Biomedical signal analysis
Electro-encephalo-graphy (EEG)
signal analysis through DSP algorithms
The recording of electrical signals emanated from human
brain, which can be collected from the scalp of the head
is called Electroencephalography (EEG). These signal's parameters
and patterns indicate the health of the brain. EEG is the
key area of biomedical data analysis.
Using Digital Signal Processing functions EEG signals can
be analyzed to properly diagnose the patient. The latest
biomedical embedded electronics systems with DSP processor
can display the computed results helping doctor to save
time in anayzing complex EEG waveforms.
The various DSP based analytic methods used to evaluate
are,
1. Spectral estimation
2. Periodogram
3. Maximum entropy method
4. AR method
5. Moving average methods
6. ARMA method
7. Maximum likelyhoodmethod
Here we provide some basics of each methods and further
links to study in more depth.
Spectral estimation: Spectral estimation helps in
finding the pulse rhythms present in the EEG signal. The
short segment of EEG data is analyzed for spectral parameters
such as location and amount of spectral energy. Waveshaping
filters are extensively used in this technique. Wave shaping
filters produce desired output signal for given input signal.
If the desired signal is a unit impulse it's called spiking
filters. Spiking filters can be used to determine the position
of the locations of the concentrated energy signal.
Such study of energy concentration in EEG signal is called
spectral estimation analysis
To know further read these papers/online materials
http://eprints.iisc.ernet.in/6870/1/Spectral_estimation_of_short_segments.pdf
http://www.commsp.ee.ic.ac.uk/~mandic/SE_ASP_LN/SE_ASP_Lecture_3_Modern_SE.pdf
http://outserv.cactus.org/~benjamin/X/marple.pdf
http://www.commsp.ee.ic.ac.uk/~mandic/SE_ASP_LN/SE_ASP_Lecture_3_Modern_SE.pdf
http://dspace.mit.edu/bitstream/handle/1721.1/4250/RLE-TR-493-15597448.pdf?sequence=1
https://darchive.mblwhoilibrary.org/bitstream/1912/2415/1/Briggs_thesis.pdf
Periodogram: Periodogram is nothing but the estimation
of spectral density. This can be obtained through estimated
correlation function.
Read these power point presentation at the link to know
about periodograms used to analyze EEG signals,
https://noppa.tkk.fi/noppa/kurssi/t-61.5080/luennot/slides.pdf
http://www.maths.bris.ac.uk/~mazlc/TSA/case02.pdf
Maximum entropy method: Helps to measure the randomness
and uncertainty associated with the EEG signal. Maximum
Entrophy method works even if any information or constraints
on a process X(n) are absent.
To know further on Maximum Entropy Method visit,,
http://www.commsp.ee.ic.ac.uk/~mandic/SE_ASP_LN/SE_ASP_Lecture_3_Modern_SE.pdf
AR method: Autoregressive (AR) is preferred when
signal's frequency has sharp peaks. AR is popular because,
the accurate estimation of PSD can be obtained by solving
linear equations. AR model is called all pole method where,
each sample of the signal can be expressed as a combination
of previous samples and an error signal.
Read this below online source to learn more on AR Method:
http://www.rgi.tut.fi/courses/materiaali/71413/71413Lecture5.pdf
ARMA (autoregressive moving average) method: This
model is suggested for modeling signals with sharp peaks
and valleys in their frequency content and also signals
with severe background noise.
The detailed explanation of ARMA model is available at,
http://www.rgi.tut.fi/courses/materiaali/71413/71413Lecture5.pdf
Maximum likelihood method: This blend the information
already available based on prior knowledge and latest measurements.
The value is optimal estimation of the actual value.
For more details on this method, read the book available
on books.google.co.in
The book name is EEG signal processing By Saeid Sanei, Jonathon
A. Chambers
To access this book visit books.google.co.in and enter the
book name and author
The other popular online reference material on biomedical
signal processing is,
Biomedical Signal Processing: Principles and Techniques
By Reddy
To access this book visit books.google.co.in and enter the
book name and author