| Summary: | 碩士 === 國立清華大學 === 電機工程研究所 === 82 === This thesis proposes some analytic methods to respiratory
physio logicalsignals. The goal of this thesis is to create
features of respiratory signals and apply them to the clinical
diagnosis and monitoring care system of respiratory system
diseases. Being dif ferent with ECG signals, respiratory
waveforms exist no unique s hape. Even for unique subject,
respiratory signals acquisited in different situation and time
will differ each others. So, the po ssible approach to identify
respiratory waveforms with different respiratory diseases is
using proper respiratory features to ind icate the
characteristics of waveforms. This thesis extracts res piratory
features from time domain , frequency domain and system
modeling. The purpose of feature extraction is to take
advantage of the difference of respiratory features for
different diseases to achieve the recognition and
classification of respiratory dis eases. This thesis
researches a general and integrated solution of respiratory
signal processing. First, through choice of respi ratory
sensors, selection and design of A/D and amplifier interf ace
circuits, and conversion of monitoring care data, we can acq
uisit respiratory data. Then , the proposed methods are used
to extract respiratory features. The extracted features are
used to train artificial neural network to activate their
classification ability. Finally, this thesis proposes some
clinical application s and conclusions based on result of
simulations.
|
|---|
