Neural Noise

Virtual brain twins for stimulation in epilepsy

2 min read

  • Patient-specific computational models (virtual brain twins) can inform stimulation design in epilepsy.1
  • Virtual brain twins support closed-loop and responsive stimulation design and have a strong implementation path (Nature).1 1

Gardner updates

  • Patient-specific computational models (virtual brain twins) for stimulation in epilepsy inform closed-loop and responsive stimulation design. 1

Weekly enrichment (2026-07-20)

  • Published in Nature Computational Science (2025), the framework builds patient-specific, high-resolution whole-brain models (“virtual brain twins”) to estimate the epileptogenic zone network (EZN) in drug-resistant focal epilepsy.23
  • Models integrate T1-weighted MRI cortical geometry and subcortical volumes with diffusion-weighted MRI tractography for structural connectivity, simulating neural source activity at ~10 mm² spatial resolution across roughly 20,284 cortical vertices.2
  • Seizure dynamics are simulated with an Epileptor-Stimulation neural mass model; patient-specific parameters (regional epileptogenicity and global network scaling) are inferred via Hamiltonian Monte Carlo Bayesian model inversion.2
  • Two stimulation modalities are modeled: invasive SEEG bipolar stimulation (1 Hz or 50 Hz, 1 ms pulses at 1–3 mA) and non-invasive temporal interference (TI) stimulation delivered through two scalp electrode pairs at ~1,000 Hz and 1,005 Hz.23
  • In an illustrative case (a 23-year-old woman with left occipital lobe epilepsy), the model consistently identified the left lateral occipital cortex (region O2 in the VEP atlas) as the EZN, matching the region resected at surgery.2
  • The approach builds on the Virtual Epileptic Patient (VEP) workflow, evaluated retrospectively in 53 patients with 187 spontaneous seizures and prospectively in the ongoing EPINOV clinical trial with 356 patients.2
  • TI stimulation can reach deeper structures than conventional transcranial direct- and alternating-current stimulation, supporting a transition from invasive to non-invasive diagnosis and treatment.2
  • The pipeline supports multimodal inference by combining scalp-EEG and SEEG and aims to optimize stimulation parameters for diagnosis and treatment, informing closed-loop and responsive neurostimulation design.23

Footnotes

  1. https://news.google.com/rss/articles/CBMiX0FVX3lxTE93TU1Dam5rd3FrQjJqNjVaSnNJd1FrdHNlUDZTOWMwM2llMjNsQTd1M1pUNFpnSU9sWnhzSkdLVkRISzdkY0xzc3prb0tLcXNfZDVUank5TGNzX1g0cG9N?oc=5 ↩ ↩2 ↩3 ↩4

  2. https://doi.org/10.1038/s43588-025-00841-6 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  3. https://link.springer.com/article/10.1038/s43588-025-00841-6 ↩ ↩2 ↩3

Graph View