On application of rational Discrete Short Time Fourier Transform in epileptic seizure classification

Epileptic Seizure Classification of EEG Time-Series Using Rational Discrete Short-Time Fourier Transform

“…We also showed that the inverse rational DSTFT achieves a reconstruction signal with a smaller mean square error compared to the classical STFT for the same number of coefficients. Finally, we compared the performance of our feature extraction procedure with 13 Cohen's class t-f distributions. Several state-of-the-art feature extraction methods were taken into account, which are commonly used in epileptic seizure detection and classification.…”

“…Furthermore, these systems were successfully applied for modeling the QRS complex [12] and for compressing the heart beats as well. We note that the basic concept of this method was originally proposed in [13]. In this paper, we extend the previous work by providing a detailed description of the time-frequency model, adding feature analysis and more comparisons with the state-of-the-art methods.…”

“…However, wavelet transforms are more appropriate for analyzing non-stationary EEG signals since they provide a simultaneous representation in the timefrequency domain. For instance, Adeli et al [6], [7] applied Daubechies wavelet transform to analyze the epileptic seizure pattern in five frequency sub-bands: delta (0-4 Hz), theta (4-7 Hz), alpha (8)(9)(10)(11)(12), beta (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30), and gamma . Using these five sub-bands and chaos theory, nonlinear dynamics of the EEG are simultaneously quantified by a classifier.…”

Analysis of EEG Signal Processing Techniques based on Spectrograms

“…SVM is applied, obtaining competitive accuracies for different datasets. On the other hand, [24] extracts Malmquist-Takenaka coefficients from Spectrograms and statistical features, with the same objective. In this methodology, STFT is in discrete form, and the classification uses an Alternating Decision Tree (ADTree) classifier with three different datasets.…”

“…GLCM y VF. From the different, it was noted that for the analysis of epilepsy diseases, there is a set of data that different authors [3,20,23,24] employ, for example use a dataset created by the University of Bonn with five different subsets (A-E), with 100 single channel signals, a duration of 26.3 seconds and a sampling frequency of 173.61 Hz. This database contains high quality information (number of studies, time and number of seizures), which were validated by experts.…”