逸奇科技 研討會課程

生醫訊號與時頻分析研討會

(課程編號:MS.VS01-01)

課程大綱

　　生活中充斥各種類型的訊號，如聲音、電波、壓力、震動、潮汐、磁場等，這些訊號常常挾帶人們渴望了解的各種訊息，但光憑觀察常常無法了解訊號中隱藏的各種資訊，於是產生訊號處理這門學問，藉此讓人們了解如何提取訊號中有用的資料。

　　 生醫訊號是生物體運作時產生之生理現象，藉由量測此訊號可推估生物體當下的狀態。
醫學上常用此訊號作為診斷病情的工具，例如腦電波(EEG)可推測大腦運作，心電圖(ECG)可評估心臟脈動，機電圖(EMG)可了解肌肉運用，除此之外還有很多可用來推估身體狀態的生理訊號，
這些訊號被解讀後，已被用在醫療診斷、狀態評估、傳達訊息等方面。

　　本次研討會將探討各種訊號分析的技巧應用於生理訊號的分析，包含HHT, 時頻分析，複雜度(Complexity)等。

膀胱肌電圖　資料來源：交通大學/電機與控制工程學系 陳右穎教授

課程內容

講者：王逸民博士 / 史丹佛大學航太所 / 逸奇科技總經理
講題：HHT: the theory, implementation and application

什麼是頻率？是週期的倒數，是一段時間發生的次數？還是傅立葉頻譜的數值？
什麼是瞬時頻率？照測不準原理，根本沒有這個觀念。那時頻分析所呈現頻率隨著時間變化的現象，瞬時頻率的表現如何解釋？是否不同頻率的定義有不同的解釋。
要改善這些問題，頻率的定義必需要有更基本的檢視！
黃鍔院士的 「赫伯特－黃變換法」 ( Hilbert-Huang Transformation ，簡稱 HHT ) ，採用 Hilbert 對瞬時頻率的定義，比較傅利葉分析，在處理非線性的問題上，有許多好處。 HHT 訊號處理法完全改變以往對於非線性、非穩態訊號幾乎束手無策的窘境。他跳脫出傳統數學理論的限制，因此運用 HHT 進行分析而得到的結論亦顛覆了傳統數學的思維。 HHT 在被提出後短短幾年內就被廣泛應用於科學、醫學、工程、社會、人文科學不同領域的數據分析，它的廣泛應用也導致一個革命性數據分析方法的演進。

講者：謝建興教授 / 元智大學機械工程學系
講題：Investigating complex patterns of blocked intestinal artery blood pressure signals by
　　　empirical mode decomposition and linguistic analysis

In this investigation, surgical operations of blocked intestinal artery have been conducted on pigs to simulate the condition of acute mesenteric arterial occlusion. The empirical mode decomposition method and the algorithm of linguistic analysis were applied to verify the blood pressure signals in simulated situation. We assumed that there was some information hidden in the high-frequency part of the blood pressure signal when an intestinal artery is blocked. The empirical mode decomposition method (EMD) has been applied to decompose the intrinsic mode functions (IMF) from a complex time series. But, the end effects and phenomenon of intermittence damage the consistence of each IMF. Thus, we proposed the complementary ensemble empirical mode decomposition method (CEEMD) to solve the problems of end effects and the phenomenon of intermittence. The main wave of blood pressure signals can be reconstructed by the main components, identified by Monte Carlo verification, and removed from the original signal to derive a riding wave. Furthermore, the concept of linguistic analysis was applied to design the blocking index to verify the pattern of riding wave of blood pressure using the measurements of dissimilarity. Blocking index works well to identify the situation in which the sampled time series of blood pressure signal was recorded. Here, these two totally different algorithms are successfully integrated and the existence of the existence of information hidden in high-frequency part of blood pressure signal has been proven.

講者：蕭又新副教授 / 政治大學應用物理所
講題：Detecting hidden information in ventricular fibrillation Background:

It is well known that the state of rhythm during resuscitation determines medical treatments.In particular, the study of pathophysiological processes of ventricular fibrillation (VF) has attracted considerable interest. To gain insight during the period of VF, various approaches,including instantaneous frequency measures in real time-domain as well as fluctuation analyses,are used to analyze real-life data.
Materials and Methods:
The surface ECG recordings from 35 VF patients were obtained from the physionet public website(www.physionet.org/physiobank/database). Two instantaneous frequency measures are used to study the VF signal. One is the Morlet transform method (MTM) and the other is the empirical mode decomposition (EMD). In addition, a popular approach in analyzing the nonstationary time-series, i.e., detrended fluctuation analysis (DFA), is used to detect statistical characteristics of the VF signal.
Results:
Both MTM and EMD exhibit the similar dominant frequency (DF) in a real time display.
However, some of extracted details are different in these two approaches.
DFA reveals that many VF patients display the uncorrelated property in long-term fluctuations,which indicates the stochastic factor would be essential for the development of VF.
In summary, these results suggest that pathophysiological processes during the period of VF should be from two independent sources at least. One of them can generate DF and its associated harmonic frequencies, thus, the coherent behavior is expected.The other has a tendancy to destruct the coherent behavior and finally makes a complicated manner of the VF signal.  

講者：李政杰 / 國立台灣大學機械所
講題：腦波與心電訊號之分析與判讀

ECG俗稱心電圖，藉由觀察心電圖之波形和頻率規律性，可檢查受測者有無各種心臟疾病，亦可了解受測者心臟的健康狀況。生理訊號是用來檢測生理機能的重要工具。量測因生理反應產生之訊息，經由某些方式判讀和分析，可推測個人的身體概況。

EEG俗稱腦波，是藉由量測頭皮表面的電位變化得到的生理訊號。在應用上，EEG常被用來評估大腦機能和精神狀態，例如：睡眠狀態檢測、腦部疾病檢測、疲勞狀態檢測、壓力檢測等。

議程

活動諮詢專線：(02) 8231-6888 #26 黃小姐
E-mail： mk@ancad.com