Abstract

Chronic pain is a debilitating and intractable disease characterized by pathological changes to the nervous system. Specifically, chronic pain is associated with changes in neural circuitry, including maladaptive neuroplasticity, cortical reorganization, and changes to descending pain modulation pathways. Neurotechnologies, or technologies that interface with the nervous system, provide a therapeutic modality that can directly target the neuropathology underlying chronic pain. For example, brain-computer interfaces (BCIs) provide a promising avenue for therapeutic neural remodeling, specifically by leveraging neurofeedback to reinforce beneficial neural patterns. However, due to the variability of the functional and structural changes in the nervous system that contribute to chronic pain, BCIs have yet to be optimized to treat chronic pain. Further, our incomplete understanding of the cortical networks that underlie pain perception limits the design and translation of effective BCI systems. This dissertation addresses these challenges through three studies. The first study introduces a novel theta-controlled BCI system, which was tested in six chronic pain patients during an open-label pilot study. Through haptic and visual neurofeedback, the BCI was able to reinforce frontal theta in this study population over a six-week intervention. Patients increased their BCI performance, reflecting thought-driven control of neurofeedback, and showed a significant decrease in pain severity and pain interference scores without any adverse events. Pain relief significantly correlated with frontal theta modulation. These findings highlight the potential of BCI-mediated cortico-sensory coupling of frontal theta with vibrotactile stimulation for alleviating chronic pain. While frontal theta has been identified as an attractive pain-relief biomarker, the neurophysiology underlying this phenomenon is poorly understood, especially in the context of pain. We hypothesized that the source of these frontal theta features was the anterior cingulate cortex. To this end, our second study provides deeper insights into the contributions of activity from the human cingulate cortex (CC) and its networks to surface EEG features. Through invasive electrical stimulation of the subregions of the cingulate cortex and concurrent recording on surface EEG, we characterized the recruitment of CC networks through electrophysiological features captured on the scalp surface. We found that stimulation of the Midcingulate cortex produced distinct surface features in frontal regions, coinciding with our spatial targets during our first study. Further, we were able to characterize the temporal features of each CC subregion on scalp EEG. The final study provides a comprehensive profile of CC effective connectivity networks. The subregions of the CC are highly implicated in the perception of pain, yet precise relationships within its effective networks remain understudied. We mapped divergent effective networks of the anterior, mid, and posterior cingulate in 15 resection-free patients using single-pulse electrical stimulation, quantifying both cortico-cortical evoked potential (CCEP) magnitude and a morphology-based coherence metric, and profiling spatial, temporal, spectral, and inter-regional waveform features with machine learning. Coherence complemented magnitude, revealing hippocampal connections undetected by magnitude, and CC hub regions with equivalent connectivity to the anterior, midcingulate, and posterior subregions. Profiling the spatial distribution of effective networks confirmed that spatial patterns of effective connectivity coincide with distributions of CC white matter tractography. Inter-regional hierarchical clustering showed the midcingulate network had the most stereotyped CCEP morphology across regions. A Random Forest model trained to classify CC effective networks using features extracted from CCEPs reached 75% accuracy and revealed that assumption-based waveform features had the lowest predictive strength. These results provide a detailed, network-level account of human CC effective connectivity and highlight the utility of multimodal, feature-driven approaches over traditional assumption-based metrics. Together, these three studies present a novel BCI modality for treating chronic pain and elucidate neurophysiological mechanisms of pain perception networks critical for the improvement of future neurotechnologies.

Committee Chair

Eric Leuthardt

Committee Members

Peter Brunner

Degree

Doctor of Philosophy (PhD)

Author's Department

Biomedical Engineering

Author's School

McKelvey School of Engineering

Document Type

Dissertation

Date of Award

12-19-2025

Language

English (en)

Download

Available for download on Friday, December 18, 2026

Included in

Neurosciences Commons

Share

COinS