以多參數評估之穿戴式步態對稱性評估系統開發

Abstract

中風所引起的偏癱現象會造成步態不對稱，以及病患生活上的不便，為了恢復患者生活水平，在臨床上主要以下肢復健來解決此問題，而患者是否適合復健、要以何種方式復健以及復健後之成效如何，都需要由步態評估來獲得，而各種步態參數中，對稱性為重要項目之一。現在有許多研究以慣性感測器來量化步態對稱性，此外已有研究將步態擺盪期與站立期之時間長進行對稱性分析，但都尚未有研究將步態運動學參數的擺盪期與站立期分開計算對稱性。根據步態這兩階段所使用之肌肉不同，以及這兩階段不對稱所代表之意義不同，分別對步態兩階段進行對稱性分析應可提供臨床更多患者的復健資訊。 因此本研究目的為開發一穿戴式步態對稱性評估系統，內含慣性感測器模組、藍芽模組以及鋰電池，可對擺盪期與站立期進行快速對稱性分析，以做為臨床步態對稱性評估之參考。在驗證本研究開發之系統方面，本研究以動作擷取系統同步慣性感測器，計算兩系統所得之大小腿矢狀面角速度，並以標準方均根誤差計算兩系統間誤差。在程式方面，本研究根據同步之動作擷取系統，設計一可經由簡單操作辨識出步態兩階段之程式。在對稱性分析方面，使用三種對稱性計算方法對兩腳大小腿矢狀面的旋轉角速度分別對站立期與擺盪期進行訊號分析，此三種方法分別為互相關係數(cross-correlation)、面積差異(area difference)與時間差異(time difference)。左右腳各一步可得4對訊號，分別為站立期之大腿、站立期之小腿、擺盪期之大腿以及擺盪期之小腿的左右腳訊號，一對訊號以三種量化法進行對稱性量化可得3種對稱值，因此 4對訊號可得12項對稱值。本研究進行八位無中風之健康男性(24±2歲，172±6公分，67±7公斤)以及兩位中風偏癱患者(男，62歲，172公分，73.5公斤，Brunnstrom stage: V；女，47歲，160公分，65.5公斤，Brunnstrom stage: III)進行步態實驗。每位受測者行走3趟，每趟取中間4步進行對稱性分析，因此可得到12步之對稱性平均值與其標準差。 本系統所得之角速度參數與動作擷取系統所得之角速度參數進行比較，兩系統所量得之誤差在5~8%之間，為可接受範圍。對稱性分析結果顯示，Brunnstrom stage為V之中風偏癱患者，以互相關係數計算所得之大腿對稱性，落在健康者的95%信賴區間範圍內，而以其他方法計算所得之對稱性表現皆比健康者低且落在健康者的95%信賴區間範圍外，顯示互相關係數有時無法辨識不對稱現象。而Brunnstrom stage為III之中風偏癱患者，12個對稱值都顯示較健康者低且在健康者的95%信賴區間範圍外。 本研究成功研發一輕便的穿戴式步態對稱評估系統，可由簡單的程式操作，得到步態在不同階段之對稱值。此系統可對大小腿之擺盪期與站立期，進行較人為評估更為仔細且客觀的對稱性評估，可供臨床評估人員作為參考，此外本系統還可進一步辨識出患者不同程度之對稱性差異。

Hemiplegia is commonly caused by stroke and usually leads to an asymmetrical gait. Both before and after rehabilitation, gait assessments need to be made to determine whether the patients can do rehabilitation or how effective the rehabilitation program was. Among all gait assessments, symmetry is an important gait characteristic. Previous researchers have used inertial sensors to quantify gait symmetry, but there has been no previous study using whole waveform patterns of one stride to quantify gait symmetry in swing phase and stance phase, separately. Based on different muscle activation sequences and different implications in two phases, it is significant to quantify them separately. Therefore, we aim to develop a wearable sensor system to determine gait symmetry of thighs and shanks in these phases, separately. This easily programmable and lightweight system can be used to evaluate symmetry immediately and be used by therapists easily. An eight-camera motion capture system was used to verify the angular velocity obtained by the IMU of our developed system in sagittal plane, and we used normalized root-mean-square error to calculate the error between the two systems. Furthermore, we evaluated gait symmetry against not only thighs and shanks but also stance phase and swing phase. Three approaches were used to evaluate gait symmetry: cross-correlation, area difference and time difference. Cross-correlation and area difference are used to evaluate the waveform patterns generated by two legs over a period of time. Four pairs of angular velocity signal, which are measured from thigh in stance phase, thigh in swing phase, shank in stance phase and shank in swing phase, can be obtained in one stride of two legs. A pair of signal can be calculated to three kinds of symmetry values by IV three approaches. Eight intact healthy males (24±2 yrs, 172±6cm, 67±7kg) and two hemiparetic subjects (one male, 62yrs, 172cm, 73.5kg, Brunnstrom stage: V; one female, 47yrs, 160cm, 65.5kg, Brunnstrom stage: III) were recruited to attend the experiments. Every subject completed three trials at comfortable speed. Four strides in the middle of gaits were evaluated in each trial. The errors between motion capture and inertial sensors are 5~8%, and these values are acceptable. In clinical test, the results of the subject in Brunnstrom stage V show that the symmetry values are lower than the 95% confidence interval (CI) of intact subjects, except the values of cross-correlation. It might represent that cross-correlation is less sensitive to asymmetry gait. Besides, all the symmetry values of the hemiparetic subject in Brunnstrom stage III are less than the symmetry values of the other hemiparetic subject and the intact subjects. The values are all lower than the 95% CI of the intact subjects. This research has successfully developed a wearable sensor system to determine gait symmetry in stance phase and swing phase, separately. The system can be used easily to evaluate gait symmetry more objective and more quantitative to be provided in clinical use.