Biomedical signal processing

A.Y. 2020/2021
Overall hours
Learning objectives
The course has the goal of providing the theoretical and practical knowledge required for processing biomedical signals and time series.
Expected learning outcomes
The student will be able to extract features from biomedical signals and time-series, also by means of spectral analysis; second, she/he will be familiar with the characteristics of the main biomedical signals; finally the student will be able to design and employ digital filters to remove noise and artifacts from signals acquired by medical applications.
Course syllabus and organization

Single session

Lesson period
First semester
Teaching methods
Classes will be held on the Microsoft Teams platform. They can be accessed either synchronously, at the normal time of class, and in asynchronous mode since they will be recorded and made available to students on the same platform.

Syllabus and reference materials
The syllabus and reference material will not be affected.

Learning verification procedures and assessment criteria
Remote examinations will be carried out with the use of the platform, with the modalities and rules described on the university portal. The written test will have the same structure of the standard one. The oral examination will take place on the Microsoft Teams channel of the class.
Course syllabus
First part
· Main biological signals and their properties
· Sampling of continuous time signals
· Linear Time Invariant (LTI) systems, impulse response, frequency response and transfer function
· Finite and Infinite Impulse Response filters (FIR & IIR)
· Design of IIR filters by direct placement of poles and zeros. Classic IIR filters
· Design of linear-phase FIR filters with the window method

Second part
Statistical characterization of signals
· Introduction to stochastic processes
· Autoregressive (AR) stochastic processes and their usage as model of biological signals
· Estimation theory basics

Spectral analysis
· Non parametric and parametric spectral estimates
· Spectral analysis of the heart rate variability (HRV) signal

Source separation
· Enhancement of repetitive patterns through averaging
· Cross-correlation& matched filters

Entropies and regularity of a signal
· Entropy rate of a signal and its use in the quantification of the regularity of a signal
· Entropy rate estimators for biological signals

Long time correlations and fractals signals
· Fractal sets and fractal dimension estimate
· Long memory and self-similar processes; 1/f noise
· Estimation of the scaling exponent in biological time series
Prerequisites for admission
The class has no specific preliminary requirement, except those necessary to enter the master degree in computer science.
Teaching methods
The course will be composed of: a) academic lectures; b) solution of exercises, previously assigned to the students; c) possible implementation in Matlab of the algorithms discussed; d) scientific paper discussions.
Teaching Resources
After each class, the learning materials (either slides of the copy of what written on the electronic board) will be made available on the Ariel course web site:

For the first part of the class, the reference textbook is:
James H. McClellan, Ronald W. Schafer, Mark A. Yoder
Digital Signal Processing First, Second edition (o DSP First, 2nd edition)
Pearson Education, 2016. ISBN-13: 978-1292113869

For the second part, there is yet no single text covering the breath and depth of the issues in this course and the materials will be provided on Ariel.
Assessment methods and Criteria
Each student can select between two ways in which the exam for the class can be taken (the selection is declared on the day of the examination, by stating it in the first page of the test):

1) Written examination on the first and second part of the class (90% of the final grade) plus a quick oral interview (10% of the final grade) where the results of the written examination are discussed. The written examination of the duration of 2 hours includes 5 open-ended questions and exercises.

2) Written examination on the first part of the class (50% of the final grade) plus the public discussion of a project (50% of the final grade). The written examination of the duration of 2 hours includes 3 open-ended questions and exercises. For the project, a student, or a party of two students (each contributing independently to a part of it) selects one of the scientific papers assigned in class and implements the algorithm proposed therein (in Matlab). The code is then tested on biological signals freely available on line, or provided by the instructor, or collected with a device. The student will submit (a few days before the day of the exam) the code, with clear explanations on how to run it, and then, during the exam, she/he will present in a 10 minutes (sharp) presentations (20 minutes for parties of two students, 10 minutes each student) the work.

At the end of the exam, the final grades are on a scale of 30 and they are assigned considering the following criteria: knowledge of the topics, ability in applying the knowledge acquired on practical problems, ability to resolve problems independently, clarity in expressing concepts.
Lessons: 48 hours
Professor: Sassi Roberto
Educational website(s)
By appointment (email or phone)
Dipartimento di Informatica, via Celoria 18, stanza 6004 (6 piano, ala Ovest), Milano