利用小波轉換於血壓及脈博訊號之異常偵測與分析

Wavelet transform in warning signal detection and analysis of blood pressure and pulse signal

張安邦(An-Pang Chang)

指導教授 ： 翁昭旼

國立臺灣大學/工學院/醫學工程學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201700279

摘要

全臺灣每年都有許多重症病患者於加護病房接受治療，但是偵測病患生理資訊的儀器時常會出現雜訊，當遇到緊急狀況時會影響醫護人員的判斷，本研究希望將加護病房的警示系統優化，讓醫護人員能夠更即時且精準的掌握病患的狀況。

為了建置此警示系統，本研究取得台灣北部醫院加護病房的真實資料，並且模擬即時狀態的生理資料序列，期望能即時或是一個時間區間內解析生理訊號來判斷是否為雜訊或是病患陷入異常狀態的訊號。

本研究嘗試尋找各種演算法與統計方法並組合各種結果來偵測訊號的漂移點，同時也利用相關係數來測試不同生理訊號間是否存在相關性，期望能透過已知數據來預測下一個時間區間內可能發生的狀況。

本研究最後的目標是從現有的病患資料中尋找各生理訊號是否存在相關性，並且當發生異常狀況時，能夠排除掉雜訊提供醫護人員更準確的警示資訊。

關鍵字

小波轉換 ； 警示系統 ； 相關係數 ； 加護病房

並列摘要

There are so many critically ill patients accept treatments in the Intensive Care Unit, but the noises are often appeared in the physiological signals monitoring which detect patients. And the judgements of the medical personnel doctors and nurses will be influenced in the emergency. This study respects that through the optimization of the warning systems in the Intensive Care Unit helps medical personnel doctors and nurses can handle the patients immediate and much more precise.

In order to build up this warning system, this study uses the real data in the Intensive Care Unit of the hospitals the north Taiwan, and simulates the physiological signals data of the patients. And hope this system can immediate or in the time zone to analysis the physiological signals whether is just the noise or occurred patients suffers in the abnormal conditions.

This study try to find many algorithms and statistical methods to combine the results, and to detect the shift points of the signals. In the meanwhile, this study also uses the correlation coefficients to detect whether the relatives are existed between different physiological signals, and hope through the data we have to prediction what happens in the next time zone.

The goal of this study is to check whether there are relatives in many physiological signals of the patients, and when the signals appear abnormal conditions, the warning system will get rid of the noises and help medical personnel doctors and nurses to get much more correct warning informations.

並列關鍵字

Wavelet ； warning system ； Correlation ； ICU

參考文獻

[5] 羅傑瀛、林彥宏、王正賢，應用灰色理論於時間序列轉折點之分析與預測，大葉學報第十一卷第二期(91)，頁115-127。

[1] Paul S Addison, The Illustrated Wavelet Transform Handbook.

[2] P. Duilleux, An implementation of the algorithme à trous to compute the wavelet transform, in Wavelets: Time-Frequency Methods and Phase Space.

[3] Mladen Victor Wickerhauser, Adapted Wavelet Analysis From Theory to Software.

[4] 廖敏治，灰色理論的GM(1,1)模型在時間數列結構轉折之研究，全球商業經營管理學報第五期，頁9-17。