iEEG-recon: A Fast and Scalable Pipeline for Accurate Reconstruction of Intracranial Electrodes and Implantable Devices

Artificial intelligence in epilepsy phenotyping

Artificial intelligence (AI) allows data analysis and integration at an unprecedented granularity and scale. Here we review the technological advances, challenges, and future perspectives of using AI for electro-clinical phenotyping of animal models and patients with epilepsy. In translational research, AI models accurately identify behavioral states in animal models of epilepsy, allowing identification of correlations between neural activity and interictal and ictal behavior. Clinical applications of AI-based automated and semi-automated analysis of audio and video recordings of people with epilepsy, allow significant data reduction and reliable detection and classification of major motor seizures. AI models can accurately identify electrographic biomarkers of epilepsy, such as spikes, high-frequency oscillations, and seizure patterns. Integrating AI analysis of electroencephalographic, clinical, and behavioral data will contribute to optimizing therapy for patients with epilepsy.

Direct cortical stimulation induces short-term plasticity of neural oscillations in humans

Patterned brain stimulation is commonly employed as a tool for eliciting plasticity in brain circuits and treating neuropsychiatric disorders. Although widely used in clinical settings, there remains a limited understanding of how stimulation-induced plasticity influences neural oscillations and their interplay with the underlying baseline functional architecture. To address this question, we applied 15 minutes of 10Hz focal electrical simulation, a pattern identical to 'excitatory' repetitive transcranial magnetic stimulation (rTMS), to 14 medically-intractable epilepsy patients undergoing intracranial electroencephalographic (iEEG). We quantified the spectral features of the cortico-cortical evoked potential (CCEPs) in these patients before and after stimulation. We hypothesized that for a given region the temporal and spectral components of the CCEP predicted the location and degree of stimulation-induced plasticity. Across patients, low frequency power (alpha and beta) showed the broadest change, while the magnitude of change was stronger in high frequencies (beta and gamma). Next we demonstrated that regions with stronger baseline evoked spectral responses were more likely to undergo plasticity after stimulation. These findings were specific to a given frequency in a specific temporal window. Post-stimulation power changes were driven by the interaction between direction of change in baseline power and temporal window of change. Finally, regions exhibiting early increases and late decreases in evoked baseline power exhibited power changes after stimulation and were independent of stimulation location. Together, these findings that time-frequency baseline features predict post-stimulation plasticity effects demonstrate properties akin to Hebbian learning in humans and extend this theory to the temporal and spectral window of interest. These findings can help improve our understanding of human brain plasticity and lead to more effective brain stimulation techniques.

Significance Statement

Brain stimulation is increasingly used to treat neuropsychiatric disorders by inducing changes in neural activity at specific brain regions. Despite their effectiveness, how these changes occur, specifically in the spectral domain, is unknown. To better understand how brain oscillations change after patterned stimulation, we performed focused stimulation in epilepsy patients and measured intracranial brain recordings. We found strong and predictable changes in brain oscillations (plasticity) after patterned stimulation. Specifically, low frequencies showing widespread effects and high frequencies exhibiting a greater magnitude of change. These changes were directly related to the temporal and spectral structure of brain responses prior to stimulation. Our study reveals that baseline brain activity patterns can predict how stimulation will induce plasticity in the spectral domain. These findings can help improve our understanding of human brain plasticity and lead to more effective brain stimulation techniques.

Highlights

We applied 15 minutes of repetitive 10Hz focal electrical stimulation and assessed the evoked brain-wide spectral changes with intracranial EEG.10Hz stimulation induced short-term plasticity in low frequency alpha evoked power broadly across regions and time windows and high frequency (beta, gamma) power specifically in early evoked time windows (10-50ms).Across patients, frequency bands, and time windows, brain regions with stronger baseline evoked power were more likely to undergo greater spectral changes after 10Hz stimulation.Post-stimulation spectral changes were specific; that is, for a given frequency band in a specific time window, baseline evoked power predicted post-stimulation change in the same frequency band and time window.Post-stimulation spectral change was driven by an interaction between direction of change and temporal window of baseline power; that is, regions exhibiting baseline evoked early (10-100ms) increases and late (100-200ms) decreases in power correlated with observed post-stimulation spectral changes.These results were independent of stimulation location.

YAEL: Your Advanced Electrode Localizer

Intracranial electroencephalography (iEEG) provides a unique opportunity to record and stimulate neuronal populations in the human brain. A key step in neuroscience inference from iEEG is localizing the electrodes relative to individual subject anatomy and identified regions in brain atlases. We describe a new software tool, Your Advanced Electrode Localizer (YAEL), that provides an integrated solution for every step of the electrode localization process. YAEL is compatible with all common data formats to provide an easy-to-use, drop-in replacement for problematic existing workflows that require users to grapple with multiple programs and interfaces. YAEL's automatic extrapolation and interpolation functions speed localization, especially important in patients with many implanted stereotactic (sEEG) electrode shafts. The graphical user interface is presented in a web browser for broad compatibility and includes an interactive 3D viewer for easier localization of nearby sEEG contacts. After localization is complete, users may enter or import data into YAEL's 3D viewer to create publication-ready visualizations of electrodes and brain anatomy, including identified brain areas from atlases; the response to experimental tasks measured with iEEG; and clinical measures such as epileptiform activity or the results of electrical stimulation mapping. YAEL is free and open source and does not depend on any commercial software. Installation instructions for Mac, Windows, and Linux are available at https://yael.wiki.