腦中風與脊髓損傷常導致運動神經訊號無法有效傳遞至肌肉，進而造成部分動作控制功能的長期喪失。功能性電刺激能啟動殘存神經路徑以恢復運動功能，傳統多採用固定強度之低頻脈衝刺激，但是由於每位受試者在電流耐受性與神經肌肉反應都不同，因此難以在不同受試者身上實施相同刺激參數（電流強度、頻率等），也使得功能性電流刺激的效果難以評估。本研究提出一套結合高頻幅度調變刺激波形與多層次濾波技術，透過高頻調變波形將刺激能量集中於高頻載波區間，降低刺激與肌電訊號頻帶重疊的干擾，提升在刺激過程中同步觀測肌電訊號的可行性，使電流刺激狀態下之肌電訊號得以觀察，並同步評估肌肉收縮狀態，使臨床人員能透過肌電訊號了解電流刺激之成效。 實驗結果顯示，肌電訊號均方根值隨刺激電流強度顯著上升，握力反應亦明顯增幅。更重要的是，在刺激過程中，肌電訊號均方根值的變化斜率於「收縮後」階段顯著高於「收縮前」，反映肌肉對電刺激反應性的改變與神經肌肉招募增益的提升。同時，本研究發現電刺激下肌電訊號均方根值與力量輸出呈非線性關係，僅依賴力量作為回饋可能忽略肌電訊號所反映的肌肉活化狀態與神經肌肉效率等關鍵資訊。 本研究的貢獻在於開發了可於高頻調變電刺激下進行肌電訊號同步觀測系統，為臨床應用提供了精準的肌肉反應回饋，研究結果強調在電刺激應用中，應考慮個體的神經肌肉招募特性之差異，作為調整刺激參數的重要依據，為未來個人化神經復健刺激策略之設計提供重要基礎。;Stroke and spinal cord injury often lead to impaired transmission of motor neural signals to muscles, resulting in long-term loss of partial motor control functions. Functional electrical stimulation (FES) can activate the remaining neural pathways to restore motor function during rehabilitation. However, conventional FES typically employs low-frequency pulses with fixed intensity, making it difficult to apply identical stimulation parameters (e.g., current intensity, frequency) across individuals with varying current tolerance and neuromuscular responses. To address this issue, this study proposes an integrated system combining a high-frequency amplitude-modulated (HF-AM) stimulation waveform with a multi-stage filtering architecture. The HF-AM waveform concentrates stimulation energy within the high-frequency carrier band, thereby reducing spectral overlap between stimulation and electromyography (EMG) signals. This design enables simultaneous EMG observation during stimulation, allowing real-time monitoring of muscle activation and contraction states. Such synchronized observation provides clinicians with more direct physiological feedback to evaluate stimulation effects.This limitation hinders the accurate evaluation of stimulation efficacy. Experimental results demonstrated that EMG root-mean-square (RMS) values increased significantly with stimulation current, accompanied by a corresponding rise in grip force. More importantly, the slope of EMG RMS during the post-contraction phase was significantly higher than that during the pre-contraction phase, reflecting enhanced muscle responsiveness and increased neuromuscular recruitment gain under stimulation. Furthermore, a nonlinear relationship was observed between EMG RMS and force output, suggesting that relying solely on force feedback may overlook critical information regarding muscle activation and neuromuscular efficiency revealed by EMG. The main contribution of this study lies in the development of an EMG observation system capable of stable, real-time monitoring under high-frequency stimulation conditions. This system provides precise feedback on muscle responses and highlights the necessity of accounting for individual differences in neuromuscular recruitment characteristics when adjusting stimulation parameters. The findings offer a solid foundation for designing personalized neurorehabilitation stimulation strategies in future clinical applications.
