Page 216 - MEGIN Book Of Abstracts - 2023

sensory evoked field(s) (SEFs) could be used to predict mal neuronal discharge after an epileptic seizure. This
seizure response to VNS. Retrospective data from forty- abnormal activity can originate from one or more cra-
eight pediatric patients who underwent VNS at two nial lobes, often travels from one lobe to another, and
different institutions were used in this study. Thirty-six interferes with normal activity from the affected lobe.
patients ("Discovery Cohort") underwent preoperative The common practice for Inter-ictal spike detection of
electrical median nerve stimulation during magneto- brain signals is via visual scanning of the recordings,
encephalography (MEG) recordings and 12 patients which is a subjective and a very time-consuming task.
("Validation Cohort") underwent preoperative pneu- Motivated by that, this article focuses on using machine
matic stimulation during MEG. SEFs and their spatial learning for epileptic spikes classification in magne-
deviation, waveform amplitude and latency, and event- toencephalography (MEG) signals. First, we used the
related connectivity were calculated for all patients. A Position Weight Matrix (PWM) method combined with
support vector machine (SVM) classifier was trained on a uniform quantizer to generate useful features from
the Discovery Cohort to differentiate responders from time domain and frequency domain through a Fast
non-responders based on these input features and test- Fourier Transform (FFT) of the framed raw MEG signals.
ed on the Validation Cohort by comparing the model- Second, the extracted features are fed to standard clas-
predicted response to VNS to the known response. We sifiers for inter-ictel spikes classification. The proposed
found that responders to VNS had significantly more technique shows great potential in spike classification
widespread SEF localization and greater functional con- and reducing the feature vector size. Specifically, the
nectivity within limbic and sensorimotor networks in proposed technique achieved average sensitivity up
response to median nerve stimulation. No difference in to 87% and specificity up to 97% using 5-folds cross-
SEF amplitude or latencies was observed between the validation applied to a balanced dataset. These samples
two cohorts. The SVM classifier demonstrated 88.9% are extracted from nine epileptic subjects using a slid-
accuracy (0.93 area under the receiver operator charac- ing frame of size 95 samples-points with a step-size of 8
teristics curve) on cross-validation, which decreased to sample-points.
67% in the Validation cohort. By leveraging overlapping
neural circuitry, we found that median nerve SEF char- IEEE journal of biomedical and health informatics (2020),
acteristics and functional connectivity could identify Vol. 24, No. 10 (32054592) (4 citations)
responders to VNS.

Keywords: Connectomics, Evoked potentials, Machine A novel method for extracting interictal
learning, SEF, VNS epileptiform discharges in multi-channel MEG: Use
of fractional type of blind source separation (2020)
NeuroImage. Clinical (2020), Vol. 26 (32070812) (13
citations) Matsubara, Teppei; Hironaga, Naruhito; Uehara, Taira;
Chatani, Hiroshi; Tobimatsu, Shozo; Kishida, Kuniharu

QuPWM: Feature Extraction Method for Epileptic Department of Clinical Neurophysiology, Neurological
Spike Classification (2020) Institute, Faculty of Medicine, Graduate School of Medical
Sciences, Kyushu University, Japan. Electronic address: tep-
Chahid, Abderrazak; Albalawi, Fahad; Alotaiby, Turky [email protected]; Emeritus of Gifu University, Japan;
Nayef; Al-Hameed, Majed Hamad; Alshebeili, Saleh; Hermitage of Magnetoencephalography, Japan
Laleg-Kirati, Taous-Meriem
OBJECTIVE Visual inspection of interictal epileptiform
ABSTRACT Epilepsy is a neurological disorder ranked discharges (IEDs) in multi-channel MEG requires a
as the second most serious neurological disease known time-consuming evaluation process and often leads to
to humanity, after stroke. Inter-ictal spiking is an abnor- inconsistent results due to variability of IED waveforms.

 ontents  Index 195
C