An evaluation of lateralizing signs in patients with temporal lobe epilepsy

Deep learning approaches for seizure video analysis: A review

Seizure events can manifest as transient disruptions in the control of movements which may be organized in distinct behavioral sequences, accompanied or not by other observable features such as altered facial expressions. The analysis of these clinical signs, referred to as semiology, is subject to observer variations when specialists evaluate video-recorded events in the clinical setting. To enhance the accuracy and consistency of evaluations, computer-aided video analysis of seizures has emerged as a natural avenue. In the field of medical applications, deep learning and computer vision approaches have driven substantial advancements. Historically, these approaches have been used for disease detection, classification, and prediction using diagnostic data; however, there has been limited exploration of their application in evaluating video-based motion detection in the clinical epileptology setting. While vision-based technologies do not aim to replace clinical expertise, they can significantly contribute to medical decision-making and patient care by providing quantitative evidence and decision support. Behavior monitoring tools offer several advantages such as providing objective information, detecting challenging-to-observe events, reducing documentation efforts, and extending assessment capabilities to areas with limited expertise. The main applications of these could be (1) improved seizure detection methods; (2) refined semiology analysis for predicting seizure type and cerebral localization. In this paper, we detail the foundation technologies used in vision-based systems in the analysis of seizure videos, highlighting their success in semiology detection and analysis, focusing on work published in the last 7 years. We systematically present these methods and indicate how the adoption of deep learning for the analysis of video recordings of seizures could be approached. Additionally, we illustrate how existing technologies can be interconnected through an integrated system for video-based semiology analysis. Each module can be customized and improved by adapting more accurate and robust deep learning approaches as these evolve. Finally, we discuss challenges and research directions for future studies.

Artificial intelligence in epilepsy phenotyping

Artificial intelligence (AI) allows data analysis and integration at an unprecedented granularity and scale. Here we review the technological advances, challenges, and future perspectives of using AI for electro-clinical phenotyping of animal models and patients with epilepsy. In translational research, AI models accurately identify behavioral states in animal models of epilepsy, allowing identification of correlations between neural activity and interictal and ictal behavior. Clinical applications of AI-based automated and semi-automated analysis of audio and video recordings of people with epilepsy, allow significant data reduction and reliable detection and classification of major motor seizures. AI models can accurately identify electrographic biomarkers of epilepsy, such as spikes, high-frequency oscillations, and seizure patterns. Integrating AI analysis of electroencephalographic, clinical, and behavioral data will contribute to optimizing therapy for patients with epilepsy.

A different look on the importance of lateralization and localization of figure 4 symptoms in epilepsy

Background

Clinical seizure semiology provides valuable information in the evaluation of focal-onset bilateral tonic–clonic seizures. In the evaluation of these patients, long-term video-EEG monitoring (VEM) and neuroimaging studies are of great significance in determining lateralization together with clinical semiology. In this study, we examined the features of the figure 4 sign that we detected in patients with refractory epilepsy whom we followed up in the VEM unit.

Methods

In the study, 175 patients followed in the VEM unit were examined. Twenty-two patients for whom the figure 4 sign was detected were included in the study. Patients with the side indicated by the figure 4 sign comply with ictal EEG compatible were named as 1st group (G1), and those not were named as 2nd group (G2). Demographic characteristics, seizure type, number of seizures per month, duration of epilepsy, number of seizures during VEM, duration of figure 4 sign, medical history, and cranial MRI characteristics were compared between the two groups.

Results

When G1 and G2 were compared, it was observed that temporal lobe seizures were statistically significantly higher in G1. When the figure 4 sign durations were examined, this period was 16.3 ± 8.2 s in frontal lobe seizures and 20.8 ± 7.4 s in temporal lobe seizures. When the duration of the figure 4 sign was examined by gender, it was found that the duration was longer in males, which was statistically significant.

Conclusion

In conclusion, it is essential to evaluate ictal EEG findings together with brain imaging while performing semiological localization and lateralization in epileptic patients.

Specific Oscillatory Power Changes and Their Efficacy for Determining Laterality in Mesial Temporal Lobe Epilepsy: A Magnetoencephalographic Study

Appropriate determination of the epileptic focus and its laterality are important for the diagnosis of mesial temporal lobe epilepsy (MTLE). The aims of this study are to establish a specific oscillatory distribution and laterality of the oscillatory power in unilateral MTLE with frequency analysis of magnetoencephalography (MEG), and to confirm their potential to carry significant information for determining lateralization of the epileptic focus. Thirty-five patients with MTLE [left (LtMTLE), 16; right (RtMTLE), 19] and 102 healthy control volunteers (CTR) were enrolled. Cortical oscillatory powers were compared among the groups by contrasting the source images using a one-way ANOVA model for each frequency band. Further, to compare the lateralization of regional oscillatory powers between LtMTLEs and RtMTLEs, the laterality index (LI) was calculated for four brain regions (frontal, temporal, parietal, and occipital) in each frequency band, which were compared between patient groups (LtMTLE, RtMTLE, and CTR), and used for machine learning prediction of the groups with support vector machine (SVM) with linear kernel function. Significant oscillatory power differences between MTLE and CTR were found in certain areas. In the theta to high-frequency oscillation bands, there were marked increases in the parietal lobe, especially on the left side, in LtMTLE. In the theta, alpha, and high-gamma bands, there were marked increases in the parietal lobe, especially on the right side in RtMTLE. Compared with CTR, LIs were significantly higher in 24/28 regions in LtMTLE, but lower in 4/28 regions and higher in 10/28 regions in RtMTLE. LI at the temporal lobe in the theta band was significantly higher in LtMTLE and significantly lower in RtMTLE. Comparing LtMTLE and RtMTLE, there were significant LI differences in most regions and frequencies (21/28 regions). In all frequency bands, there were significant differences between LtMTLE and RtMTLE in the temporal and parietal lobes. The leave-one-out cross-validation of the linear-SVM showed the classification accuracy to be over 91%, where the model had high specificity over 96% and mild sensitivity ~68–75%. Using MEG frequency analysis, the characteristics of the oscillatory power distribution in the MTLE were demonstrated. Compared with CTR, LIs shifted to the side of the epileptic focus in the temporal lobe in the theta band. The machine learning approach also confirmed that LIs have significant predictive ability in the lateralization of the epileptic focus. These results provide useful additional information for determining the laterality of the focus.

The Preoperative Evaluation of Drug-Resistant Epilepsy

Antiepileptic drugs afford good seizure control for approximately 70% of individuals with epilepsy. Epilepsy surgery is extremely helpful for appropriate individuals with drug resistance. Since antiquity, trephination was a crude and invasive technique to manage epilepsy. The late 1800s saw the advent of a more evidence-based approach with attempts to define seizure foci and determine areas of function. Seizure localization initially required direct brain stimulation during surgery before resection. Fortunately, improved knowledge of seizure semiology and advancements in preoperative investigations have enabled epilepsy specialists to better analyze the benefit of seizure reduction versus risk of functional harm. This preoperative phase and the investigative techniques used to analyze surgical candidacy will be discussed in this article.

Hand posture as localizing sign in adult focal epileptic seizures

OBJECTIVE

The aim of this study was to identify specific ictal hand postures (HPs) as localizing signs of the epileptogenic zone (EZ) in patients with frontal or temporal lobe epilepsy.

METHODS

In this study, we retrospectively analyzed ictal semiology of 489 temporal lobe or frontal lobe seizures recorded over a 6-year period at the Seizure Disorder Center at University of California, Los Angeles in the USA (45 patients) or at the C. Munari Epilepsy Surgery Center at Niguarda Hospital in Milan, Italy (34 patients). Our criterion for EZ localization was at least 2 years of seizure freedom after surgery. We analyzed presence and latency of ictal HP. We then examined whether specific initial HPs are predictive for EZ localization.

RESULTS

We found that ictal HPs were present in 72.5% of patients with frontal and 54.5% of patients with temporal lobe seizures. We divided HPs into 6 classes depending on the reciprocal position of the fingers ("fist," "cup," "politician's fist," "pincer," "extended hand," "pointing"). We found a striking correlation between EZ localization and ictal HP. In particular, fist and pointing HPs are strongly predictive of frontal lobe EZ; cup, politician's fist, and pincer are strongly predictive of temporal lobe EZ.

INTERPRETATION

Our study offers simple ictal signs that appear to clarify differential diagnosis of temporal versus frontal lobe EZ localization. These results are meant to be used as a novel complementary tool during presurgical evaluation for epilepsy. At the same time, they give us important insight into the neurophysiology of hand movements. ANN NEUROL 2019;86:793-800.

Voluntary and involuntary movements: A proposal from a clinician

In this communication, I first summarize the mechanisms underlying human voluntary movements and define the involuntary movements (medical term).

CLASSIFICATION OF HUMAN MOVEMENTS

Human movements are classified into two main kinds: intentional movements and non-intentional movements in which the involuntary movements are included. Non-intentional movements have many kinds of movement: normal non-intentional movements (associate movements, mirror movements or juggling knees etc.), several reflexes (spinal tendon, spinal flexion, spino-bulbo-spinal, cortical reflexes and startle response) and pathological non-intentional movements which should be treated (so called "involuntary movements" in clinical practice, medical term of involuntary movement).

VOLUNTARY MOVEMENTS

The final motor commands for movements are mediated by several descending motor pathways. These final pathways are modified, regulated by two main loops (basal ganglia loop and cerebellar loop).

INVOLUNTARY MOVEMENTS (MEDICAL TERM)

The involuntary movements are produced by a non-intentional, pathological activation anywhere within the final common pathways or the above two loops. I would like to personally divide those into four major groups.

TREMOR

Some oscillation mechanisms may produce tremor: one site oscillation or loop oscillation.

MYOCLONUS

Sudden, brief, shock-like involuntary movements arising from anywhere from the cortex to the muscle.

CHOREA/BALLISM

Suddenly appearing, irregular, phasic movements which are usually mimicked by normal subjects.

DYSTONIA/ATHETOSIS

Sustained, long duration muscle contraction sometimes associated with torsion components.

Prognostic factors of postoperative seizure outcome in patients with temporal lobe epilepsy and normal magnetic resonance imaging

PURPOSE

To retrospectively analyse a single-centre consecutive surgical series of patients with temporal lobe epilepsy (TLE) and negative MRI. To identify factors associated with postoperative seizure outcome among several presurgical, surgical and postsurgical variables.

METHODS

Clinical records of 866 patients who received temporal lobe resections and with a minimum follow-up of 12 months were retrospectively searched for MRI-negative cases. Anamnestic, clinical, neurophysiological, surgical, histopathological and postsurgical data were collected. Seizure outcome was categorised as favourable (Engel's class I) and unfavourable (Engel's classes II-IV). Uni- and multivariate statistical analysis was performed to identify variables having a significant association with seizure outcome.

RESULTS

Forty-eight patients matched the inclusion criteria. 26 (54.1%) patients required invasive EEG evaluation with Stereo-electro-encephalography (SEEG) before surgery. Histological evaluation was unremarkable in 34 cases (70.8%), revealed focal cortical dysplasias in 13 cases and hippocampal sclerosis in 2. 28 (58.3%) patients were in Engel's class I after a mean follow-up of 82 months (SD ± 74; range 12-252). Multivariate analysis indicated auditory aura, contralateral diffusion of the discharge at Video-EEG monitoring and use of ¹⁸F-FDG PET as variables independently associated with seizure outcome.

CONCLUSION

Carefully selected patients with MRI-negative TLE can be good candidates for surgery. Surgery should be considered with caution in patients with clinical features of neocortical seizure onset and contralateral propagation of the discharge. Use of ¹⁸F-FDG PET may be helpful to improve SEEG and surgical strategies. The presented data help in optimising the selection of patients with MRI-negative TLE with good chances to benefit from surgery.

Deep facial analysis: A new phase I epilepsy evaluation using computer vision

Semiology observation and characterization play a major role in the presurgical evaluation of epilepsy. However, the interpretation of patient movements has subjective and intrinsic challenges. In this paper, we develop approaches to attempt to automatically extract and classify semiological patterns from facial expressions. We address limitations of existing computer-based analytical approaches of epilepsy monitoring, where facial movements have largely been ignored. This is an area that has seen limited advances in the literature. Inspired by recent advances in deep learning, we propose two deep learning models, landmark-based and region-based, to quantitatively identify changes in facial semiology in patients with mesial temporal lobe epilepsy (MTLE) from spontaneous expressions during phase I monitoring. A dataset has been collected from the Mater Advanced Epilepsy Unit (Brisbane, Australia) and is used to evaluate our proposed approach. Our experiments show that a landmark-based approach achieves promising results in analyzing facial semiology, where movements can be effectively marked and tracked when there is a frontal face on visualization. However, the region-based counterpart with spatiotemporal features achieves more accurate results when confronted with extreme head positions. A multifold cross-validation of the region-based approach exhibited an average test accuracy of 95.19% and an average AUC of 0.98 of the ROC curve. Conversely, a leave-one-subject-out cross-validation scheme for the same approach reveals a reduction in accuracy for the model as it is affected by data limitations and achieves an average test accuracy of 50.85%. Overall, the proposed deep learning models have shown promise in quantifying ictal facial movements in patients with MTLE. In turn, this may serve to enhance the automated presurgical epilepsy evaluation by allowing for standardization, mitigating bias, and assessing key features. The computer-aided diagnosis may help to support clinical decision-making and prevent erroneous localization and surgery.

Genital automatisms: Reappraisal of a remarkable but ignored symptom of focal seizures

OBJECTIVES

Genital automatisms (GAs) are uncommon clinical phenomena of focal seizures. They are defined as repeated fondling, grabbing, or scratching of the genitals. The aim of this study was to determine the lateralizing and localizing value and associated clinical characteristics of GAs.

METHODS

Three hundred thirteen consecutive patients with drug-resistant seizures who were referred to our tertiary center for presurgical evaluation between 2009 and 2016 were investigated. The incidence of specific kinds of behavior, clinical semiology, associated symptoms/signs with corresponding ictal electroencephalography (EEG) findings, and their potential role in seizure localization and lateralization were evaluated.

RESULTS

Fifteen (4.8%) of 313 patients had GAs. Genital automatisms were identified in 19 (16.4%) of a total 116 seizures. Genital automatisms were observed to occur more often in men than in women (M/F: 10/5). Nine of fifteen patients (60%) had temporal lobe epilepsy (right/left: 4/5) and three (20%) had frontal lobe epilepsy (right/left: 1/2), whereas the remaining two patients could not be classified. One patient was diagnosed as having Rasmussen encephalitis. Genital automatisms were ipsilateral to epileptic focus in 12 patients and contralateral in only one patient according to ictal-interictal EEG and neuroimaging findings. Epileptic focus could not be lateralized in the last 2 patients. Genital automatisms were associated with unilateral hand automatisms such as postictal nose wiping or manual automatisms in 13 (86.7%) of 15 and contralateral dystonia was seen in 6 patients. All patients had amnesia of the performance of GAs.

CONCLUSION

Genital automatisms are more frequent in seizures originating from the temporal lobe, and they can also be seen in frontal lobe seizures. Genital automatisms seem to have a high lateralizing value to the ipsilateral hemisphere and are mostly concordant with other unilateral hand automatisms. Men exhibit GAs more often than women.