Adaptive fusion of accelerometer and ultrasound signals for real-time motion tracking in therapeutic ultrasound stimulation
Year: 2026
Type: article
Abstract: Objective . Ultrasound (US) stimulation of the spleen has shown promise in activating anti-inflammatory neural reflexes for treating conditions such as rheumatoid arthritis and sepsis. However, clinical translation remains challenging due to respiratory-induced motion, which risks off-target stimulation and often requires manual adjustment. While US imaging offers key advantages, tracking reliability is reduced by acoustic window loss and occlusion. This study introduces a robust tracking framework that enhances US-based motion monitoring via adaptive sensor fusion. Approach . A wearable accelerometer-based motion tracker was developed to monitor spleen displacement during US imaging. The model was trained and calibrated on short segments of tri-axial accelerometry data to estimate organ motion in the US imaging plane. Performance was evaluated within and across sessions and for different sensor placements. The optimal configuration was integrated into an adaptive tracker that automatically switches between image- and sensor-based estimates using a frame-wise tracking confidence module, ensuring continuous motion monitoring under loss of the acoustic window. Main results . The standalone accelerometer model achieved an average error of 2.0 ± 0.8 mm, computed as Euclidean distance to a landmark, across sessions in 27 participants. Under challenging imaging conditions, this model achieved error of 2.3 ± 0.6 mm and the adaptive US-accelerometer tracker significantly improved performance, reducing the error to 0.5 ± 0.1 mm. Significance . A compact, low-cost, body-mounted accelerometer can enhance US-guided therapies by providing robust and continuous motion estimates. The proposed framework is scalable and generalizable to other abdominal targets, supporting autonomous US-based interventions and enables at-home wearable therapeutic applications.
Source: Physics in Medicine and Biology
Institution: University of Minnesota
Language: English
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $3,325
Funder: University of Minnesota
OpenAlex ID: W7141241522