Virtual Epileptic Patient Cohort (v1.0)

The virtual epileptic patient cohort is derived from a dataset of 30 drug-resistant epilepsy patients from La Timone hospital in Marseille, France. These patients have undergone invasive SEEG electrode implantation to record seizure activity. This dataset contains simulated SEEG recordings of spontaneous seizures, stimulation-induced seizures and interictal activity with spikes. In addition, patient-specific information including whole-brain connectivity, reconstructed SEEG electrode implantation and epileptogenic zone hypothesis are provided. The patient’s epileptogenic zone (EZ) hypothesis is used to parametrize their whole-brain model used for simulating epileptic brain activity. The purpose of this virtual cohort containing ground truth epileptogenic zone information is for facilitating evaluation of diagnostic data analysis methods in clinical epilepsy research.
We used The Virtual Brain platform to implement personalized brain models by integrating T1-MRI, diffusion-weighted MRI and CT-scans. For each brain model, the EZ was set using the clinical hypothesis in one scenario and the Virtual Epileptic Patient pipeline’s estimation of EZ in another scenario. We then ran simulations and validated the synthetic SEEG data against the empirical SEEG recordings by comparing for each electrode the seizure’s spatial and temporal propagation, the stimulation response and the interictal spike distribution. This dataset is BIDS-iEEG compatible, and the code used to generate it is available here: [https://github.com/BalanceKey/virtual_epilepsy_patient_cohort](https://github.com/BalanceKey/virtual_epilepsy_patient_cohort).