UC San Diego

Vagus nerve electrical stimulation is known to improve learning processes and augment anti-inflammatory treatment [1]. Current methods include the surgically implanted cuff electrode, or the noninvasive electric field that penetrates from the skin to the nerve. The former ensures target accuracy yet can be fraught with complications due to surgical implant, whereas the latter avoids procedural complications but may result in off target neurostimulation (i.e., other neuronal stimulation besides vagus nerve)[2]. Off target neuronal stimulation can alter cardiac function and cerebral blood flow patterns [3] [4]. Converging lines of evidence now support non-invasive high temporal and spatial resolution focused ultrasound can result in brain and peripheral nerve stimulation [4] [5] [6]. Therefore, focused ultrasound is proposed to replace electrical nerve stimulation to effectively modulate the vagus nerve as a safe and noninvasive modality. To guarantee safety and efficacious vagus nerve focused ultrasound stimulation, real-time image guidance is required to provide efficient and accurate tracking of the vagus nerve. In this work, we propose to: 1) develop real time B-mode ultrasound imaging to visualize the vagus nerve and 2) employ neural networks capable of tracking the nerve location within the view. We implement the real-time beamforming process from raw data to the reconstructed B-mode image and send it to the vagus nerve detector that is trained using convolutional neural networks. We successfully extract the location of the nerve from the beamformed data, validating the idea of real-time guidance. Future work will focus on improving the efficiency of nerve detection algorithms that are then used as feedback for an eventual autonomous closed loop focused ultrasound nerve stimulation system.