Uporaba valčkov pri analizi EKG signalov : magistrsko delo

Bjelkić, Ajdina (author), Knez, Marjetka (mentor)

Elektrokardiogram (EKG) vsebuje številne informacije o delovanju srca, s pomočjo katerih lahko napovemo in zaznamo različne srčne bolezni. Glavni korak pri samodejni analizi EKG signalov je detekcija QRS kompleksov, s pomočjo katerih zaznavamo srčne utripe. V tem magistrskem delu se bomo seznanili z različnimi valčki in valčnimi transformacijami ter predstavili metodo za detekcijo QRS kompleksov, ki temelji na uporabi valčkov oz. diskretne valčne transformacije. Opisali bomo tudi postopek za predobdelavo signalov s pomočjo diskretne valčne transformacije, s katerim odstranimo nekatere šume. Metodo za detekcijo bomo vrednotili na dveh bazah, MIT-BIH bazi aritmij ter CU bazi ventrikularnih tahiaritmij. Poleg tega bomo primerjali delovanje metod pri uporabi treh vrst valčkov, Daubechiesinih valčkov, coifletov in symletov. Na koncu pa bomo metodo tudi kvalitativno vrednotili na izbranih posnetkih iz MIT-BIH baze, ki vsebujejo določene značilnosti.

• Language: Slovenian
• Keywords: valčki, EKG, analiza, diskretna valčna transformacija
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FMF - Faculty of Mathematics and Physics; FRI - Faculty of Computer and Information Science
• Year: 2021
• PID: 20.500.12556/RUL-131667
• UDC: 519.6
• COBISS.SI-ID: 79049219
• Publication date in RUL: 01.10.2021

Secondary language

• Language: English
• Title: Application of wavelets to ECG signal analysis

Abstract

An electrocardiogram contains a lot of information about heart function, which we can use to predict and detect different types of heart diseases. The main step of automatic analysis of ECG sygnals is to find QRS complexes that are used to detect heart beats. In this master thesis we will learn about different wavelets and wavelet transforms. Moreover, we will present a method for QRS detection based on the use of wavelets and discrete wavelet transform. We will also describe a discrete wavelet transform based algorithm for a noise reduction. The detection method will be evaluated on two databases: MIT-BIH arrhythmia database and CU ventricular tachyarrhythmia database. In addition, we will compare the performance of the methods using three types of wavelets: Daubechies wavelets, coiflets and symlets. Finally, we will qualitatively evaluate the method on selected ECG recordings from the MIT-BIH database that contain certain features.

• Keywords: wavelets, ECG, analysis, discrete wavelet transform