Advanced Biosignal Analysis

Presentation

Following the research-based spirit of our master, this course is centered around some of the work done at one of the participating research groups: The Nonlinear Time Series Analysis group (https://www.upf.edu/web/ntsa/), and it is taught by Simone Tassani.

The multifactorial and multidimensional signal analysis framework allows us to characterize complex systems and explore different causes that can lead to the pathological modification of signals. Importantly, these approaches allow us to study several factors at once, avoiding confounding the effect of each independent variable. Application of such methodology to recordings of human movement allows us to identify different causes of gait modification and contribute to our understanding of human motion.

In this course, we will focus on the particular case of movement and posture analysis recorded from both healthy volunteers and osteoarthritic subjects. The World Health Organization (WHO) reports that musculoskeletal diseases are responsible for 4% of health years lost due to disability worldwide. This percentage increased to 8% in High-Income Countries. Most worrying is the fact that about 10% of the people presenting musculoskeletal diseases are younger than 30.

Movement is the expression of many human characteristics, both from the bio-mechanical and social point of view. Many movements, like gait, are repeated several times a day. Every repetition is similar, but different from the others. There is growing evidence that the study of movement variability and coordination can provide useful information for the improvement of diagnosis and treatment of several musculoskeletal diseases and can also give an insight into the emotional state of the patients. Time series approaches, like statistical parametric maps and phase portraits, will be introduced to study the variability and coordination of human movement.

One example, to which the Nonlinear Time Series Analysis group is contributing, is the classification of osteoarthritic subjects requiring or not total knee replacement, or the study of posture and stability of young subjects.

We will learn about the potential, but also about the limitations, of multifactorial and multivariate approaches.

In the theory classes, we will learn about these signal analysis techniques, and their applications to study movement variability and coordination, as well as some basics about gait dysfunctions, and gait analysis from osteoarthritic and healthy subjects. In the lab sessions, we will apply the analysis techniques to real gait and posture recordings acquired in the motion laboratory of UPF. A focus will be placed on a meaningful interpretation of the derived results.

Prerequisites

The theory of the course is designed to follow the techniques introduced in the data science course of the first trimester. With this premise, it can be followed by students with any engineering bachelor's degree. This includes, but is not limited to, biomedical engineers. Students from physics, mathematics, or life sciences will also be able to follow the course. All these bachelor's degrees equip the students with the necessary technical and mathematical skills to take this course.

No prior knowledge of multidimensional or continuous analysis is required. The students should, however, be familiar with fundamental concepts of statistics. No prior knowledge of biomechanics of human movement is needed.

We will use Matlab for the data analysis done in the lab sessions. The students need therefore to be able to write and understand simple source code in this programming language.

Contents

The theory is organized in four blocks:

1. Overview
   • We will start with an overview and motivation of the course content. It will be highlighted why its content is relevant for biomedical engineers, regardless of whether they aim at a career in academia or in industry.
2. Movement and Movement Analysis
   • Movement is the expression of our life, and its analysis can give us several insights about our condition, the way to avoid musculoskeletal problems, or how to structure a rehabilitation process. This block will give us the theoretical basis to understand the data that will be used during the laboratory sessions and interpret the results of our analysis.
3. Analysis of Covariance and Multivariate Analysis of Variance
   • The Analysis of Variance introduced in the Data Science course will be expanded to include the Analysis of Covariance, and we will later start to analyze data multidimensionally. The concepts related to principal component analysis (PCA) introduced during the Data Science course will be related to Multivariate Analysis of Variance to explore several aspects of human motion.
4. Concepts of variability and techniques to explore it.
   • Variability over time will also be considered. First, repeated measure analysis will be introduced to study how signals can evolve over time studying specific discrete time points. Secondly, the time-continuous methodology will be introduced to explore variation of motion. Statistical parametric maps will be used to compare motion between groups and finally, phase portrait and angle-angle plots will be introduced to study variability of the single subjects.
   • Finally, different segments of the body do not move independently. Coordination of movement is often a biomarker that can allow us to understand if the subject is healthy or not. In this block, students will learn to use tools based on the Hilbert transform to explore such coordination.

Laboratory sessions

In the first lab session, the students will be provided with gait recordings from osteoarthritic patients or healthy subjects. They will work in groups of two or three students. Data will be in Matlab files, and in the first few labs, the student’s task will be to write some code to visualize the data, explore the data, and recognize the different phases of gait and their characteristics. Subsequently, the students will have to develop the code to implement the analysis introduced during the theory classes. Students will implement, in order, multifactorial, multivariate, and continuous analysis of the signal obtained. Finally, the students will implement a combination of techniques to explore the data provided.

Evaluation Methods

The students have to deliver:

• A well-commented source code, which they developed for the analysis implemented in each laboratory.
• A final written report on their exploration of the data and the tools they wrote for this purpose.

This material will not be directly evaluated, but their submission is mandatory. A final public presentation based on the material delivered will be given by each working group and will be evaluated.

The final exam will be individual and oral, and based on the material submitted. The grades will be derived from the quality of the presentation and the result of the final exam.