Title :
EEG brain map reconstruction using blind source separation
Author :
Sanei, Saeid ; Leyman, A.R.
Author_Institution :
Sch. of Electr. & Electron. Eng., Singapore Polytech., Singapore
fDate :
6/23/1905 12:00:00 AM
Abstract :
EEG-based brain maps are very useful in anatomical, functional and pathological diagnosis. These images are projections of the energy of the signals in four different frequency bands. Joint approximate diagonalization of eigenmatrices (JADE) is used as an effective tool in the deconvolution of EEG signals prior to spectrum estimation. The algorithm also, restores the noise from the signal as a result of higher order statistics (HOS) estimation. The spectrum is estimated using autoregressive (AR) modelling and pseudo-hot colours are used to represent brain activities. The results show a great enhancement in diagnostic features in the reconstructed images. The overall system also enables real-time reconstruction of the images for patient monitoring purposes
Keywords :
autoregressive processes; bioelectric potentials; biomedical imaging; deconvolution; eigenvalues and eigenfunctions; electroencephalography; higher order statistics; image reconstruction; matrix algebra; medical image processing; parameter estimation; patient monitoring; random noise; spectral analysis; EEG brain map reconstruction; EEG signal deconvolution; HOS; anatomical diagnosis; approximate diagonalization; autoregressive modelling; blind source separation; brain activities; diagnostic features; eigenmatrices; frequency bands; functional diagnosis; higher order statistics; image enhancement; pathological diagnosis; patient monitoring; pseudo-hot colours; spectrum estimation; Blind source separation; Brain mapping; Deconvolution; Electroencephalography; Frequency; Image reconstruction; Image restoration; Joints; Pathology; Spectral analysis;
Conference_Titel :
Statistical Signal Processing, 2001. Proceedings of the 11th IEEE Signal Processing Workshop on
Print_ISBN :
0-7803-7011-2
DOI :
10.1109/SSP.2001.955265
