跳到主要內容

臺灣博碩士論文加值系統

(44.222.189.51) 您好!臺灣時間:2024/05/18 19:23
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:柯正雲
研究生(外文):Ko, Cheng-Wen
論文名稱:腦電圖棘波自動辨識系紮之研製
論文名稱(外文):Implementation of Automatic Spike Detection System for EEG
指導教授:詹國禎詹國禎引用關係鍾孝文---
指導教授(外文):Gwo-Jen JanHsiao-Wen Chung
學位類別:碩士
校院名稱:國立臺灣大學
系所名稱:電機工程學系研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:1998
畢業學年度:86
語文別:中文
論文頁數:50
中文關鍵詞:棘波偵測腦電圖半徑基準函式類神經網路事件矩陣
外文關鍵詞:spike detectionelectroencephalographyradial basis functionartificial neural networkincidence matrix
相關次數:
  • 被引用被引用:1
  • 點閱點閱:277
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:1
本研究乃提出一套以類神經網路為架構的腦電圖多腔道棘波自動辨識系統。類神經網
路採用半徑基準函式來進行單一腔道之辨識,並經由 ROC分析方法來達到模型之最佳
化。波形簡化可除去大量的雜訊,並可因此將特徵值之數目減至三個來當作類神經網
路的輸入。為了降低誤判之機率,我們將多腔道之間的幾何相關性也以事件機率的方
式來納入考慮。類神經網路所輸出的結果必須經由閾值的判斷以完成棘波之分類。閾
值的選擇必須同時考慮靈敏度與挑選度。在經過臨床人員的比對之後,四個病人之十
六腔道連續腦電圖記錄裡面,百分之八十三的棘波可以被辨識出來,然而,誤判的機
率卻偏高。整個系統運算的時間相當短暫,幾乎可視為即時的辨識。我們相信,這個
系統是相當穩定的,而且,對於即時處理的潛力也是不容忽視的。
An automatic spike detection algorithm for classification of multi-channelelec
troencephalographic (EEG) signals based on artificial neural network is presen
ted. Radial basis function (RBF) neural network was chosen for single channel
recognition, with model optimization using receiver operating characteristics
analysis. Waveform simplification was employed for high noise immunity. Fea
ture extraction with as few as three parameters was used as preparation for th
e inputs to the neural network. Identification of multi-channel geometric cor
relation was performed to further lower the false-positive rate by using an in
cidence matrix. Threshold value for spike classification was chosen for simul
taneous maximization of detection sensitivity and selectivity. Evaluation wit
h visual analysis in this preliminary study showed a 83% sensitivity using 16-
channel continuous EEG records of four patients, while a high false positive r
ate was found, which was believed to arise from the extensive andexhaustive vi
sual analysis process. The computation time required for spike detection was
significantly less than that needed for online display of the signals on the m
onitor. We believe that the algorithm proposed in this study is robust and th
at the simple structure of RBF neural network yields high potential for real-t
ime implementation.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top