Alzheimer’s Disease Classification using DTCWT, PCA and Feed-forward Neural Network in MRI
- Author(s)
- 자 데베쉬
- Issued Date
- 2017
- Abstract
- Alzheimer’s disease (AD), the most familiar form of dementia, is a neurodegenerative disorder of the brain that causes memory loss to the elderly people. Magnetic resonance imaging (MRI) of the brain provides comprehensive diagnostic information for diagnosis. The error-free diagnosis of Alzheimer’s disease (AD) from Healthy control (HC) at the early stage is a major concern, because the knowledge of the severity and the development risks allows the subjects to take precautionary measures before irretrievable brain damages are shaped. Recently, there have been great interests for computer-aided diagnosis for magnetic resonance image (MRI) classification. However, identifying the distinctions between Alzheimer′s brain data and healthy brain data in older adults is challenging due to highly similar brain patterns and image intensities. Recently, cutting-edge feature extraction technologies have been very quickly expanding into numerous fields, incorporating medical image analysis. In this paper, we proposed dual tree complex wavelet transform (DTCWT) for extracting features from the image. The dimensionality of feature vector is reduced by using principal component analysis (PCA). The reduced feature is sent to feed-forward neural network (FNN) to distinguish MR images into AD and NC.
These proposed and implemented pipelines, which demonstrate an improvement in classification output when compared to other studies, resulted in high and reproducible accuracy rates of 90.06±0.1 % with sensitivity of 92.00±0.40 %, a specificity of 87.78±0.40 % and a precision of 89.6 ± .03 % with 10-fold cross validation.
|알츠하이머병(Alzheimer’s disease)은 노인들에게 기억 상실을 일으키는 뇌의 퇴행성 신경 장애로서 치매의 가장 친숙한 형태이다. 뇌의 자기공명영상(MRI)은 포괄적인 진단 정보를 제공한다. 건강관리를 통한 알츠하이머병의 오류 없는 초기 진단은 주요 관심사이며, 이는 심각한 발달 장애 및 뇌 손상으로 인한 지능 저하 전에 피험자가 사전 예방 조치를 취할 수 있게 해주기 때문이다.최근에는, 자기공명영상 분류를 위한 컴퓨터 보조 진단에 많은 관심이 일어나고 있다. 그러나 고령자의 경우에는 알츠하이머병의 뇌 데이터와 건강한 뇌 데이터의 차이를 확인할 때, 매우 유사한 뇌 패턴과 이미지 강도로 인해 많은 어려움이 있다. 최근에는 최첨단 특징 추출 기술이 의료 영상 분석을 포함하여, 다양한 분야로 빠르게 확대되고 있다.본 논문에서는 이미지로부터 특징을 추출하기 위해 DTCWT(Dual Tree Complex Wavelet Transform)를 제안한다. DTCWT를 수행한 대역의 계수들을 주성분 분석(PCA)을 사용하여 특징 벡터의 차원을 감소시킨다. 감소된 특징은 피드 포워드 뉴럴 네트워크(FNN)로 전송되어 MR 영상을 알츠하이머병과 일반영상으로 구분한다.다른 연구와 비교했을 때 분류 결과가 개선되었음을 입증하는 제안 및 구현된 파이프 라인은 민감도 92.00±0.40%, 특이성 87.78±0.40% 및 정확도 90.06±0.1%의 높은 재현성 정확도를 나타낸다. 10배 교차 유효성 확인으로는 89.6±0.03%가 실험결과에 의해 증명된다.
- Alternative Title
- MRI 영상에서 DTCWT와 PCA 그리고 Feed-forward Neural Network 알고리듬을 이용한 알츠하이머 병변 분류
- Alternative Author(s)
- Jha Debesh
- Department
- 일반대학원 정보통신공학과
- Advisor
- Goo-Rak Kwon
- Awarded Date
- 2017-08
- Table Of Contents
- TABLE OF CONTENTS
Table of Contents……….................................................................................i
List of Figures………....................................................................................iii
List of Tables……........................................................................................iv
Acronyms…………………………………………………………………...v
Abstract (ENGLISH) .............................................................................viii
한 글 요 약………..........................................................................................x
1. INTRODUCTION:............................................................................1
1.1 Overview and motivation………....................................................1
1.2 Objective ........................................................................................2
1.3 Contribution.....................................................................................2
1.4 Structure of the thesis…………………………………………….3
2. BACKGROUND:.............................................................................4
2.1 Alzheimer’s disease……………………………………………….......4
2.2 Detection Techniques………………………………………….7
2.3 Magnetic Resonance Imaging………………………………….7
2.4 MRI pros and Cons………………………………………….....8
2.5 Accuracy of MRI……………………………………………....8
2.6 Image preprocessing and normalization…………………….....9
3. LITERATURE REVIEW………………………….......................12
4. WAVELET TRANSFORM…………………………………….....13
4.1 2D-Discrete wavelet Transform………………………………13
4.2 Dual-tree complex wavelet Transform…………………….....15
4.3 Principal component analysis……………………………........19
4.4 Probabilistic principal component analysis…………………..21
4.5 Feed-forward neural network…………………………………22
4.5.1 Training method…………………………………….23
4.6 Performance estimation……………………………………….26
4.7 Cross validation ……………………………………………....27
5. PROPOSED METHOD………………………………….............28
6. OVERVIEW OF THE EXPERIMENTAL DATA……………..32
7. EXPERIMENT, RESULTS AND DISCUSSIONS……………..35
7.1 Parameter estimation for S…..………......................................35
7.2 Feature extraction…………………………………….............35
7.3 Feature reduction………………………………………..........36
7.4 BPNN training ……………….................................................37
7.5 Statistical analysis………………………………….……........39
7.6 Performance evaluation…………………………….………...39
7.7 Comparison to other state-of-the-art approaches……………..41
7.8 Computational time……………………………………….......43
8. CONCLUSION AND FUTURE RESEARCH………………….45
REFERENCES……….........................................................................46
LIST OF PUBLICATION……………………………………………52
- Degree
- Master
- Publisher
- Chosun University, Department of Information and Communication
- Citation
- 자 데베쉬. (2017). Alzheimer’s Disease Classification using DTCWT, PCA and Feed-forward Neural Network in MRI.
- Type
- Dissertation
- URI
- https://oak.chosun.ac.kr/handle/2020.oak/13242
http://chosun.dcollection.net/common/orgView/200000266254
-
Appears in Collections:
- General Graduate School > 3. Theses(Master)
- Authorize & License
-
- AuthorizeOpen
- Embargo2017-08-25
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.