以數學模型分析膝關節振動訊號之嵌入式系統

人體膝蓋關節為一個控制下肢彎曲或伸張動作的重要關節，對行進和運動都有重大影響，但卻是個一般人容易忽略保養且受傷退化的關節。本研究配合臺大醫院骨科門診的幫忙，提供我們多名受測者寶貴資料與專業知識，還有藉著精良的測試設備作量測，最後由李枝宏教授與本所學生不斷地分析研究，終於發展出一系列適用於分析此膝蓋關節信號的演算法，如均方根值、自回歸模型、可適性區段、主極點功率比、平均內部群組距離等。經過取樣有效且足夠的臨床診斷經驗，本組可以用簡單非侵入式的量測方式，進而得到個體資料輸入數學模式，並求出比較各特徵參數，再使用one-way anova統計分析來判別群組間的差異性，雖然信號的時域圖與時域上的均方根值無法直接判斷該並人是否異常，但經過信號處理的技巧即可達到一定程度的區分效果。本研究討論主體是要用硬體來發展此醫療診斷儀器。因有鑑於現今的系統在演算法部分都是由軟體處理，而我們知道軟體處理運算的速度慢，尤其是大量資料運算的時候，所以本論文是以模組設計的方式來建構數位信號處理功能並加以說明。

關鍵字

膝關節振動訊號；嵌入式系統

並列摘要

We know that knee joint is important for us and we need one instrument to tell us what condition of it. We got some kinds of Bio-medical signals from hospital of National Taiwan University and we use one exact instrument to measure and store them. After that, our team have to research this signal of Bio-medical since 1995. Finally, we already developed one algorithm and we can use it to analysis signal of Bio-medical. We have used MATLAB to design this algorithm, so we can get some references of character in computer. They includes “Root Mean Square”、”spectral power ratio of dominant pole”、”one-way ANOVA”. If we get new data from hospital, we’ll let them input to computer and run MATLAB to get new references. In this flow, we still exist one issue is spending more time to analysis these signals. How to use hardware to develop new instrument is my research. We know if we want to shorten running time of algorithm, we need to use hardware to replace software. This thesis will introduce how to use Dev-C++ to design in one embedded system. After finishing that, I used the result to compare with the one before.