Advancing Vagus Nerve Therapies: Ultrasound Motion Tracking for End-Organ Stimulation Therapy and Multimodal Analysis of the Effects of Implanted Cervical Vagus Nerve Stimulation Across Body Systems

Abstract

The vagus nerve, the tenth cranial nerve, provides extensive parasympathetic innervation to thoracic and abdominal organs and plays a central role in regulating cardiovascular, immune, metabolic, and neurological functions. Because dysregulation of vagal activity contributes to diverse diseases, vagus nerve stimulation (VNS) has emerged as a promising therapeutic strategy. Two major approaches are currently investigated: cervical VNS (cVNS), delivered through implanted electrodes on the cervical trunk, and ultrasound (US)-based stimulation, a noninvasive technique that targets polysynaptic pathways within end-organs such as the spleen and liver. Both approaches face critical challenges. For US stimulation, respiratory-induced motion, acoustic shadowing, and occlusions complicate consistent targeting of moving organs. For cVNS, the lack of knowledge about human vagus nerve physiology, fiber selectivity, and multi-organ effects limits optimization of stimulation parameters and therapeutic efficacy. This thesis addresses these gaps through two parallel contributions. First, novel tracking frameworks were developed to enable accurate, robust motion monitoring of visceral organs for integration into US stimulation platforms. An enhanced KLT (Kanade–Lucas–Tomasi) tracker combined with a long short-term memory (LSTM) predictor (EKLT-LSTM) achieved high accuracy in liver and spleen tracking, with mean errors of 0.4 ± 0.4 mm for shallow breathing and 1.37 ± 0.9 mm for deep breathing in spleen recordings, and 0.3 ± 0.2 mm in a benchmark liver dataset. In addition to providing real-time tracking, this framework can also support annotation tasks by generating high-quality motion labels. To further enhance robustness in real-world conditions, a hybrid accelerometer-fusion tracker (CKLT-AF) was developed, integrating respiratory motion signals from a chest-mounted sensor with a conditioned KLT image tracker. This adaptive system dynamically switched between modalities and is particularly responsive to sudden breathing changes and maintains continuous sub-millimeter accuracy even under occlusion and out-of-plane motion, achieving 0.5 ± 0.1 mm error in cross-session validation. These contributions establish the first motion tracking solutions specifically designed for the challenges of US stimulation. Second, the thesis contributes to the REVEAL (Research Evaluating Vagal Excitation and Anatomical Linkages) study, the first clinical investigation of implanted cVNS across multiple physiological systems simultaneously. While most prior work has focused on isolated effects such as heart rate modulation or immune biomarkers, REVEAL collects a uniquely comprehensive dataset. This includes acute and chronic measurements across cardiovascular, metabolic, immune, and nervous systems: heart rate, blood pressure, ECG, respiration, muscle sympathetic nerve activity (MSNA), and 24-hour ambulatory recordings, in addition to blood biomarkers. Participants undergo testing under six distinct stimulation settings, enabling systematic evaluation of cVNS effects across multiple physiological domains. At the time of writing, 10 out of 144 participants have completed the full study protocol. By analyzing these multimodal responses, this thesis advances mechanistic understanding of vagal physiology in humans and lays the foundation for precision, whole-health applications of VNS. In summary, this work introduces novel US tracking frameworks to support the translation of US stimulation therapies, and provides one of the first multimodal analyses of cVNS in humans. Together, these contributions address key technical and physiological knowledge gaps, moving the field closer to precise, safe, and effective vagus nerve–based bioelectronic medicine.