Localization of Epileptic Zone Based on Adptive Dynamic Cortical Functional Networks
Abstract number :
2.187
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2024
Submission ID :
1193
Source :
www.aesnet.org
Presentation date :
12/8/2024 12:00:00 AM
Published date :
Authors :
Siqi Zhang, PhD – Harbin University of Science and Technology
Yueqian Sun, PhD – Beijing Tiantan Hospital
Yakang Dai, PhD – Suzhou Insititute of Biomedical Engineering and Technology
Qun Wang, MD – Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, China
Presenting Author: Yan Liu, PhD – Northeast Forest University
Rationale: Epileptic zone localization is important for drug-resistant epilepsy surgical treatment. The traditional epileptic zone localization methods based on the abnormal discharge center of cortex is gradually evolving to the methods based on the abnormal topological characteristics of cortical functional networks of the whole brain. Among them, the dynamic cortical functional networks constructed by combining scalp electroencephalography (EEG) of seizure with personalized brain magnetic resonance imaging can display the spatio temporal dynamic changes of epileptic electrical activity with high spatio-temporal resolution, and capture the key regions leading to seizures, which has important research significance in the localization of epileptic zone.
Methods: An accurate epileptic zone localization method based on adaptive dynamic cortical functional networks analysis was proposed, based on 10-seconds 21-channel EEG on seizures and 1-mm personally MR. Firstly, a dynamic cortical functional network conforming to the law of spontaneous switching of brain functions was constructed on an adaptive time scale based on the theory of EEG microstates. Then, the virtual seizure state, virtual seizure preparation state and virtual seizure similarity state were calculated based on the topological characteristics of the constructed cortical functional networks with the technique of source imaging with EEG and MR. Finally, epileptic zone was determined according to the changes of node strength of the three states. For validation of our method, we compared the lcoalization errors from our method and those from the method computing dynamic network with fixed window length (1s and 0.1s) and from source imaging (without network).
Results: Validation was performed based on the clinical data of 5 patients with drug-resistant epilepsy from Beijing Tiantan Hospital. First, with our proposed adptive dynamic network, 2 patients with localization errors of 7.5972 mm and 6.9361 mm, and other 3 patients with localization errors of 47.3089 mm, 48.1023 mm and 45.2535 mm. The mean loclaization error was 31.0396 mm. Second, with the fixed window length dynamic network (1s-network and 0.1-network), the mean localization errors were respectively 32.3998 mm (1s-network) and 50.5107 mm (0.1s-network ). Finally, with source imaging, the mean localization error was 66.1087 mm. Therefore, the proposed adaptive dynamic network demonstrated its superiority in localizing epileptic zone.
Conclusions: This proposed adaptive dynamic network conhensively considers and excludes the influence of spontaneous brain states on cortical functional networks, and hence it improves the current situation that the influence of spontaneous brain function switching on epileptic zone location cannot be excluded due to the unclear time scale selection rules in the construction of dynamic cortical functional networks, and realizes better epileptic zone location.
Funding: This work was supported by the National Natural Science Foundation of China (62271481,82371449); the National Key R&D Program of China grant (2022YFC2503800); Jiangsu International Cooperation Project (BZ2022028).
Neurophysiology