wavelet Transform (DWT) via filter banks
• Parametric Spectral Estimation Methods
o Autoregressive Methods (review)
o Spectral estimation methods based on eigenanalysis
• Electrocardiographic (ECG) and Photoplethysmographic (PPG) Signal Processing
o Physiology of ECG and PPG signals
o Heart Rate (HR) and Heart Rate Variability (HRV)
o Time and frequency domain parameters
o Lomb-Scargle Periodogram
o Pan-Tompkins Algorithm
o Application Examples
• Electrodermal Activity (EDA) Signal Processing
o Physiology of the EDA signal
o Methods for extracting Tonic (SCL) and Phasic (SCR) components
o Continuous Decomposition Analysis applied to EDA signals
o Ledalab Algorithm
o Application Examples
• Electrooculographic (EOG) Signal Processing
o Physiology of the EOG signal
o Methods for extracting the Eye Blink Rate (EBR) parameter
o Methods for correcting ocular artifacts in EEG signals
o Gratton & Coles Regression Algorithm
o Application Examples
• Supplementary Seminars on the application (or advancement) of some processing methods covered in class.
• Practical Examples in MATLAB implementing the analysis and processing methods discussed during the course.

• Semmlow and Griffel, Biosignal and Medical Imaging Processing
• Teaching material provided by the instructor:
• Lecture slides (current and previous year)
• Prerequisite review slides from EDSB1
• Lecture recordings (audio with screen sharing)
• MATLAB code related to the in-class examples

• Semmlow and Griffel, Biosignal and Medical Imaging Processing
• Teaching material provided by the instructor:
• Lecture slides (current and previous year)
• Prerequisite review slides from EDSB1
• Lecture recordings (audio with screen sharing)
• MATLAB code related to the in-class examples

Teaching is delivered through traditional lectures, enriched with numerous practical examples in MATLAB, aimed at demonstrating the practical implementation of the signal processing algorithms covered in class. Lectures are recorded (voice + slide screen sharing) and made available to students, along with the teaching materials, including slides and MATLAB code developed step-by-step during the course.
Additional seminars are offered, focusing on practical applications of the algorithms (or advanced versions thereof), to provide students with concrete examples of how the developed techniques are used. Live demonstrations of biomedical measurement systems are also included.

Lectures will be conducted in-person. Exercises will be carried out in the Matlab environment, through practical examples of processing and classification of real biomedical data. Attendance is strongly recommended.

Multiple choice questions
Open-ended questions

Evaluation criteria:
The examination will be conducted in written mode and will consist of two parts.
The first part will consist of multiple-choice answers on the entire syllabus, 20 multiple-choice questions will be submitted, with 0.75 points for each correct answer, -0.25 points for each incorrect answer, 0 points for each answer not given.
The second part will consist of 3 open-ended questions of various types based on the arguments and examples developed in class. A maximum of 5.5 points will be awarded for each appropriately discussed answer.
In this second part, the following will be evaluated:
• Structured organisation of the response
• Synthesis skills, where required
• Appropriateness of terms used
• Ability to highlight connections between different parts of the programme.
• Proposal of original reflections made during the study of the subject Particularly brilliant examinations (≥31) will compete for laude.

• Biomedical Measurement System (review)  - 5 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Digital filters - 12 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Multivariate Analysis - 8 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Time-Frequency Spectral Methods - 8 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Wavelet Transform - 8 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Parametric Spectral Estimation Methods - 8 hours
  • Books: Semmlow and Griffel, Biosignal and medical Imaging Processing, Third Edition, CRC Press, 2014


• Autonomic signal processing (PPG/ECG/EDA/EOG) - 5 hours
  • Books: Slides


• Supplementary Seminars on the application (or advancement) of some processing methods covered in class - 6 hours
  • Books: Slides