# Sonomyography accurately captures joint kinematics during volitional and electrically stimulated motion in healthy adults and an individual with cerebral palsy

## TL;DR

Sonomyography (SMG) accurately captures joint movements during both voluntary and electrically stimulated motions in healthy individuals and someone with cerebral palsy.

## Contribution

This study demonstrates the technical feasibility of using sonomyography to estimate joint kinematics during both volitional and electrically stimulated movements.

## Key findings

- Sonomyography showed lower RMSE (0.10–0.25) compared to EMG (0.34–0.57) for joint angle prediction.
- SMG performance was consistent across both volitional and electrically stimulated conditions.
- SMG results were comparable to previously reported values in the literature.

## Abstract

Despite significant advances in biosignal extraction techniques for studying neuromotor disorders, there remains an unmet need for a method that effectively links muscle structure and dynamics to muscle activation. Addressing this gap could improve the quantification of neuromuscular impairments and pave the way for precision rehabilitation. In this study, we demonstrate the proof of concept of recording multimodal signals from the brain, muscles, and resulting limb kinematics. We also explore the use of ultrasound imaging to extract limb kinematics.

We collected data from three healthy volunteers and one individual with cerebral palsy during single degree-of-freedom ankle and wrist movements. Participants performed range of motion (ROM) tasks at approximately 1-second intervals, either volitionally or through functional electrical stimulation. We simultaneously recorded electroencephalography, surface electromyography (EMG), continuous ultrasound imaging, and motion capture data. Joint kinematics were computed from ultrasound imaging using a technique called sonomyography (SMG), and we evaluated the technical feasibility of estimating joint kinematics from both sonomyography and surface EMG signals.

The technical feasibility study evaluated joint angle prediction using EMG and SMG under volitional (FES-OFF) and electrically stimulated (FES-ON) conditions. Root mean squared error (RMSE) between predicted and measured joint angles was computed for multiple methods of extracting kinematics from EMG and SMG. EMG-based RMSE ranged from 0.34 to 0.57 (FES-OFF) and 0.43–0.51 (FES-ON). SMG-based RMSE ranged from 0.10 to 0.25 across all conditions and methods. Linear regression analysis produced \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$R^2$$\end{document}R2 values between 0.31 and 0.81 depending on joint, condition, and method. No significant RMSE difference was found between FES-ON and FES-OFF conditions within SMG. SMG RMSE values were also comparable to previously reported values (10-25%) in prior literature.

Our findings suggest that sonomyography can be used as a noninvasive method for estimating joint kinematics when the joint movement is driven either by volition or by functional electrical stimulation. This technique can potentially be be useful in evaluating altered muscle dynamics and driving assistive and rehabilitation devices in individuals with neuromotor disorders such as cerebral palsy.