|
|
 |
|
| Cristin Bigan |
 |
Unirii 15 bl.3 ap.9 sect.5 Bucharest, Romania
e-mail: ibigan@pcnet.pcnet.ro |
Poster Presentation: |
|
| TIME - FREQUENCY METHOD FOR EEG BRAIN SIGNALS ANALYSIS AND MAPPING. |
| Cristin Bigan | |
|
Politehnica University of Bucharest, Romania |
|
| 1. Introduction
EEG recording is used for understanding the normal or pathological problems involved in human brain activity. Identifying an EEG with a specific event and his nature can help support a diagnosis, and may also be used to classify the type of specific event (normal, artefact, spike, seizure, K-complexes, sleep spindles, etc.). From this work, some useful clinical interpretation results were obtained on event detection. Time-frequency, pre-processing results were features used to a NN input. The system performs a NN and knowledge based detection. Results from the proposed method of time-frequency analysis can be also used for the dynamical representation of brain instantaneous frequency mapping on a single picture.
2. The EEG Recording of Brain Neural Activity
The EEG research investigations are trying to improve the techniques of extracting the useful signal from the high level of external or internal noise. Next step is the data pre-processing in order to enhance the features for further extraction.
3. Instantaneous Frequency detection
Using a proposed polynomial decomposition method an instantaneous frequency is obtained from each short time EEG epoch with a much better resolution than classical FFT.
We define the spectral epoch centroid to be the vector that reduces to a minimum the frequency error inside the epoch.
The result is the optimal epoch’s frequency given by the corresponding centroid. We can consider that the EEG signal is represented at this stage by the temporal successive spectra (frequency values). The initial EEG signal can be represented by means of the characteristic features array.
4. Automatic detection and mapping
Artificial neural networks can be employed to perform the recognition of certain EEG frequency feature having as inputs the features array. Alternatively each epochs frequency can be used to a color code scale in representing a frequency brain mapping.
|
|
|
 |
|
|
|
