English  |  正體中文  |  简体中文  |  Items with full text/Total items : 68069/68069 (100%)
Visitors : 23154225      Online Users : 205
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version

    Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/81925

    Title: 應用深度學習之睡眠分期辨識
    Authors: 吳佳俊;Wu, Jia-Jyun
    Contributors: 電機工程學系
    Keywords: 睡眠腦波;深度學習;小波轉換;長短期記憶;雲端;穿戴式;brain wave;deep learning;wavelet transform;Long Short-Term Memory(LSTM);cloud;wearable
    Date: 2019-10-15
    Issue Date: 2020-01-07 14:38:45 (UTC+8)
    Publisher: 國立中央大學
    Abstract: 患有睡眠障礙的患者可能影響他們的整體健康,安全和生活品質。睡眠不佳與高血壓、動脈粥樣硬化(膽固醇阻塞的動脈)、心力衰竭、心臟病發作、中風、糖尿病和肥胖有關。然而,監測睡眠狀況需要醫生檢查大量的腦電圖數據,需要花費大量的時間,可能會有主觀判斷的問題。因此,開發用於自動且準確地檢測EEG數據中睡眠狀況的客觀平台是很重要的。在本文中,我們開發了一種基於深度學習的腦電數據睡眠分析系統。我們採用深度學習網路進行睡眠EEG分析,受測者的EEG進行小波轉換將訊號的頻帶特徵取出,再將其八個特徵值(最大值、最小值、中位數、平均值、標準差、全距、偏度、峰度)放入深度學習網路來做分析,其網路架構為LSTM,經由其網路所得到得準確度高達到83%,將取出八個特徵值以及直接時頻分析的結果做比較,發現取出八個特徵的效果比較好。接下來的目標為運用在硬體上,搭配雲端醫療系統,更能降低醫療資源,進而達到完善的治療以及資料的建檔。;Patients who suffer from sleep disorder can affect their overall health, safety and quality of life. Poor sleep has been linked with high blood pressure, atherosclerosis (cholesterol- clogged arteries), heart failure, heart attack,stroke , diabetes, and obesity. However, monitoring of sleep conditions requires physicians to examine huge amount of EEG data which is time-consuming and might fall into the pitfall of subjective judgement. Accordingly, development of an objective platform for automatic and accurate detection of sleep condition in EEG data is important. In this paper, we developed a deep learning based sleep evaluation system for EEG data. We adopted deep learning techniques to analyze sleep EEG data. The EEG data recorded from Fp1, Fp2,Pz and Oz positions were firstly transformed into temporal-frequency domain using wavelet analysis as training features. Eight characteristic features (maximum, minimum, median, average, standard deviation, range, skewness, kurtosis) on temporal-frequency domain were used as input data to train a LSTM network. The accuracy of the network is 83%. Comparing the eight eigenvalues and the results of the direct time-frequency analysis, we found that the performance of extracting eight features is better. The next goal is to use on hardware and the cloud medical system. The development of this sleep EEG analysis technique can reduce medical resources and facilitate the documentation of subject’s sleep history for better treatment in the future.
    Appears in Collections:[電機工程研究所] 博碩士論文

    Files in This Item:

    File Description SizeFormat

    All items in NCUIR are protected by copyright, with all rights reserved.

    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback  - 隱私權政策聲明