Page 197 - MEGIN Book Of Abstracts - 2023

not rely on any subjective channel selection and thus by two experts. The experimental results show that our
contribute towards making source localization more new detector outperforms multiple traditional machine
unbiased and automatic. We show that the two dipolar learning models. In particular, our method can achieve
methods, SESAME and RAP-MUSIC, generally agree a mean accuracy of 89.3% and an average area under
with dipole fitting in terms of identified cerebral lobes the receiver operating characteristic curve of 0.88 in 50
and that the results of the former are closer to the fitted repeats of random subsampling validation. In addition,
equivalent current dipoles than those of the latter. In we experimentally demonstrate the effectiveness of
addition, for all the tested methods and particularly virtual sample generation, attention mechanism, and
for SESAME, concordance with surgical plan is a good architecture of neural network models.
predictor of seizure freedom while discordance is not
a good predictor of poor post-surgical outcome. The IEEE transactions on neural systems and rehabilitation
results suggest that the dipolar methods, especially engineering: a publication of the IEEE Engineering
SESAME, represent a reliable and more objective alter- in Medicine and Biology Society (2020), Vol. 28, No. 8
native to manual dipole fitting for clinical applications (32746301) (5 citations)
in the field of epilepsy surgery.

Keywords: Bayesian methods, Dipole modeling, Epilepsy, Temporal-plus epilepsy in children: A connectomic
Magnetic source imaging, Magnetoencephalography analysis in magnetoencephalography (2020)

Brain topography (2020), Vol. 33, No. 5 (32770321) (4 Martire, Daniel J; Wong, Simeon; Workewych, Adriana;
citations) Pang, Elizabeth; Boutros, Sarah; Smith, Mary Lou; Ochi,
Ayako; Otsubo, Hiroshi; Sharma, Roy; Widjaja, Elysa;
Snead, O Carter; Donner, Elizabeth; Ibrahim, George M
Automatic and Accurate Epilepsy Ripple and Fast
Ripple Detection via Virtual Sample Generation and Program in Neuroscience and Mental Health, Hospital for
Attention Neural Networks (2020) Sick Children Research Institute, Toronto, Ontario, Canada;
Institute of Biomaterials and Biomedical Engineering,
Guo, Jiayang; Li, Hailong; Pan, Yijie; Gao, Yuan; Sun, University of Toronto, Toronto, Ontario, Canada; Division
Jintao; Wu, Ting; Xiang, Jing; Luo, Xiongbiao of Neurology, Hospital for Sick Children, Toronto, Ontario,
Canada; Division of Psychology, Hospital for Sick Children,
ABSTRACT About 1% of the population around the University of Toronto, Toronto, Ontario, Canada; Department
world suffers from epilepsy. The success of epilepsy of Diagnostic Imaging, Hospital for Sick Children, Toronto,
surgery depends critically on pre-operative localization Ontario, Canada; Institute of Medical Science, University of
of epileptogenic zones. High frequency oscillations Toronto, Toronto, Ontario, Canada; Division of Neurosurgery,
including ripples (80-250 Hz) and fast ripples (250- Department of Surgery, Hospital for Sick Children, University
500 Hz) are commonly used as biomarkers to localize of Toronto, Toronto, Ontario, Canada
epileptogenic zones. Recent literature demonstrated
that fast ripples indicate epileptogenic zones better OBJECTIVE Seizure recurrence following surgery for
than ripples. Thus, it is crucial to accurately detect fast temporal lobe (TL) epilepsy may be related to extra-
ripples from ripples signals of magnetoencephalog- temporal epileptogenic foci, so-called temporal-plus
raphy for improving outcome of epilepsy surgery. (TL+) epilepsy. Here, we sought to leverage whole brain
This paper proposes an automatic and accurate ripple connectomic profiling in magnetoencephalography
and fast ripple detection method that employs virtual (MEG) to identify neural networks indicative of TL+
sample generation and neural networks with an atten- epilepsy in children.
tion mechanism. We evaluate our proposed detector on
patient data with 50 ripples and 50 fast ripples labeled

 ontents  Index 176
C