Título de la tesis


Análisis multimodal de las señales fotopletismográfica de pulso y electrocardiográfica para el diagnóstico del síndrome de apnea obstructiva durante el sueño en niños

Doctorando


Eduardo Gil Hernando

Director


Pablo Laguna

Descripción


A multimodal study of pulse photoplethysmography (PPG) and electrocardiogram (ECG)

signals and physiological measures derived from them as heart rate variability (HRV)

and pulse transit time (PTT) has been tackled in this PhD thesis to help sleep disorders

diagnosis in children by means of the autonomic nervous system (ANS) activity analysis.

The main goal consists of studying the existing relationship between these signals and sleep

apnea to deine new clinical indexes to help obstructive sleep apnea syndrome (OSAS)

diagnosis using a more economical and less complex technique than the gold standard

diagnostic method.


First of all, the relationship between sleep apnea and decreases in the amplitude luc-

tuations of PPG (DAP) events has been studied. Our hypothesis is that DAP events are

directly related to arousals as a consequence of an apnea by means of a sympathetic system

activation. In this way, an automatic detection of DAP events might be useful in indirectly

quantifying sleep apnea. A DAP events detector, based on detecting the envelope reduction

of the PPG by means of an adaptive threshold, which also includes an artifact detector,

and an apnea detector from respiratory signals, where respiratory low cessation and its

association with oxygen desaturation are detected, have been developed for this study.


Afterwards, optimum values of detectors parameters have been determined and evaluated

using simulated signals and data bases of real registers. Next, a study has been carried out

to analyze the relationship between DAP events and apneas and the efect of polysom-

nographic segments length for classifying has been studied. The diagnostic performance

of the index derived from PPG has been analyzed by means of a clinical study over the

MSDBII database, which is composed of polysomnographic real signals of children recor-

ded in the Children Hospital "Miguel Servet" (Zaragoza). The obtained results show that

a relationship between DAP and apnea exists verifying our hypothesis and therefore it has

been proved that DAP events provide important information for OSAS diagnosis.


Nevertheless, since not all DAP events are associated with an apnea, a HRV analysis

during DAP events has been tackled as an alternative criteria for discriminating between

DAP events associated with apnea and those without that association. As HRV signal

during apnea is clearly non stationary, time-frequency speciic methods have been used for

the frequency analysis of the signal.


A set of both temporal and frequencial indexes derived from HRV have been deined.

ANS variations during DAP and the performance of these indexes to discriminate between
apneic and non apneic DAP events have been studied by means of two Kruskal-Wallis non

parametric statistical test. DAP events have been classiied as apneic or non apneic by

a linear discriminant analysis based on these indexes. The diagnostic performance of the

index derived from PPG and HRV has been analyzed by means of a clinical study over

the MSDBII database. The obtained results show that an increase in sympathetic activity

occurs during DAP events and that HRV makes discrimination between apneic and non

apneic DAP events possible, which improves the diagnostic method performance.


PPG signal has been directly related to the cardiac function, giving as a result the

pulse transit time (PTT) signal. PTT variability (PTTV), which is supposed to mainly

relect sympathetic activity, analysis during DAP events has been carried out as another

alternative criteria to discriminate between apneic and non apneic DAP events.

A study, similar to the analysis of HRV, has been carried out. Firt of all, PTT signal

has been obtained as the time interval between the peak of R wave in ECG and the 50 %

peak value of the associated PPG wave. Time-frequency techniques have been used for

PTTV frequencial analysis because this signal is non stationary.


A set of both temporal and frequencial indexes derived from PTTV has been deined.

ANS variations during DAP and the performance of these indexes to discriminate between

apneic and non apneic DAP events have been studied by means of two Kruskal-Wallis

non parametric statistical test. DAP events have been classiied as apneic or non apneic

by a linear discriminant analysis based on these indexes. The diagnostic performance of

the index derived from PPG and PTTV has been analyzed by means of a clinical study

over the MSDBII database. The obtained results show that PTTV relects sympathetic

changes more clearly than HRV. In addition, including information of ANS improves the

performance of the diagnostic method. Results are very similar when this information is

derived from PTTV or from HRV, although the diagnosis is more robust and less dependent

on the detectors parameters in case of PTTV.




Estado


Defendida - 2009

Calificación


Sobresaliente Cum Laude



Validación de código - XHTML 1.1
Validación de CSS
Validación de Accesibilidad Web