TL;DR
This study bridges a critical gap in understanding Cranial Electrical Stimulation (CES) by performing the first direct, in vivo measurements of electric fields (EFs) in the human brain using implanted electrodes. The findings validate personalized computational models, showing they can reliably predict CES-induced EFs across different ages and highlight the significant impact of electrode placement on EF distribution.
What Problem Does This Paper Address?
Cranial Electrical Stimulation (CES) is a non-invasive neuromodulation technique used for various neurological conditions, but its effectiveness is inconsistent. A key hypothesis is that the spatial electric fields (EFs) generated within the brain are responsible for its biological effects. While computational models exist to predict these EFs, they have largely lacked validation with direct in vivo measurements in humans. This is particularly challenging because obtaining such data requires invasive procedures. Furthermore, the influence of factors like electrode montage, stimulation parameters (frequency, intensity), and individual differences (age) on these EFs in real brains remains unclear. This paper aims to address this by directly measuring CES-induced voltage changes in epilepsy patients with implanted electrodes and comparing these to simulations.
Methods
The researchers employed a multi-faceted approach:
- In Vivo Measurement: They utilized stereotactic-electroencephalographic (sEEG) electrodes, which are surgically implanted to record brain activity, in 21 epilepsy patients (16 adults, 5 children). These electrodes allowed for direct measurement of voltage changes induced by CES.
- CES Protocol: A standardized CES protocol was applied, varying stimulation intensity (5-500 µA) and frequency (0.5-200 Hz) for some participants.
- Personalized Computational Modeling: For each participant, individual T1-weighted MRI and CT scans were used to create personalized head models. The ROAST software was employed to simulate EF distributions based on these models.
- Data Analysis: Measured voltage and calculated EF data from the sEEG electrodes were compared with the simulated values. Statistical analyses, including Pearson correlation coefficients and linear regression, were used to assess the agreement between measured and simulated data. The influence of stimulation parameters and age on EFs was also investigated.
Datasets
The study used individual T1-weighted MRI and CT scans from the 21 epilepsy participants for creating personalized computational models. No new benchmark datasets were introduced.
Key Insights and Ideas
- Validation of Simulation Models: The study provides strong evidence that personalized computational models can accurately predict CES-induced electric fields in the live human brain. This is a crucial step towards reliable neuromodulation planning.
- Importance of Individualization: The high correlation between measured and simulated data underscores the necessity of using individual anatomical data (MRI/CT scans) for accurate EF prediction.
- Electrode Montage Matters Significantly: The placement of electrodes has a substantial impact on the distribution of EFs, even with minor variations. This suggests that optimizing electrode placement is critical for targeted CES.
- Age is Less of a Factor than Expected: While anatomical differences exist between adults and children, the simulation models showed comparable predictive accuracy across age groups, suggesting that current modeling approaches can generalize well.
- Frequency Independence: Within the tested range (0.5-200 Hz), CES-induced voltages showed minimal changes with varying frequencies, indicating that the brain might behave as a resistive conductor at these frequencies.
Key Results
- High Correlation: Measured voltages and EFs were highly correlated with predicted values from personalized models across all participants (voltages: r = 0.93, p < 0.001; EFs: r = 0.73, p < 0.001).
- Linear Intensity Response: In white and gray matter, measured voltages increased linearly with stimulation intensity from 5 to 500 µA.
- Frequency Stability: Stimulation frequency (0.5-200 Hz) had no significant impact on measured voltages, with an average coefficient of variation less than 4.10%.
- Montage Effect: Electrode montage significantly affected EF distribution (p < 0.01), with pre-auricular montages generally producing stronger peak EFs than infra-auricular ones.
- Age Effect: Age did not significantly affect the distribution of EFs, although simulated EFs were slightly higher in children, consistent with previous modeling studies.
Limitations and Future Work
The authors acknowledge several limitations:
- Tissue Segmentation and Conductivity: The simulation models used a simplified segmentation of brain tissues and relied on commonly used isotropic conductivity values. Future work should incorporate more detailed tissue segmentation and explore anisotropic conductivity, especially with the availability of diffusion tensor imaging (DTI) data.
- Limited Electrode Coverage: While sEEG provided unprecedented access, the coverage was still limited to specific brain regions.
- Sample Size for Children: The number of child participants (n=5) was relatively small, limiting definitive conclusions about age-specific effects.
- Clinical Artifacts: The study was conducted in epilepsy patients, and while efforts were made to minimize artifacts, the presence of implanted electrodes and potential seizure activity could introduce some noise.
Future work should focus on improving simulation accuracy through more refined tissue properties and segmentation, exploring anisotropic conductivity, and expanding studies to larger and more diverse pediatric populations. The findings also pave the way for optimizing CES protocols and developing personalized neuromodulation strategies.
Summary updated