Design of a Wireless and Non-Invasive EMG-Based Embedded Measurement System for Monitoring Muscle Activation Asymmetries in Scoliosis

Adolescent idiopathic scoliosis (AIS) is a three-dimensional spinal deformity typically diagnosed and monitored through radiographic assessment of the Cobb angle. While imaging remains the clinical gold standard, it involves radiation exposure and is typically performed at limited intervals, making it less suitable for frequent, short-term monitoring of treatment response. Prior research has shown that paraspinal muscle activation asymmetries - particularly increased convex-side activation near the apex of the curve - are consistently observed in AIS and may serve as early, objective markers of curve progression and treatment response, as demonstrated in prospective cohort and machine-learning-based studies.

This study will use a custom-developed, low-power, wireless, embedded surface EMG (sEMG) measurement system, consisting of skin-surface EMG electrodes, an analog front-end (AFE) for signal conditioning and digitization, and a microcontroller-based wireless communication module. EMG signals will be sampled at a minimum of 1000 Hz using a differential (bipolar) electrode configuration to minimize noise and common-mode interference.

Electrodes will be placed bilaterally on the paraspinal muscles at the level of the apical vertebra of the major spinal curve, approximately 3 cm lateral to the spinous process, oriented parallel to the muscle fibers. Standard skin preparation procedures (shaving, alcohol cleaning) will be used to reduce skin impedance prior to electrode placement.

EMG recordings will be performed under two task conditions: a static task, in which participants sit upright with hips and knees flexed at 90°, arms at their sides, and feet on the floor; and a dynamic task, in which participants perform active trunk extension in the prone position for a defined number of repetitions. These tasks are designed to capture both static postural and dynamic movement-related muscle activation patterns.

Recorded EMG signals will be processed to calculate root mean square (RMS) and mean absolute value (MAV) parameters, with frequency-domain (FFT) analysis performed as needed. Convex- and concave-side RMS values at each vertebral level will be used to derive muscle activation ratios and asymmetry indices. Measurements will be repeated at different time points over the course of physiotherapy treatment to evaluate longitudinal changes in muscle activation patterns, and associations between EMG-based indices and clinical treatment outcomes will be examined using appropriate statistical methods.