Unsupervised Graph Neural Network Reveals Spatiotemporal Subtypes in Focal Epilepsy
Poster No:
1354
Submission Type:
Late-Breaking Abstract Submission
Authors:
Yafei Liu1, Ke Dong2, Xueqi Wei1, Yue Zhang1, Zhongyi Xie3, Tianjun Wang1, Chunyan Cao4, Limin Sun1
Institutions:
1Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai, Shanghai, 2Shanghai University, shanghai, Shanghai, 3China Medical University, Shenyang, Liaoning, 4Ruijin Hospital, Shanghai, China
First Author:
Yafei Liu
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Co-Author(s):
Xueqi Wei
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Yue Zhang
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Tianjun Wang
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Limin Sun
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
Shanghai, Shanghai
Late Breaking Reviewer(s):
Rosanna Olsen
Rotman Research Institute, Baycrest Academy for Research and Education
Toronto, Ontario
Rotman Research Institute, Baycrest Academy for Research and Education
Toronto, Ontario
Introduction:
Interictal epileptiform discharges (IEDs) are critical biomarkers for diagnosing, monitoring and treatment of focal epilepsy[1,2]. However, despite their importance, the classification of IEDs and their role in revealing the underlying spatial-temporal mechanisms of epilepsy remains largely unexplored[3,4].
Methods:
This study introduces 4 unsupervised graph embedding models - DeepWalk, Node2Vec, graph convolutional network and graph attention-based network- combined with 2 clustering methods (k-means and sepctural clustering) to identify IEDs subtypes in source-level magnetoencephalography (MEG) data [5]. 55 patients with focal epilepsy were enrolled in this study. Resting-state MEG data and T1-weighted MRI were collected through the clinical routine examination for each subject. 8059 IEDs were identified and extracted manually. The source-level networks for each IED were computed by using 3-layer BEM head model and phase locking values. Global and nodal level network features were represented by graph theory[6], including degree, path length, clustering coefficient and efficiency[7,8].
·Fig. 1 General data processing
Results:
Unsupervised clustering analysis identified four IED subtypes based on source-level MEG networks using the graph attention-based network with k-means algorithm, supported by optimal clustering indices (highest Silhouette Score and Calinski-Harabasz Index, lowest Davies-Bouldin Index). Subtype-specific differences in global and nodal network efficiency were observed: Subtype 3 exhibited the highest global efficiency, Subtype 2 the lowest, and Subtype 1 outperformed Subtype 4. Nodal analysis identified unique hub regions for each subtype: Subtype 1 was characterized by the left temporal transverse gyrus and the left posterior dorsal cingulate gyrus; Subtype 2 by the left middle posterior cingulate gyrus and the left orbital H-shaped sulcus; and Subtype 3 by the bilateral subcallosal gyri. Subtype 4 lacked distinctive nodal features. Furthermore, subtype frequency correlated with cortical thickness and age, revealing significant age-related patterns.
·Fig. 2 Unique nodes for each subtype
Conclusions:
This study demonstrates that leveraging the spatiotemporal dynamics of IEDs through unsupervised clustering enhances the understanding of shared network features across focal epilepsy types. Our findings provide a foundation for developing treatment strategies and highlight the potential of IED classification in advancing epilepsy research and clinical practice.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Modeling and Analysis Methods:
EEG/MEG Modeling and Analysis 1
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Anatomy and Functional Systems
Novel Imaging Acquisition Methods:
MEG
Keywords:
Data analysis
Epilepsy
MEG
Other - Graph Neural Network; Interictal Epileptiform Discharges
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
2.Ung, H.,(2017). Interictal epileptiform activity outside the seizure onset zone impacts cognition. Brain, 140(8), 2157–2168.
3.Royer, J.,(2022). Epilepsy and brain network hubs. Epilepsia, 63(3), 537–550.
4.Conrad, E. C.,(2020). Spatial distribution of interictal spikes fluctuates over time and localizes seizure onset. Brain, 143(2), 554–569.
5.Abadal, S.,(2025). Graph neural networks for electroencephalogram analysis: Alzheimer’s disease and epilepsy use cases. Neural Networks, 181, 106792.
6.Vetkas, A.,(2022). Identifying the neural network for neuromodulation in epilepsy through connectomics and graphs. Brain Communications, 4(3), fcac092.
7.Liu, Y.,(2023). Neuromorphic transcutaneous electrical nerve stimulation (nTENS) induces efficient tactile-related cortical networks in forearm amputees. Science China Technological Sciences, 66(5), 1451–1460.
8.Liu, Y., (2022). Effect of neuromorphic transcutaneous electrical nerve stimulation (nTENS) of cortical functional networks on tactile perceptions: An event-related electroencephalogram study. Journal of Neural Engineering, 19(2), 026017.
9.Finn, E. S.,(2023). Functional neuroimaging as a catalyst for integrated neuroscience. Nature, 623(7986), 263–273.
10.Kural, M. A.,(2020). Criteria for defining interictal epileptiform discharges in EEG. Neurology, 94(20), e2139–e2147.