Clinical investigations: the Davis Lab at Penn Neurology

Led by Kathryn Davis, MD, MTR, the Davis Lab at Penn Neurology is currently conducting clinical studies to advance the fields of invasive neurophysiology and neuroimaging in medication-refractory epilepsy patients. Among the goals of these studies are the development of objective noninvasive imaging biomarkers to enable epileptologists to better identify the epileptic networks responsible for generating and spreading seizures (localization), and to better predict responses to intervention (e.g., seizure control devices, resective surgery, or continued medical management) in individual patients. Other objectives include the integration of AI into the management and tracking of epileptogenesis and ictal pathology.

Background

Of individuals with focal epilepsy, about one third will have drug-resistant disease (DRE). Of these patients, 60 percent can expect to gain seizure freedom or improvement in quality of life if the source of their seizures can be accurately identified and addressed through surgery or ablation.

The importance of noninvasive modalities is evident in the evaluation of patients with DRE. For pre-surgical evaluation, the majority of DRE patients will have intracranial EEG (iEEG), of which two forms are available. Electrocorticography (ECoG), which uses subdural grid electrodes placed directly on the brain, is considered among the most invasive diagnostic procedures in modern medicine. Somewhat less invasive is stereotactic EEG (sEEG), which places rigid electrodes through 2-3 ml holes in the skull. In the hands of advanced specialists, iEEG is highly accurate in detecting and localizing brain events and is considered a critical tool in the prediction of surgical outcomes in epilepsy.

However, iEEG is much more invasive than the tools used to diagnose most neurological disorders, including the movement disorders, and more invasive than the adjuncts used in epilepsy diagnosis (e.g., MRI, PET, SPECT).

Moreover, iEEG is expensive ($95,000 – $106,000 including surgical implantation, and multi-day ICU/hospital monitoring), and its availability is largely limited to medical academic centers and large tertiary care hospitals.

Importantly, at this time clinicians lack methods to quantitatively map, or integrate, high-resolution anatomical and functional data (such as MRI, PET, or CT) with the high-temporal-resolution electrical recordings from iEEG electrodes implanted in the brain. This has prompted a critical demand to validate whole-brain noninvasive neuroimaging network-based biomarkers to guide precise placement of electrodes and translate noninvasive network neuroimaging in epilepsy management.

To address this need, Davis Lab investigators and their partner institutions are conducting a study with the eventual objective of enabling epileptologists to better localize epileptic networks and predict surgical outcomes from the integration of MRI and iEEG data.

Integrating iEEG with high resolution imaging

The Davis Lab is currently conducting the clinical trial Optimized Intracranial EEG Targeting in Focal Epilepsy based upon Neuroimaging Connectomics (NCT04649008), which aims to better define epileptogenic networks, particularly for temporal lobe epilepsy. The study represents a substantive departure from the status quo by directly connecting noninvasive multimodal imaging with measures of functional network dynamics in iEEG.

Working on the hypothesis that noninvasive measures of structure and function relate to and can predict the intricate functional dynamics captured on iEEG, the investigators’ overall objective is to develop open-source noninvasive imaging tools that map epileptic networks by integrating MRI and iEEG data. It is expected that successful completion of these aims will yield personalized strategies for iEEG targeting based on noninvasive neuroimaging.

This hypothesis will be evaluated in patients undergoing iEEG targeting the temporal lobe network by tracing the patient-specific white-matter tracts between brain regions (structural connectome) to iEEG seizure onset and propagation. Seizure onset and propagation will then be correlated with network measures derived from resting state functional MRI (rsfMRI). Finally, these structural and functional elements will be integrated with standard qualitative clinical data to predict iEEG network dynamics and surgical outcomes.

Protocol: Patients will undergo diffusion tensor imaging (DTI) prior to stereotactic iEEG. The functional iEEG network will be mapped to DTI, defining how seizures are constrained by the underlying structural connectome as they propagate.Functional network measures from rsfMRI and iEEG in patients with temporal lobe epilepsy will be co-registered, and rsfMRI will be used to predict functional EEG ictal and interictal networks. In the final aim, two models predicting iEEG network dynamics and epilepsy surgical outcomes will be created building off of methods developed in the previous procedures.

Recent Publications from the Davis Lab, et al

The Davis Lab and its partners at the University of Pennsylvania and elsewhere have collectively reported study findings on a wide range of topics relevant to epilepsy seizure analysis and management, a selection of which is present herein. A further overview of recent investigations from Dr. Davis and her colleagues is available at PubMed.

AI-Driven Mapping of Seizure Spread Patterns (Ann Neurol 2026;00:1–15), for example, saw the development of deep learning algorithms on a small subset of patients to detect seizure activity and deploy these algorithms across 275 seizures in 71 patients. The intent was to analyze the patterns of seizure spread (extent, timing, surgical outcomes, and common patterns) and incorporate diffusion-weighted imaging to understand how these patterns relate to the structural connections of the brain.

Results: Using physician annotations as a benchmark, deep learning algorithms were found to outperform single features (line length, absolute slope, and power) in ranking seizure onset contacts. In addition, poor outcome patients had more extensive brain regions involved in their seizures and more rapid spread between temporal lobes. Incorporating diffusion-weighted imaging, the investigators found an association between an increase in structural connectivity between temporal lobes and quicker seizure spread. Finally, clusters of spread patterns common across patients based on spread timing, location, and extent were identified.

White matter (WM) signals reflect information transmission between brain regions during seizures (Brain 2026;149:77-89) explored the electrophysiological understanding of white matter (WM) among individuals with DRE who underwent stereo EEG for surgical evaluation .

The study involved white matter recordings in 29 patients with drug-resistant epilepsy who underwent stereo EEG for surgical evaluation. Among the findings of the investigation were the following:

• WM intracranial EEG power is lower during seizures and elevated postictally;
• WM signals are more correlated with each other than grey matter signals and these correlations increase more than grey matter signals during seizures;
• Patients with poor outcomes have higher ictal SEEG white matter connectivity than patients with good outcomes;
• Ictal functional connectivity between ablated and white matter regions relates to patient outcomes.

These results support the distributed epileptic network hypothesis, which states that the true seizure onset zone is not a single focal anatomical location per se (in some cases), but rather a distributed epileptic network where many brain regions interact together, allowing a patient to enter a seizure state. Thus, WM electrophysiology and connectivity may represent information on the spatial distribution of epilepsy and its amenability to intervention, whether through focal ablation or modulation through devices. Overall, WM functional recordings might provide a wealth of currently untapped knowledge about the neurobiology of disease and could guide clinical decision-making in treatment of drug-resistant epilepsy patients.

Further information about epilepsy clinical research efforts at Penn Neurology is available online at the Penn Physician Hub.

About the Davis Lab at Penn Neurology

Located at the University of Pennsylvania, the Davis Lab is led by Kathryn A. Davis, MD, MTR, the Medical Director of both Penn’s Epilepsy Monitoring Unit and Epilepsy Surgical Program. Comprising specialists from Penn Neurology and the University of Pennsylvania, as well as post-doctoral fellows, graduate students, and clinical research coordinators, the Lab conducts research in the advancing fields of invasive neurophysiology and neuroimaging to better localize epileptic networks in medication refractory epilepsy patients.