TIE-EEGNet: Temporal Information Enhanced EEGNet for Seizure Subtype Classification.

Electroencephalogram (EEG) based seizure subtype classification is very important in clinical diagnostics. However, manual seizure subtype classification is expensive and time-consuming, whereas automatic classification usually needs a large number of labeled samples for model training. This paper proposes an EEGNet-based slim deep neural network, which relieves the labeled data requirement in EEG-based seizure subtype classification. A temporal information enhancement module with sinusoidal encoding is used to augment the first convolution layer of EEGNet. A training strategy for automatic hyper-parameter selection is also proposed. Experiments on the public TUSZ dataset and our own CHSZ dataset with infants and children demonstrated that our proposed TIE-EEGNet outperformed several traditional and deep learning models in cross-subject seizure subtype classification. Additionally, it also achieved the best performance in a challenging transfer learning scenario. Both our code and the CHSZ dataset are publicized.

[Role of nerve stimulation at Erb point in early diagnosis of Guillain-Barré syndrome in children].

OBJECTIVE: To study the role of proximal nerve stimulation at Erb point in the early diagnosis of Guillain-Barré syndrome (GBS) in children. METHODS: Thirty-two children who were diagnosed with GBS between October 2013 and December 2014 received neurophysiological examination. Thirty healthy children were used as controls. Compound muscle action potentials and distal motor latency of the median and ulnar nerves were determined and analyzed after nerve stimulation at the wrist, elbow, and Erb point in the two groups. Moreover, F-wave latency of the median nerve and H-reflex latency of the tibial nerve were measured and analyzed in the two groups. RESULTS: The F-wave and H-reflex latencies were significantly longer in the patient group than in the control group (P<0.05). In thirty-two patients, the numbers of patients with abnormal amplitude, abnormal latency, and conduction block at Erb's point were 24 (75%), 22 (69%), and 20 (62%), respectively. The patient group had significantly lower amplitudes but significantly longer latencies of the ulnar and median nerves at Erb point than the control group (P<0.05). There were no significant differences in the amplitudes and latencies at the wrist and elbow between the two groups (P>0.05). CONCLUSIONS: The nerve stimulation at Erb point holds promise as a routine examination for the early diagnosis of GBS.

[A study of brain-computer interface paradigm based on mental arithmetic].

In the traditional P300 brain-computer interface (BCI) system, the electroencephalogram (EEG) signals can only provide limited information with a low signal-to-noise ratio. A BCI paradigm under visual stimulus was proposed in our study aiming to effectively activate the related brain areas and response signal while dealing with specific cognitive task (mental arithmetic task), so as to enhance the EEG signals. The result was compared with the traditional P300 counting task paradigm. Then the collected EEG data were preprocessed including extracting signal features with coherent averaging method, and analyzing the influences of different experimental paradigms on main components of event related potential (ERP). In the improved paradigm experiments the average increasing rate of P300 amplitude was 6. 83MV (73. 94%). The brain activity from 400ms was more active and lasted longer. Besides, unlike traditional counting task, mental arithmetic task appeared to have apparent activation at 650ms. The results showed that the improved paradigm could activate the related brain areas better and enhance the characteristics of signal. This provides a new system paradigm for BCI.