University of Ioannina
Ανάπτυξη ευφυούς συστήματος για τη διάγνωση του ηλεκτρομυογραφήματος
Abstract
dc:descriptionIn this PhD thesis, a methodology for the automated and quantitative analysis of the electromyographic signal and the development of an intelligent system (and a methodology), for the affective assessment of individuals is presented. Initially, the way the neural and muscle cells function, as well as their structure is described. Then the features of the electromyographic signal, the elements used for its recording and the implementation fields of the electromyogram are outlined. Moreover, the diagnostic value of the electromyogram and the main difficulties that appear during its analysis are reported, while the concept of quantitative analysis of the electromyogram, the techniques for quantitative analysis that have appeared in the literature and their limitations are presented. Then, the structure of the proposed methodology for the automated quantitative analysis of the electromyogram is described. The methodology includes four stages: (i) Preprocessing of the signal aiming in the automated detection of regions of interest (candidate motor unit action potentials), (ii) Clustering of the motor unit action potentials (MUAPS) and calculation of the representative MUAP (template MUAP) for each group. Moreover in this stage the composite action potentials are automatically located and analyzed in their constituent MUAPS. (iii) Calculation of features for every template MUAP (amplitude, duration, rising time, area and number of phases), (iv) Classification of the extracted template MUAPs in normal, myopathic and neuropathic. In this stage two different approaches are used. In the first, a system based on support vector machines is used, while the second is based on a hybrid two-stage classifier. This classifier consists of a radial basis function artificial neural network (RBF ANN) for the classification of the template MUAPs into normal and pathological, and of a decision tree for the classification of the pathological MUAPs into myopathic and neuropathic. The extracted results are compared with the results from other methods mentioned in the literature. The way emotions affect vital signals of a person as well as the developed computational system for the automated estimation of the affective assessment of individuals and its implementation fields, are described. The system consists of three components: (i) The wearable subsystem, which consists of 16 facial electromyographic sensors, a piezoelectric respiration sensor, the electrocardiographic sensor and the skin conductivity sensor, (ii) The data acquisition and wireless communication subsystem, which is responsible for the safe reception and broadcasting of the selected biosignals and (iii) the centralized subsystem. The central unit processes the biosignals and produces a “diagnosis” concerning the emotional situation of the user. It also shows a 3-D model of the face of the user where his expressions are depicted in real time and it activates an alert system in the case the user experiences an “extreme” emotional situation. The proposed methodology for the estimation of basic emotional states of individuals consists of 3 stages, (i) The acquisition and preprocessing of biosignals (surface electromyograms of selected muscles of the face, electrocardiogram, respiration and electric conductance of the skin), (ii) the feature extraction and (iii) the assessment of the affective state of the user. In the third stage, the classification of the user’s affective state takes place into one of the predetermined emotional states. For this purpose a support vector machine system is used as well as an Adaptive Neuro-Fuzzy System (ANFIS). The proposed computational system is applied and validated in two different fields of interest: (i) In simulated conditions of car racing driving and (ii) in the framework of a pro-clinical study. The performance of the proposed system was found high compared to the performance of other systems appeared in literature.
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 2008
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Katsis, C.
- Κατσής, Χρήστος
Subjects
dc:subject × 20- Ηλεκτρομυογράφημα, Ποσοτική ανάλυση
- Επεξεργασία ιατρικού σήματος
- Πρότυπα, Αναγνώριση
- Αναγνώριση συναισθηματικών καταστάσεων
- Τεχνητή νοημοσύνη
- Συστήματα διανυσμάτων υποστήριξης
- Δίκτυα, Τεχνητά νευρωνικά ακτινικών συναρτήσεων βάσης
- Δέντρα αποφάσεων
- EMG, Quantative decomposition
- Template matching
- biomedical signal processing
- Affective state recognition
- Artificial intelligence
- Support vector machines
- Networks, Radial basis function artificial neural
- Decision trees
- Ιατρική και Επιστήμες Υγείας
- Βασική Ιατρική
- Medical and Health Sciences
- Basic Medicine
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/20645