Aristotle University Of Thessaloniki (AUTH)
Ανάλυση αναπνευστικών ήχων με χρήση μη γραμμικών μεθόδων επεξεργασίας σήματος
Abstract
dc:descriptionThe dissertation deals with wheezes, i.e., pathological breath sounds whose presence is related to airways' obstruction. In particular, the dissertation addresses two problems corresponding to wheezes: their detection and the analysis of their characteristics. For their detection, two methods are proposed. In the first method, which is based on Short-Time Fourier Transform, different detection criteria are set for different frequency bands, since the spectrum of breath sounds is not the same in their whole frequency range. Additionally, through the inverse transform the representation in time of the signal which has been processed in the time-frequency domain is introduced. The proposed method exhibits high detectability, sensitivity and specificity (TDR=90.2%, SE=92.1%, SP=89.8%). Despite the high percentages, the resolution in time and in frequency of the suggested method is limited by the selected window length. The second method for wheeze detection is based on the combination of the ensemble empirical mode decomposition and the instantaneous frequency. The kernel of this method is the observation that the instantaneous frequency of a signal remains constant when the signal is dominated by particular frequency components, such as wheezes. The proposed method exhibits high detectability, sensitivity and specificity (TDR=94.0%, SE=94.9%, SP=98.4%), which are improved comparing to the previous method. For the analysis of wheeze characteristics, and in particular the detection of non-linear phenomena, third-order spectra, which preserve the phase information of the signal, are combined with the wavelet transform, which allows capturing non-linearities across time. Through wavelet bispectrum and bicoherence and the corresponding instantaneous values a new analysis domain is introduced. For the investigation of statistically significant differences between wheezes from patients with asthma and COPD, a set of features is introduced based on these mathematical tools. Summarizing, the current dissertation deals with the problem of wheeze detection and wheeze characteristics analysis based on combinatory approaches. The proposed methods may contribute in a diagnosis-assisting device based on breath sounds.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2010
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Taplidou, Styliani
- Ταπλίδου, Στυλιανή
Subjects
dc:subject × 13- Επεξεργασία βιοακουστικών σημάτων
- Συρίττοντες
- Εμπειρική αποσύνθεση ενδογενών ρυθμών συνόλου
- Φάσμα/συνάφεια κυματιδίων τρίτης τάξης
- Biacoustic signal processing
- Wheezes
- Ensemble empirical mode decomposition
- Wavelet bispectrum
- Wavelet bicoherence
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/28932