Page 206 - MEGIN Book Of Abstracts - 2023

Genoa, Italy; Neuroscience Center, HiLIFE-Helsinki Institute of results provide evidence for interareal CFS and PAC
Life Science, University of Helsinki, Finland; Claudio Munari being 2 distinct mechanisms for coupling oscillations
Epilepsy Surgery Centre, Niguarda Hospital, Italy; Centre for across frequencies in large-scale brain networks.
Cognitive Neuroimaging, Institute of Neuroscience & Psychol-
ogy, University of Glasgow, United Kingdom PLoS biology (2020), Vol. 18, No. 5 (32374723) (25
citations)
ABSTRACT Phase synchronization of neuronal oscil-
lations in specific frequency bands coordinates ana-
tomically distributed neuronal processing and com- A Combination of Particle Swarm Optimization
munication. Typically, oscillations and synchronization and Minkowski Weighted K-Means Clustering:
take place concurrently in many distinct frequencies, Application in Lateralization of Temporal Lobe
which serve separate computational roles in cognitive Epilepsy (2020)
functions. While within-frequency phase synchroniza-
tion has been studied extensively, less is known about Jamali-Dinan, Samira-Sadat; Soltanian-Zadeh, Hamid;
the mechanisms that govern neuronal processing Bowyer, Susan M; Almohri, Haidar; Dehghani, Hamed;
distributed across frequencies and brain regions. Such Elisevich, Kost; Nazem-Zadeh, Mohammad-Reza
integration of processing between frequencies could
be achieved via cross-frequency coupling (CFC), either Department of Mathematics and Computer Science, Amir
by phase-amplitude coupling (PAC) or by n:m-cross- Kabir University of Technology, Tehran, Iran; Research Ad-
frequency phase synchrony (CFS). So far, studies have ministration, Radiology, Henry Ford Health System, Detroit,
mostly focused on local CFC in individual brain regions, MI, 48202, USA; Neurology Departments, Henry Ford Health
whereas the presence and functional organization System, Detroit, MI, 48202, USA; Department of Industrial
of CFC between brain areas have remained largely and Systems Engineering, Wayne State University, Detroit, MI,
unknown. We posit that interareal CFC may be essential USA; Medical Physics, and Biomedical Engineering Depart-
for large-scale coordination of neuronal activity and ment, Tehran University of Medical Sciences (TUMS), Tehran,
investigate here whether genuine CFC networks are Iran; Department of Clinical Neurosciences, Spectrum Health,
present in human resting-state (RS) brain activity. To College of Human Medicine, Michigan State University,
assess the functional organization of CFC networks, we Grand Rapids, MI, 49503, USA; Research Center for Molecular
identified brain-wide CFC networks at mesoscale reso- and Cellular Imaging, Research Center for Science and Tech-
lution from stereoelectroencephalography (SEEG) and nology in Medicine, Tehran University of Medical Sciences
at macroscale resolution from source-reconstructed (TUMS), Tehran, Iran. [email protected]
magnetoencephalography (MEG) data. We developed
a novel, to our knowledge, graph-theoretical method ABSTRACT K-Means is one of the most popular clus-
to distinguish genuine CFC from spurious CFC that may tering algorithms that partitions observations into
arise from nonsinusoidal signals ubiquitous in neuronal nonoverlapping subgroups based on a predefined
activity. We show that genuine interareal CFC is present similarity metric. Its drawbacks include a sensitivity
in human RS activity in both SEEG and MEG data. Both to noisy features and a dependency of its resulting
CFS and PAC networks coupled theta and alpha oscil- clusters upon the initial selection of cluster centroids
lations with higher frequencies in large-scale networks resulting in the algorithm converging to local optima.
connecting anterior and posterior brain regions. CFS Minkowski weighted K-Means (MWK-Means) addresses
and PAC networks had distinct spectral patterns and the issue of sensitivity to noisy features, but is sensitive
opposing distribution of low- and high-frequency to the initialization of clusters, and so the algorithm
network hubs, implying that they constitute distinct may similarly converge to local optima. Particle Swarm
CFC mechanisms. The strength of CFS networks was Optimization (PSO) uses a globalized search method to
also predictive of cognitive performance in a separate solve this issue. We present a hybrid Particle Swarm Op-
neuropsychological assessment. In conclusion, these timization (PSO) + MWK-Means clustering algorithm to

 ontents  Index 185
C