使用多感測器系統實作手指顫抖分析並應用於神經退化疾病

Abstract

手指顫抖對於神經退化疾病來說是一個很重要的特徵，透過手指顫抖的情形可以幫助醫師判斷疾病。Tremogram 是目前醫院所使用的偵測儀器，其原理是使用加速規來做偵測，並且將獲得的數據經由快速傅立葉轉換來得知手指顫抖的頻率。但是 Tremogram 目前用的是單一且單軸的加速規，對於病患來說，無法一次就做全面性的手指顫抖檢查。

之前有論文實作 Leap Motion 與 Tremogram 的比較。但是經過 Leap Motion 版本的更新後，敏感度較先前的版本低。因此本論文主要使用多感測器系統的G-sensor來做手指顫抖的資料蒐集，並且設計演算法來衍伸出更多的資料分析。

論文中分析了三位介於 24 ~ 28 歲的正常人及兩位有手指顫抖病徵的病人。由實驗結果發現，使用多感測器系統得到的分析結果很接近目前醫院使用的 Tremogram，而且還可以一次得到五支手指頭在空間中三維的資料，也藉著不同的演算法將顫抖程度量化。本論文成功的利用較低成本的機器實作出比 Tremogram 更多元、並且比 Leap Motion 敏感度更高的分析結果。期待本論文所實作出來的系統可以進一步應用在臨床醫學上。

Finger trembling is an important feature for neurodegenerative disease, through the trembling of the fingers can help the doctor to determine the disease. Tremogram is currently used by the hospital detection equipment. Then transfer the data to frequency domain by using fast Fourier transform to get the frequency of finger trembling. However, Tremogram is currently using a single and uniaxial acceleration gauge, which make patients, can not do a comprehensive finger trembling check.

There is a research comparison of Leap Motion and Tremogram. But after update the new version of Leap Motion.The sensitivity is lower then the previous version. Therefore, the purpose of my research is to explores the other tools to do finger trembling data collection, and design algorithms to analysis the data.

There are three normal people and two patients with finger tremors as experimental subjects in my research. From the experiment results, it is found that not only the analysis result obtained by using the multi-sensor system is very close to Tremogram, but also my system can quantize the amplitude of all fingers trembling on 3-axis at once. My research succeeded in using lower-cost machines to produce more diverse information than Tremogram and more sensitive than Leap Motion. It is expected that the system can applied in the clinical in the future.