Advanced

Time-Frequency Analysis of the Auditory Brainstem Response

Johansson, Henrik (2013) FMS820 20132
Mathematical Statistics
Abstract (Swedish)
This thesis is about time-frequency analysis of the brainstem auditory evoked
potential (BAEP). The work can be divided into two parts. One part where
a model is built up from a very simple example to a more complex model
resulting in a model consisting of a sum of sinusoids with stochastic starting
points and amplitudes. Dierent time-frequency methods have been evaluated
for these models and the multi window spectrogram with Hermitian
base functions performs the best in a real life situation with more than one
component and a high level of noise.
The second part consists of investigating real BAEP data. BAEP data from
ve patients were available. Each patient has two data sets which have
been studied. One while the patient is... (More)
This thesis is about time-frequency analysis of the brainstem auditory evoked
potential (BAEP). The work can be divided into two parts. One part where
a model is built up from a very simple example to a more complex model
resulting in a model consisting of a sum of sinusoids with stochastic starting
points and amplitudes. Dierent time-frequency methods have been evaluated
for these models and the multi window spectrogram with Hermitian
base functions performs the best in a real life situation with more than one
component and a high level of noise.
The second part consists of investigating real BAEP data. BAEP data from
ve patients were available. Each patient has two data sets which have
been studied. One while the patient is awake and one while it is asleep.
A hypothesis is that there exists some sort of dierence between these two
datasets. It turns out that it does. The earlier peaks dier slightly in latency
and the later peaks for the sleeping data seem to disappear. This result is
concluded from dierent time frequency methods, where the spectrogram and
the multi-window spectrogram are the most successful methods. An attempt
to make a bootstrap simulation in order to estimate the mean and condence
bounds of each peak is also made for one dataset. (Less)
Please use this url to cite or link to this publication:
author
Johansson, Henrik
supervisor
organization
course
FMS820 20132
year
type
H2 - Master's Degree (Two Years)
subject
language
English
id
4113810
date added to LUP
2013-10-21 14:53:46
date last changed
2013-10-21 14:53:46
@misc{4113810,
  abstract     = {This thesis is about time-frequency analysis of the brainstem auditory evoked
potential (BAEP). The work can be divided into two parts. One part where
a model is built up from a very simple example to a more complex model
resulting in a model consisting of a sum of sinusoids with stochastic starting
points and amplitudes. Dierent time-frequency methods have been evaluated
for these models and the multi window spectrogram with Hermitian
base functions performs the best in a real life situation with more than one
component and a high level of noise.
The second part consists of investigating real BAEP data. BAEP data from
ve patients were available. Each patient has two data sets which have
been studied. One while the patient is awake and one while it is asleep.
A hypothesis is that there exists some sort of dierence between these two
datasets. It turns out that it does. The earlier peaks dier slightly in latency
and the later peaks for the sleeping data seem to disappear. This result is
concluded from dierent time frequency methods, where the spectrogram and
the multi-window spectrogram are the most successful methods. An attempt
to make a bootstrap simulation in order to estimate the mean and condence
bounds of each peak is also made for one dataset.},
  author       = {Johansson, Henrik},
  language     = {eng},
  note         = {Student Paper},
  title        = {Time-Frequency Analysis of the Auditory Brainstem Response},
  year         = {2013},
}