Neural Signals-Based Respiratory Motion Tracking: A Surface Electromyography Study.

Abstract

Neural signals-based respiratory motion tracking offers a potential solution to the system latency issue of medical linear accelerators in respiratory motion tracking radiation therapy. However, decoding respiratory-related neural signals from scalp electroencephalography in real-time is challenging. Herein we propose a clinically applicable neural signals-based respiratory motion tracking approach using surface electromyography (sEMG). Neural signal and respiratory motion of 15 healthy subjects were simultaneously recorded using an sEMG system and a pressure sensor embedded in a stretchy belt. Cross-correlation analysis was performed to characterize the time dependencies between the respiratory-related neural signal extracted via an offline analysis method and respiratory motion. Combined with recurrent neural networks-based online smoothing, this offline analysis method was adapted into an online analysis framework to enable real-time prediction of respiratory motion. Using the respiratory motion as a reference, the resulting signal from the online analysis was compared using the mean absolute error and root mean square error. The correlation coefficients between the offline-extracted respiratory-related neural signal and respiratory motion consistently exceeded 0.90, with an average precursor time of 319 ms. No statistically significant difference was observed between the precursor time of the first 1-minute interval and that of the subsequent 9-minute intervals. In the online analysis, the proposed method achieved an mean absolute error of 0.075 ± 0.021 and root mean square error of 0.098 ± 0.028. We have proposed a clinically applicable neural signals-based respiratory motion tracking method using sEMG. The proposed online analysis extracts respiratory-related neural signals with minimal latency while maintaining high accuracy. These findings suggest that neural signals-based respiratory motion tracking using sEMG is a promising solution to the system latency issue of medical linear accelerators in cancer radiation therapy.