在骨科的臨床上，發現當膝關節發生病變時，其活動時會產生異常的聲音，即膝關節在擺動下所產生的振動訊號，vibration arthrometry（VAM）即是藉由分析此一振動訊號來診斷膝關節的病變。而本研究即是針對退化性關節炎的振動訊號來分析。由於VAM是一種非侵襲性的檢查工具，因此極具發展潛力。因此本研究利用時域(time domain)及頻域(frequency domain) 數學理論為獨立成份分析及經驗模態建立訊號的特徵，對照後找出退化性關節炎的特徵。我們發現膝關節於站立及蹲下所產生的振動訊號可以用來區分正常者與退化性關節炎患者。本研究首次嘗試將獨立成份分析及經驗模態分析引入退化性關節炎振動訊號的診斷，並針對所提供之ICA及HHT 及受測姿勢-位置的選擇上加以驗證，來區別退化性關節炎患者與正常者具有最大的差異性。關節振動測量術是一種非侵襲性且簡單方便低成本的膝關節病變診斷工具。本項技術的持續發展，將可成為醫生診斷時另一種重要的工具，藉由選用適當的治療方式，不僅解除了病患的痛苦，也可以避免醫療資源的浪費。本研第二部分究使用希伯特黃轉換技術診斷膝蓋關節振動號，在實驗過程中結合希伯特黃轉換方法作驗證。且所提出的方法確實可以有效運用於膝蓋關節退化之非侵入性診斷。A phenomenon can be found which abnormal joint sound arises from knee joint disorder during knee motion in the clinical diagnosis. The knee joint could produce vibration signals from a standing position to a squatting position, and the vibration arthrometry（VAM）could diagnose the disorders of the knee joint by analyzing these vibration signals. In this study we will apply VAM to the patients of the normal and degenerative arthritis. Because VAM is a noninvasive diagnostic tool, it has great potential. The main methods in the thesis we apply VAM to the vibration signals of the normal and degenerative arthritis, utilizing Independent Component Analysis (ICA) and Empirical Mode Decomposition (EMD) to establish the mathematical model after adaptive decomposition, and try to find out the characteristic parameters of the vibration signals in these diseases. Vibration arthrometry (VAM) provided a noninvasive, simple, and cheap clinic tool for diagnosing knee joints. Appropriate therapy can be given to the patients with correct diagnosis. The performance of the combined ICA/HHT technique is verified experimentally. The experimental results show that proposed ICA/HHT approach has better recognition performance than that obtained using other traditional methods.
