The maximum length sequence (m-sequence) has been successfully used to study the linear/nonlinear components of auditory evoked potential (AEP) with rapid stimulation. However, more study is needed to evaluate the effect of the m-sequence order in terms of the noise attenuation performance. This study aimed to address this issue using response-free electroencephalogram (EEG) and EEGs with nonlinear AEPs. We examined the noise attenuation ratios to evaluate the noise variation for the calculations of superimposed averaging and cross-correlation, respectively, which constitutes the main process in the deconvolution method using the dataset of spontaneous EEGs to simulate the cases of different orders (order 5 to 12) of m-sequences. And an experiment using m-sequences of order 7 and 9 was performed in true cases with substantial linear and nonlinear AEPs. The results demonstrate that the noise attenuation ratio is well agreed with the theoretical value derived from the properties of m-sequences on the random noise condition. The comparison of waveforms for AEP components from two m-sequences showed high similarity suggesting the insensitivity of AEP to the m-sequence order. This study provides a more comprehensive solution to the selection of m-sequences which will facilitate the feasible application on the nonlinear AEP with m-sequence method.
The existing epilepsy seizure detection algorithms have problems such as overfitting and poor generalization ability due to high reliance on manual labeling of electroencephalogram’s data and data imbalance between seizure and interictal periods. An unsupervised learning detection method for epileptic seizure that jointed graph attention network (GAT) and Transformer framework (GAT-T) was proposed. In this method, channel correlations were adaptively learned by GAT encoder. Temporal information was captured by one-dimensional convolution decoder. Combining outputs of the two mentioned above, predicted values for electroencephalogram were generated. The collective anomaly score was calculated and the detection threshold was determined. The results demonstrated that GAT-T achieved the average performance exceeding 90% (or 99%) with a 0.25 s (or 2 s) time segment length, which could effectively detect epileptic seizures. Moreover, the channel association probability matrix was expected to assist clinicians in the initial screening of the epileptogenic zone, and ablation experiments also reflected the significance of each module in GAT-T. This study may assist clinicians in making more accurate diagnostic and therapeutic decisions for epilepsy patients.