|
|
Optimization of a Deep Neural Network Label Encoding in a Multi-Label Problem.
Zaťko, Martin ; Novotná, Petra (referee) ; Hejč, Jakub (advisor)
The aim of the diploma thesis is to propose a method of deep learning for the classification of arrhythmias from ECG recordings and to compare the effect of coding its outputs on the overall quality of the model. A 1D convolutional neural network was selected and methods of label coding using one-hot coding, ordinal coding, the method using an autoencoder and the word embbeding method were tested and compared on it. The obtained results show that the use of the word embbeding method can increase the classification capacity of the proposed network.
|
|
|
Generative Adversial Network for Artificial ECG Generation
Šagát, Martin ; Ronzhina, Marina (referee) ; Hejč, Jakub (advisor)
The work deals with the generation of ECG signals using generative adversarial networks (GAN). It examines in detail the basics of artificial neural networks and the principles of their operation. It theoretically describes the use and operation and the most common types of failures of generative adversarial networks. In this work, a general procedure of signal preprocessing suitable for GAN training was derived, which was used to compile a database. In this work, a total of 3 different GAN models were designed and implemented. The results of the models were visually displayed and analyzed in detail. Finally, the work comments on the achieved results and suggests further research direction of methods dealing with the generation of ECG signals.
|
|
|
Computer Simulation of Quadrupole Mass Filter
Kandra, Mário ; Kolářová, Jana (referee) ; Hejč, Jakub (advisor)
At our thesis, we deal with problematics of mass spectrometry. First, we are trying to explain basic physical principles, which are used by quadrupole filter or are appearing in concomitant phenomenons. In following section we will describe theoretical nature of quadrupole mass filter, stabilization and optimalization of device. Main focus of this thesis is dedicated to simulation of quadrupole mass filter and influence of parameters on trajectory of ion movement.
|
|
|
QRS complex detection in long electrograms
Novotná, Petra ; Hejč, Jakub (referee) ; Kozumplík, Jiří (advisor)
This bachelor thesis deals with the detection of the QRS complexes in long electrograms which were acquired by scanning isolated rabbit hearts. The first section of this thesis contains information about general detection principles. Some of those principles are de- scribed into details in the fourth chapter. The practical part consists of: the introduction of processed data, the description of specific algorithms used while solving the task and the realisation of those algorithms in the language of technical computing (MATLAB).
|
|
|
Membrane Oxygenator
Jindra, Jakub ; Harabiš, Vratislav (referee) ; Hejč, Jakub (advisor)
This bachelor thesis deals with the creation of an oxygenator for isolate rabbit’s hearts. In this work is created a mathematical model of the oxygenator and calculate optimal parameters for this application. Design of the construction solution leads to creating a 3D model and print of the oxygenator proposal and creation of the functional prototype. The conclusion of this work is the evaluation of the functionality of the equipment and the discussion of the achieved results.
|
|
|
PVC detection in ECG
Imramovská, Klára ; Hejč, Jakub (referee) ; Ronzhina, Marina (advisor)
The thesis deals with problems of automatic detection of premature ventricular contractions in ECG records. One detection method which uses a convolutional neural network and LSTM units is implemented in the Python language. Cardiac cycles extracted from one-lead ECG were used for detection. F1 score for binary classification (PVC and normal beat) on the test dataset reached 96,41 % and 81,76 % for three-class classification (PVC, normal beat and other arrhythmias). Lastly, the accuracy of the classification is evaluated and discussed, the achieved results for binary classification are comparable to the results of methods described in different papers.
|
|
|
Exercise-based predictors of atrial fibrillation recurrence in patients undergoing catheter ablation.
Mátych, Martin ; Pešl, Martin (referee) ; Hejč, Jakub (advisor)
Atrial fibrillation (AF) is the most frequently treated heart arrhythmia. Radiofrequency catheter ablation is a treatment option with a success rate ranging from 60 % to 80 % for paroxysmal AF. This work aimed to determine parameters associated with AF recurrence to identify high-risk patients. Data from 98 patients who underwent pulmonary vein isolation were analyzed. Out of these patients, 19 experienced AF recurrence. Exercise and echocardiographic parameters differed significantly between the recurrence and non-recurrence groups and were used in regression analysis. Peak oxygen consumption (pVO2) was found to be a strong predictor of AF recurrence after adjusting for gender and age (hazard ratio 0.43). Four parameters were identified as the ideal combination in multivariable analysis: pVO2, septal peak late diastolic mitral annulus velocity, post-exercise systolic blood pressure, and left atrial volume index. These findings highlight the importance of stress and echocardiographic parameters in predicting the success of ablation procedures.
|
|
|
ECG-based biomarkers for estimation of sudden death risk in patients with accessory atrioventricular pathway.
Tomková, Lucia ; Ředina, Richard (referee) ; Hejč, Jakub (advisor)
The aim of this work is to acquaint the reader with the delta wave manifestation on the electrocardiogram in pediatric patients with Wolf-Parkinson-White syndrome who are at risk of sudden cardiac death. Subsequently, in the work we deal with the processing of data from the electrophysiological examination and the creation of algorithms for the extraction of biomarkers from the ECG signal in the Python programming language. The work includes statistical analysis of biomarkers, focusing on the difference in datasets for at-risk and non-at-risk patients. Biomarkers that have a statistically significant difference in the datasets are used to create and select a suitable model for predicting the risk of the accessory pathway in patients.
|
|
|
Analysis of Atrial Fibrillation Heart Rate Dynamics
Tesařová, Tereza ; Ronzhina, Marina (referee) ; Hejč, Jakub (advisor)
Bachalor thesis focuses on clinical tracking of heart rate abnormalities to specify the level of variability in diverse-lengthy ECG signals, with a direct way to analyze atrial fibrillation and accuracy increasing its detection of actual available methods. The principle of heart electrophysiology and also the impulse genesis is mentioned in a introductory part of work, including elementary classification of cardiac arrhytmias and their deflection from normal sinus rhythm. The physiological analysis of atrial fibrilation dynamic structure helps with the application of detection algorithms in a MATLAB program and sets an underlying basis for an appropriate discussion in a final statistical ROC analysis.
|
|
|
Optimization of a Deep Neural Network Label Encoding in a Multi-Label Problem.
Zaťko, Martin ; Novotná, Petra (referee) ; Hejč, Jakub (advisor)
The aim of the diploma thesis is to propose a method of deep learning for the classification of arrhythmias from ECG recordings and to compare the effect of coding its outputs on the overall quality of the model. A 1D convolutional neural network was selected and methods of label coding using one-hot coding, ordinal coding, the method using an autoencoder and the word embbeding method were tested and compared on it. The obtained results show that the use of the word embbeding method can increase the classification capacity of the proposed network.
|
guest ::
