基於共空間模式與通道相關同步分析之前期癲癇預測;Pre-seizure Prediction Based on Common Spatial Pattern and Channel-Dependent Synchronization Analysis

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Electrical Engineering > Electronic Thesis & Dissertation > Item 987654321/89892

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/89892

Title:	基於共空間模式與通道相關同步分析之前期癲癇預測;Pre-seizure Prediction Based on Common Spatial Pattern and Channel-Dependent Synchronization Analysis
Authors:	王裕森;Wang, Yu-Sen
Contributors:	電機工程學系
Keywords:	癲癇;腦電圖;共同空間形樣法;支持向量機;Epilepsy;Electroencephalography;common spatial patterns;support vector machine
Date:	2022-07-28
Issue Date:	2022-10-04 12:03:43 (UTC+8)
Publisher:	國立中央大學
Abstract:	腦部為人類重要的器官之一，當其中一部份發生問題將造成患者極大的威脅，例如大腦受損的話將導致智力退化、記憶力喪失；小腦受損則是有小腦萎縮症Spinocerebellar Ataxia(SCA)，將造成運動失調等；而癲癇發作(Epilepsy)為上述疾病中的共同現象，會造成反覆抽蓄、失神等症狀，導致患者在生活中遇到危險。本論文藉由擷取少量電極之腦電圖(Electroencephalography, EEG)訊號，基於共空間形樣法(Common Spatial Patterns, CSP)、功率譜密度(Power Spectral Density, PSD)以及大腦電極通道間的同步相關性做為特徵，之後使用支持向量機(Support Vector Machine, SVM)進行分類。最後再藉由投票法進行最終的分類，以降低癲癇誤報率，提前預測癲癇發生。論文中以公開資料集CHB-MIT腦電圖數據庫進行演算法測試驗證，實驗分別在前期長度10分鐘以及30分鐘的情況下得到預測率91.3%、誤報率0.097以及預測率92.9%、誤報率0.108，可有效預測癲癇的發生。 ;Brain is one of the most important organs of human beings. When a problem happens in one part of it, it will pose a great threat to the patient. For example, cerebrum injury will cause intelligence degeneration, memory loss; cerebellum injury will cause movement disorders, called Spinocerebellar Ataxia(SCA). While epilepsy is the common phenomenon of above disorders, lead to repeated withdrawal, absence and other symptoms cause patients to encounter danger in life. In this paper, by extracting EEG signals with few channels, based on Common Spatial Patterns(CSP), Power Spectral Density(PSD) and the synchronous correlation between brain electrode channels as features, then using Support Vector Machine(SVM) for classification. Finally using the voting method for the final classification, reducing the false alarm rate of epilepsy and predict the occurance of epilepsy in advance. Public dataset, CHB-MIT EEG database is used in this paper for algorithm verification. The result shows that the sensitivity reach 91.3%, 92.9% and average false prediction rate 0.097, 0.108 when preictal length is set to 10mins and 30mins respectively, which can predict seizure from happen efficiently.
Appears in Collections:	[Graduate Institute of Electrical Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	58	View/Open

社群 sharing

Loading...