Virtual brain twin for diagnosis of epileptogenic networks using temporal interference stimulation
Poster No:
87
Submission Type:
Abstract Submission
Authors:
Chloe DUPRAT1, Borana Dollomaja2, Jan Paul Triebkorn3, Jean-Didier Lemaréchal4, Fariba Karimi5, Maxime Guye6, Fabrice Bartolomei7, Huifang WANG3, Viktor Jirsa8
Institutions:
1AMU, MARSEILLE, BOUCHES-DU-RHONE, 2Institut de Neurosciences des Systemes UMR1106, Marseille, Marseille, 3AMU, INS, INSERM U1106, Marseille, PACA, 4AMU INS, MARSEILLE, BOUCHES-DU-RHONE, 5ETH, ZURICH, ZURICH, 6Aix Marseille Université, Marseille, PACA, 7AMU, INS, INSERMU1106, Marseille, PACA, 8nstitut de Neurosciences des Systèmes, Marseille, N/A
First Author:
Co-Author(s):
Introduction:
The exploration of brain activity in the context of refractory epilepsy has been attempted through modeling and simulation to provide patient-specific diagnosis. The efficacy of whole-brain models, such as the Virtual Epileptic Patient (VEP), has been demonstrated in simulating seizure-patterned brain activity, specifically induced by intra-cortical stimulation [1]. Recent studies have introduced the innovative technique of temporal interference (TI) to stimulate the brain non-invasively that effectively targets deep brain tissue with focal precision, avoiding overlying brain tissue [2]. Diagnosing epilepsy is challenging, especially in complex focal cases where specific brain areas, known as epileptogenic zones (EZ), are responsible for seizure onsets. Inducing seizures through brain stimulation, such as stereo-electroencephalography (SEEG) implantation based on the EZ hypothesis, is a current diagnosis approach [3]. However, this method has limitations: not every area can be implanted due to cognitive considerations, the location scheme established before induced-ictal recording cannot be modified between stimulations, the stimulated area is sensitively dependent to electrode contacts location inside the brain region and implantations can involves complications. The temporal interference method offers the same advantages i.e. depth and focality, without the aforementioned invasiveness-related challenges. Thus, combining temporal interference with scalp EEG allows simultaneous non-invasive monitoring of the whole brain activity. This study aims to assess the diagnosis performance of temporal interference stimulation in identifying epileptogenic networks.
Methods:
A high-resolution virtual brain, built from a patient's magnetic resonance and diffusion images (MRI and dMRI), incorporates the anatomical specifics and structural connectivity of the individual. The whole brain was discretized as a Neural Mass Model (NMM) for the subcortical regions and as a Neural Field Model (NFM) for the pial surface of the cortical regions [4]. Temporal interference fields are calculated and linearly summed to derive the modulation envelope of interfering electric fields. This resulting field, directly responsible for stimulation, is interpolated onto the virtual brain. The effective neural response is computed through the Epileptor-Stimulation model [5]. Simulations are conducted considering EEG, SEEG, and simultaneous EEG-SEEG recording methods. The results are fitted to the real time-series, and the model is inferred [6] to extract corresponding region parameters. These parameters are crucial, as they are intricately linked to properties of the brain, such as excitability associated with seizure onsets (Fig. 1). The parameters inferred from TI-induced ictal recordings are validated using ground truth parameters obtained from spontaneous seizure recordings. To assess the efficiency of EEG recording, we compare parameters inferred from TI-EEG with those from TI-SEEG and identify any additional information contributed by EEG during concurrent SEEG-EEG co-recording (Fig. 2).
Results:
The workflow was performed on a cohort of 20 patients from La Timone Hospital in Marseille. All patients experienced spontaneous and SEEG-induced seizures, all recorded by SEEG. The recorded ictal activity is then mapped at the source-level by the inverse solution and mapped again at the EEG-level by the forward solution on the virtual brain. We compared the performance of model inversion using SEEG, EEG and SEEG-EEG simultaneously. Results show that their performance depends on the different locations of EZNs and the seizure types.
Conclusions:
The efficacy of the TI stimulation modeling approach attests to its potential as a valuable tool for inducing seizure for drug-resistant epilepsy diagnosis. This study can contribute to the development of effective, targeted, low-volume, and cost-effective treatment options for refractory-epilepsy patients, providing a safer non-invasive therapeutic option.
Brain Stimulation:
Deep Brain Stimulation
Non-Invasive Stimulation Methods Other 1
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism)
Modeling and Analysis Methods:
Bayesian Modeling 2
Connectivity (eg. functional, effective, structural)
Keywords:
Computational Neuroscience
Degenerative Disease
Electroencephaolography (EEG)
Epilepsy
Machine Learning
Modeling
Other - Temporal Interference Stimulation
1|2Indicates the priority used for review
Provide references using author date format
[2] Grossman, N., Bono, D., Dedic, N., Kodandaramaiah, S. B., Rudenko, A., Suk, H. J., Cassara, A. M., Neufeld, E., Kuster, N., Tsai, L. H., Pascual-Leone, A., & Boyden, E. S. (2017). 'Noninvasive Deep Brain Stimulation via Temporally Interfering Electric Fields'. Cell, 169(6), 1029–1041.e16. https://doi.org/10.1016/j.cell.2017.05.024
[3] Bartolomei, F., Lagarde, S., Wendling, F., McGonigal, A., Jirsa, V., Guye, M., et al. (2017). 'Defining epileptogenic networks: Contribution of seeg and signal analysis'. Epilepsia 58, 1131–1147. doi:10. 1111/epi.13791
[4] Proix, T., Jirsa, V., Bartolomei, F., Guye, M., Truccolo, W. (2018). 'Predicting the spatiotemporal diversity of seizure propagation and termination in human focal epilepsy'. Nature communications, 9(1), 1088. https://doi.org/10.1038/s41467-018-02973-y
[5] Wang, H. E., Dollomaja, B., Duma, G. M., Lemarechal, J-D., Williamson, A., Makhalova, J., Triebkorn, P., Bartolomei, F., Jirsa, V. (2023). 'Personalized high-resolution virtual brain modeling for stimulation in epilepsy'. Frontiers. Submitted
[6] Jha, J., Hashemi, M., Vattikonda, A., Wang, H., Jirsa, V. (2022). 'Fully Bayesian estimation of virtual brain parameters with self-tuning Hamiltonian Monte Carlo'. Machine Learning: Science and Technology. 3. 10.1088/2632-2153/ac9037