Persistent homology to analyse disruptions of functional and effective brain connectivity

0568775 - ÚI 2023 RIV IT eng A - Abstrakt
Hlinka, Jaroslav - Pidnebesna, Anna - Caputi, Luigi
Persistent homology to analyse disruptions of functional and effective brain connectivity.
The 11th International Conference on Complex Networks and their Applications - Book of Abstracts. Palermo: 11th International Conference on Complex Networks and their Applications, 2022. s. 513-514. ISBN 978-2-9557050-6-3.
[COMPLEX NETWORKS 2022: The International Conference on Complex Networks and their Applications /11./. 08.11.2022-10.11.2022, Palermo]
Grant CEP: GA ČR(CZ) GA21-17211S
Institucionální podpora: RVO:67985807
Klíčová slova: Persistent homology * Connectivity * fMRI * Electrophysiology * Epilepsy * Schizophrenia
Obor OECD: Neurosciences (including psychophysiology
https://dx.doi.org/10.5072/zenodo.1154242

ZÁKLADNÍ ÚDAJE: The 11th International Conference on Complex Networks and their Applications - Book of Abstracts. Palermo: 12th International Conference on Complex Networks and their Applications, 2023. s. 513-514. ISBN 978-2-9557050-6-3. KONFERENCE: COMPLEX NETWORKS 2023: The 12th International Conference on Complex Networks and their Applications. 28.11.2023-30.11.2023, French Riviera]. ABSTRAKT: Topological Data Analysis (TDA [1]), despite its relative novelty, has already been applied to study network connectivity structure across fields. We propose that its prominent tool of persistent homology (PH) may apart from the more common dependence networks (functional connectivity – FC) be applied also to directed, causal, networks – known as effective connectivity (EC) in neuroscience. We test the PH discriminatory power in two archetypal examples of disease-related brain connectivity alterations: during epilepsy seizures (captured by electrophysiology – EEG) and in schizophrenia patients (using functional magnetic resonance imaging - fMRI). We employ a range of PH-based features and quantify ability to distinguish healthy from diseased brain states by applying a support vector machine (SVM), a relatively standard method of choice for similar data situations, used also previously in similar context. We compare this novel approach to using standard undirected PH applied to the functional connectivity matrix, as well as comparing the (D)PH approach to using the raw EC/FC matrices [2]
Trvalý link: https://hdl.handle.net/11104/0340035