English  |  正體中文  |  简体中文  |  Items with full text/Total items : 43312/67235
Visitors : 2108582      Online Users : 16
RC Version 5.0 © Powered By DSPACE, MIT. Enhanced by NTU/NCHU Library IR team.
National Chung Hsing University Institutional Repository - NCHUIR > 工學院 > 通訊工程研究所 > 依資料類型分類 > 碩博士論文 >  非監督式多發性硬化症偵測和分類方法在腦部核磁共振成像

Please use this identifier to cite or link to this item: http://nchuir.lib.nchu.edu.tw/handle/309270000/153926

標題: 非監督式多發性硬化症偵測和分類方法在腦部核磁共振成像
Unsupervised MS Lesion Detection And Classification Method on Brain MRI Images
作者: 張智凱
Chang, Chih-Kai
Contributors: 歐陽彥杰
Yen-Chieh Ouyang
通訊工程研究所
關鍵字: 自動目標物產生過程;純度像素索引;限制能量最小化法;正交子空間投影法
Automatic target generation process;pure pixel index;constrained energy minimization;orthogonal subspace projection
日期: 2013
Issue Date: 2013-11-19 12:32:01 (UTC+8)
Publisher: 通訊工程研究所
摘要: 分類一般而言需要一系列的訓練點,分為監督式或是非監督式的方法取決於從資料中訓練點的產生方式是預先得知或是沒有預先得知資訊。非監督式法在沒有任何預先資訊的情況下來幫助分類的程序比起監督式法更具有挑戰性。為了達成分割需要一系列訓練點來進行分類,在本論文中我們使用了兩種非監督式的演算法,自動目標產生過程和純度像素索引來找尋訓練點。然後應用了限制能量最小化法將多發性硬化症從腦部核磁共振影像中分類出來,限制能量最小化法在沒有得知資料的模型和背景雜訊統計下只需要最少的子樣本數便可以有很好的表現,接著使用正交子空間投影法來做其餘的組織分類。實驗結果顯示本方法在腦部核磁共振影像中有很好的效果。
The classification generally requires a set of training samples, which can be carried out in a supervised or an unsupervised manner depending upon how training samples are produced a priori using prior knowledge or a posteriori obtained directly from the data. Unsupervised methods which do not assume any prior scene knowledge can be learned to help classification process are obviously more challenging than the supervised ones. In order for segmentation to perform classification, a set of training samples is required. In this thesis we present two unsupervised target detection methods, automatic target generation process (ATGP) and pure pixel index (PPI), to find training samples. Then we apply constrained energy minimization (CEM) method to classify multiple sclerosis (MS) lesion for MR brain image. This CEM method is perfectly used due to the fact that it requires the least amount of information about subsample target of interest without making assumptions on signal model and noise/background statistics. After that the orthogonal subspace projection (OSP) method is also applied to classify the rest of tissues. Experimental results show that these approaches have great promise in MR brain image classification.
Appears in Collections:[依資料類型分類] 碩博士論文

Files in This Item:

File SizeFormat
index.html0KbHTML124View/Open


 


學術資源

著作權聲明

本網站為收錄中興大學學術著作及學術產出,已積極向著作權人取得全文授權,並盡力防止侵害著作權人之權益。如仍發現本網站之數位內容有侵害著作權人權益情事者,請權利人通知本網站維護人員,將盡速為您處理。

本網站之數位內容為國立中興大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用。

聯絡網站維護人員:wyhuang@nchu.edu.tw,04-22840290 # 412。

DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU/NCHU Library IR team Copyright ©   - Feedback