|
Clustering of ECG cycles
Ředina, Richard ; Smíšek, Radovan (referee) ; Ronzhina, Marina (advisor)
The bachelor thesis explores the aplication of cluster analysis on diferent ECGs in order to create a reliable algorithm for detecting different QRS complexes. Algorithm comprises filtration, R-wave positions adjustment, model cycle creation and comparasion based on mean square error and correlation. Both, correlation and mean square error, become data for k-means clustering. The number of clusters is derived from silhouette values for diferent numbers of clusters.
|
|
Automatic stress detection using non-EEG biological signals
Malina, Ondřej ; Kolářová, Jana (referee) ; Smíšek, Radovan (advisor)
This work deals with the problem of stress detection using non-EEG biosignals. The first part deals with the definition of stress and related concepts. Describes possible views of the phenomenon of stress, mentions possible causes of stress, as well as physiological and psychological manifestations of short and long-term effects of stress. In addition, this work deals with several different methods used to detect stress with non-EEG signals. For this purpose, a short search of articles dealing with this topic is available in this paper. The last chapter of this work describes the algorithm design using the c-mean fuzzy method for detecting stress values in data obtained form five different non-EEG signals.
|
|
Comparison of heart activity sensing devices
Babicová, Martina ; Smital, Lukáš (referee) ; Smíšek, Radovan (advisor)
The goal of this work is comparison of heart activity sensing devices. However, an ECG record cannot be evaluated with the presence of muscle interference. Removing this noise is one of the needs for device success. The theoretical part represents electrophysiology of the heart, electrocardiography, various interferences types, theoretical basis for recording of biosignals including used devices and methods signal quality estimation. The practical part is SNR (signal-to-noise ratio) calculation. The Wavelet filter and Wiener filter-based wavelet domain are used to separate the useful and noise component.
|
|
ECG signal quality annotation
Waloszek, Vojtěch ; Smíšek, Radovan (referee) ; Vítek, Martin (advisor)
This thesis gives basic information summary about electrophysiology of heart and electrocardiography and overview of several signal quality assessment methods. It also presents a new method for evaluating ECG quality, shows how signal quality indices are extracted and how the quality annotation is performed. It also gives test results of how the signal quality indices reflect the presence of corresponding noise and whether the quality annotation is correct.
|
|
Classification of free living data
Rychtárik, Martin ; Vítek, Martin (referee) ; Smíšek, Radovan (advisor)
The topic of this bachelor thesis is classification of free living data, captured by the accelerometer sensor of a smart phone. The first part of the thesis deals with the possibilities of recording daily activity using accelerometer and subsequent classification by neural network. In the next section, the data of eight different daily activities were recorded on ten people. An algorithm containing a neural network was created for the data in the MATLAB programming environment to automatically identify the activities. In the last part of the work the algorithm classification was compared with manually recorded reference and the results were statistically evaluated.
|
|
Automatic detection of strict left bundle branch block
Němčáková, Jesika ; Ronzhina, Marina (referee) ; Smíšek, Radovan (advisor)
The aim of this paper is to introduce the theory behind electrophysiology of hearth and pathology of left bundle branch. Furthermore, algorithms for automatic detection of left bundle branch block (LBBB) according to strict criteria are proposed. Algorithms are tested based on data from THEW databes within the Matlab interface. The final part of the bachelor’s thesis is devoted to the evaluation of the success of individual algorithms and the comparison of results on training data with test data.
|
|
ECG quality estimation
Pospíšil, Jan ; Smíšek, Radovan (referee) ; Smital, Lukáš (advisor)
This bachelor thesis deals with the question of estimation of the quality of the ECG signals, which is a key parameter for determining the diagnosis. The theoretical part deals with the basic knowledge concerning cardiac physiology, electrocardiography and finally the types of interferences that can occur during the measurement. The following practical part will deal with the published methods and the proposal of methods for estimating signal quality and their testing on artificial and real data.
|
|
Recommendations for ECG Acquisition Using BITalino
Němcová, Andrea ; Maršánová, Lucie ; Smíšek, Radovan
Cardiovascular disorders are still the most common cause of death in Western countries. In addition to cardiologists´ care, patient self-examination is the growing area nowadays. For this purpose, mobile devices for ECG signal acquisition are suitable. From these devices, we selected low-cost system called BITalino. In this work, we tested the quality of ECG signals acquired with BITalino under various conditions. We tested 2 values of sampling frequency, 12 electrodes placements and 6 types of exercises. We recommend using sampling frequency of 1,000 Hz, two electrodes placements – on chest and on wrists for both resting and exercise ECG.
|
|
ECG signal classification based on SVM
Smíšek, Radovan
Cardiovascular diseases nowadays represent the most common cause of death in Western countries. Long-term ECG recording is modern method, because it allows to detect sporadically occurring pathology. We designed an automatic classifier to detect five pathologies (AAMI standard) by SVM method. The classifier was tested on the entire MIT-BIH Arrhythmia Database with an accuracy of 99.17 %. We also compared the quality of parameters entering the classifier.
|
|
Detection of ventricular extrasystoles using high-frequency components of ECG
Budíková, Barbora ; Smital, Lukáš (referee) ; Smíšek, Radovan (advisor)
Pathology of the ventricular extrasystole is commonly detected by comparing the width and other parameters of the QRS complex. This work represents recognition of extrasystoles using high-frequency ECG components, captured under special conditions, and rating of succes of this method. The detection is performed by algorithm in the Matlab programming environment and its output is an impulse propagation map and a decision, whether the complex is a ventricular extrasystole or a physiological complex QRS.
|