Resting-state precision functional mapping corresponds with behavioral effects of intracranial electrical stimulation

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Christopher Cyr¹ (christopher.cyr@northwestern.edu), Ania Holubecki¹, James Kragel², Christina Zelano¹, Joel Voss², Joshua Rosenow¹, Stephan Schuele¹, Elizabeth Johnson¹, Rodrigo Braga¹; ¹Northwestern University, ²University of Chicago

Intracranial electrical stimulation (iES) can provide causal information regarding the function of a stimulated brain region. Non-invasive mapping procedures, such as those based on functional MRI (fMRI), provide complementary information to iES. However, the accuracy of fMRI has been limited because typical procedures (i) collect insufficient amounts of data for reliable within-individual estimates and (ii) rely on patients performing tasks that typically target a single functional domain or network, and can yield variable results depending on the task design and patient's strategy. In contrast, resting-state functional connectivity (FC) can estimate multiple networks at once and does not rely on an active task. Here, we investigated whether within-individual FC estimates from extensively sampled patients correspond with behavioral effects elicited by clinical iES. Six patients with drug-resistant epilepsy who were scheduled for invasive monitoring completed up to 4 MRI sessions, providing 35-245 minutes of fMRI rest data. Three patients additionally provided 10-35 minutes of language task-based fMRI data. Electrodes were localized using a post-surgery CT image co-registered to MRI data. Stimulation (bipolar, 50 Hz, 1-15 mA, 300 µs pulse width) was applied while patients read passages aloud. Positive (e.g., hand jerks) and negative (e.g., speech arrest) effects were documented by an epileptologist. Across participants, an optimization analysis indicated that the FC-estimated language network displayed 64-82% correspondence with reading deficits, while a task-based language map displayed 45-79% correspondence, suggesting that resting-state FC-based network estimation provides similar information to task-based language mapping. Ongoing work is assessing correspondence of FC-defined networks in additional functional domains.

Topic Area: OTHER