Institutional Repository
Technical University of Crete
EN  |  EL



My Space

Compressed sensing methods for analysis of epileptic seizures

Tsekos Nikolaos

Full record

Year 2020
Type of Item Diploma Work
Bibliographic Citation Nikolaos Tsekos, "Compressed sensing methods for analysis of epileptic seizures", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2020
Appears in Collections


This thesis investigates the preservation of Electroencephalogram’s (EEG) focal epileptic seizure’s information after compression and reconstruction of the signal through the Compressed Sensing (CS) technique and the effectiveness of Common Spatial Patterns (CSP) analysis of EEG signals on the automatic detection of focal epileptic seizures. Epilepsy is a neurological disorder characterized by an enduring predisposition to generate epileptic seizures with great neurological, cognitive, psychological and social consequences. According to the World Health Organization (WHO), it is estimated that in 2019, epilepsy affects around 50 million people worldwide which is quite common in childhood. In 2017, the prevalence and incidence of epilepsy are estimated to be 6.38 and 0.67 per 1000 persons respectively. In this thesis, we use the Discrete Cosine Transform (DCT) to have a sparse representation of the information, in order to be able to apply the CS technique. We further reduce the power of the EEG in order to have a sparser signal. After compressing the signals, we use the Basis Pursuit algorithm to reconstruct the sparse DCT signal and then the inverse Discrete Cosine transform to return to the time domain. Then we apply the Fourier transform and the Approximate Entropy to check the preservation of the original information of the seizure. CSP analysis has been frequently used in literature for multichannel EEG signal separation between -mainly- two states. In the present thesis, the EEG recordings from 10 subjects aged 6.8±5.9 years, including 63 seizures, were analyzed with respect to seizure detection and discrimination between inter-ictal and ictal signal periods. Machine learning techniques of feature selection and classification were used in the analysis.

Available Files