Construction of an epileptic seizure prediction model using a semi-supervised method of generative adversarial and long short term memory network combined with Stockwell transform / 南方医科大学学报

OBJECTIVE@#To propose a semi-supervised epileptic seizure prediction model (ST-WGAN-GP-Bi-LSTM) to enhance the prediction performance by improving time-frequency analysis of electroencephalogram (EEG) signals, enhancing the stability of the unsupervised feature learning model and improving the design of back-end classifier.@*METHODS@#Stockwell transform (ST) of the epileptic EEG signals was performed to locate the time-frequency information by adaptive adjustment of the resolution and retaining the absolute phase to obtain the time-frequency inputs. When there was no overlap between the generated data distribution and the real EEG data distribution, to avoid failure of feature learning due to a constant JS divergence, Wasserstein GAN was used as a feature learning model, and the cost function based on EM distance and gradient penalty strategy was adopted to constrain the unsupervised training process to allow the generation of a high-order feature extractor. A temporal prediction model was finally constructed based on a bi-directional long short term memory network (Bi-LSTM), and the classification performance was improved by obtaining the temporal correlation between high-order time-frequency features. The CHB-MIT scalp EEG dataset was used to validate the proposed patient-specific seizure prediction method.@*RESULTS@#The AUC, sensitivity, and specificity of the proposed method reached 90.40%, 83.62%, and 86.69%, respectively. Compared with the existing semi-supervised methods, the propose method improved the original performance by 17.77%, 15.41%, and 53.66%. The performance of this method was comparable to that of a supervised prediction model based on CNN.@*CONCLUSION@#The utilization of ST, WGAN-GP, and Bi-LSTM effectively improves the prediction performance of the semi-supervised deep learning model, which can be used for optimization of unsupervised feature extraction in epileptic seizure prediction.

The earth mover's distance and Bayesian linear discriminant analysis for epileptic seizure detection in scalp EEG

Since epileptic seizure is unpredictable and paroxysmal, an automatic system for seizure detecting could be of great significance and assistance to patients and medical staff. In this paper, a novel method is proposed for multichannel patient-specific seizure detection applying the earth mover's distance (EMD) in scalp EEG. Firstly, the wavelet decomposition is executed to the original EEGs with five scales, the scale 3, 4 and 5 are selected and transformed into histograms and afterwards the distances between histograms in pairs are computed applying the earth mover's distance as effective features. Then, the EMD features are sent to the classifier based on the Bayesian linear discriminant analysis (BLDA) for classification, and an efficient postprocessing procedure is applied to improve the detection system precision, finally. To evaluate the performance of the proposed method, the CHB-MIT scalp EEG database with 958 h EEG recordings from 23 epileptic patients is used and a relatively satisfactory detection rate is achieved with the average sensitivity of 95.65% and false detection rate of 0.68/h. The good performance of this algorithm indicates the potential application for seizure monitoring in clinical practice.

Ictal Scalp EEG in Mesial Temporal Lobe Epilepsy Patients with Unilateral versus Bilateral Interictal Epileptiform Discharges

BACKGROUND: Interictal and ictal scalp EEG is the most often used and relied-on method of noninvasive presurgical evaluation in temporal lobe epilepsy. We attempted to compare unitemporal interictal epileptiform discharges (UIED) with bitemporal interictal epileptiform discharges (BIED) groups for lateralizing value and propagation patterns of ictal scalp EEGs. METHODS: We investigated ictal scalp EEGs in 48 patients who had undergone anterior temporal lobectomies. We divided them into UIED and BIED groups by cut-off value of 90% in the laterality of IED. RESULTS: We analyzed ictal EEG patterns in 201 seizures of 32 patients with UIED and 86 seizures of 16 patients with BIED. Ictal scalp EEG was correctly lateralized significantly more often in the UIED group compared with the BIED group; 93.5% versus 57.0% in seizures and 93.8% versus 62.5% in patients, respectively. Bilateral independent onset was seen more frequently in the BIED group whereas the maintenance of ictal discharges in the unilateral temporal or hemisphere occurred more frequently in the UIED group. Lateralization by a later significant pattern (LSP) presents additional value of correct lateralization rather than pattern at onset (PAO) only. CONCLUSIONS: Interpretation of ictal scalp EEGs must be considered more cautiously in patients with BIED because BIED may be correlated with bitemporal excitability.

Predicting Factors of Long-term Surgical Outcome for Mesial Temporal Lobe Epilepsy with Unilateral Hippocampal Atrophy on MRI / 대한간질학회지

PURPOSE: To identify the clinical and electroencephalographic factors which are independently predictive of a postoperative seizure-free outcome for 4 years. We compared the outcomes of the first 2 years with the subsequent 2 years one after anterior temporal lobectomy (ATL) for mesial temporal lobe epilepsy (MTLE) with unilateral hippocampal atrophy (HA) on MRI. METHODS: We studied 51 consecutive operated patients who had above 4 years of follow-up and had MTLE with definite unilateral HA on MRI. The surgical outcome was classified as either seizure-free or not seizure-free in the first postoperative 2 years and the subsequent 2 years. Several clinical variables were included. The scalp EEG parameters included the lateralization of interictal epileptiform discharges, ictal onset location, ictal onset frequency, ictal EEG lateralization, and ictal scalp EEG propagation (bitemporal asynchrony or switch of lateralization). Variable factors were subjected to univariate analysis. RESULTS: Overall, 36 patients (71%) became seizure-free during the postoperative 4 years. On univariate analysis, only one factor was significantly associated with poor outcome (p<0.05): ictal scalp EEG propagation pattern such as bitemporal asynchrony or switch of lateralization. The seizure-free outcome was seen in 88.9% of patients without bitemporal asynchroncy, or switch of lateralization while only 54.5% of patients with those patterns (p=0.007) during the postoperative third and fourth year. However, those propagation patterns did not show the prognostic value during the first 2 years (p=0.449). Other variable factors were found not to be predictive of prognosis on early or late recurrence. CONCLUSIONS: Bitemporal asynchrony or a switch of lateralization in the ictal scalp EEG might be a highly predictive factor for an undesirable surgical outcome, late recurrence of seizure during a follow-up period after ATL, and probably an index of bitemporal epileptogenicity in MTLE.

Effect of Combination of Ictal Scalp EEG and Seizure Semiology on Seizure Lateralization in Temporal Lobe Epilepsy / 대한간질학회지

PURPOSE: To compare the reliability of lateralization between seizure semiology and ictal scalp EEG findings in mesial temporal lobe epilepsy (MTLE) patients, and to examine the advantage of the combined use of these two methods. METHODS: We independently reviewed the ictal scalp EEG recordings and clinical seizure semiology of 243 seizures recorded in 58 consecutive MTLE patients. All patients were seizure-free for at least 1 year postoperatively. Each seizure was lateralized on the basis of ictal semiology and ictal scalp EEG patterns according to strictly defined criteria, respectively. Individual patients were also lateralized based on these data. RESULTS: Seizure semiology analysis lateralized 64.6 % of seizures and 82.8 % of patients. Ictal scalp EEG analysis lateralized 74.5% of seizures and 74.1% of patients. Combination of the information from the two methods allowed for lateralization in a greater portion of both seizures (79.8%) and patients (89.7%). CONCLUSION: This study suggests that combination of ictal scalp EEG findings and seizure semiology improves the lateralization of individual seizures and patients. Therefore, it is worth lateralizing with standardized combined ictal EEG and semiology analysis for noninvasive presurgical evaluation in TLE patients.

Interictal scalp EEG in intractable temporal lobe epilepsy

BACKGROUNDS: The localization validity of interictal scalp EEG abnormalities in the patients with temporal lobe epilepsy(TLE) has been a subject of much debate. This study was an attempt to assess the general characteristics and the lateralization value of interictal scalp EEG abnormalities of TLE. We also examined the possible etiologic factors of bitemporal independent epileptiform discharges(BIED). METHODS: We investigated the interictal scalp EEGs of 59 patients. All patients have been seizure free but might have had persistent auras (Engel's classification, class I) after anterior temporal lobectomy(ATL) with minimum follow-up of 1 year. Long term scalp EEG monitoring records were evaluated for interictal EEG abnormalities in all patients. RESULTS: Scalp EEGs from 59 patients exhibited clear epileptiform discharges. Strictly unitemporal epileptiform discharge(UED) was present in 23 patients (39%). It was concordant with the side of seizure origin in 22 patients (95%), and discordant in 1 patient (5%). Thirty six patients (61%) had BIED. Twenty two patients with BIED showed lateralized preponderance which was defined as at least 80% laterality. It was concordant with the side of seizure origin in 21 patients (95%), and discordant in 1 patient (5%), and 14 patients were not lateralized. Bilateral synchronous epileptiform discharges were present in 4 patients (7%), and extratemporal spike in only 1 patient (2%). Localized temporal slow waves were shown in 33 patients (56%). It was concordant with the side of seizure origin in 28 patients, discordant in 1 patient. Four of 33 patients had bilateral temporal slow waves. There was no statistically significant difference in age at seizure onset, duration of epilepsy between the groups with UED and with BIED. Febrile seizures occurred similarly in both groups. CONCLUSIONS: It seems obvious that patients with UED or lateralized interictal temporal spike or sharp waves have a strong likelihood of ictal onset from the ipsilateral temporal region.