CHOSUN

Synergistic Structure of SOMs and GMKL Based on MR Images for Alzheimer's disease Diagnosis

Metadata Downloads
Author(s)
말락 아흐람
Issued Date
2016
Keyword
Computer Engineering
Abstract
Alzheimer’s disease (AD) is a complicated progressive neurodegenerative disease and the most famous type of dementia in elderly people worldwide[1] the prevalence of clinically manifest AD is about 2% at the age of 65
years but increases to about 30% at the age of 85 years, recent research [2].
Diagnosis of the AD still remains a challenge, especially in the first stages.Thus, new methods for data extraction and early diagnosis by using noninvasive methods are desired. From here the need of Computer-Aided diagnosis (CAD) methods arises. There is little research on the comparison
among machine learning approaches to find a better classification framework and further analysis of brain imaging features in the study of Alzheimer’s
disease.
To deal with this issue, we carry out an exhaustive integrative study to evaluate the efficiency of different machine learning approaches by applying them individually and by combining them together, in order to find the best
possible structure to study the disease.
In this work, many supervised and unsupervised machine learning approaches are combined in many different experiments. Among them, classification methods including (i) support vector machines (SVM), (ii)
multiple kernel learning (MKL), and (iii) generalized multiple kernel learning (GMKL). Moreover, self-organizing maps (SOM) and principal component analysis (PCA) are used as feature selection machine learning approaches.
The novelty of this work is demonstrated by the combination of the mentioned earlier methods that is newly used to serve the purpose of Alzheimer’s disease classification using MR images. Especially, the
combination of SOM as a feature selection and dimensionality reduction method, and GMKL as a classification method, in which this combo has
shown the highest classification accuracy among all other combined feature selection and classification methods used.|알츠하이머병은 만성적으로 신경퇴행성 질병이며, 전 세계 노인들이 앓는 치매 중에 가장 흔한 형태를 가지고 있다. 최근의 연구결과에서는 임상적으로 드러난 AD 유병률이 65세에는 약 2%였으나, 85세에는 약 30%까지 상승했다. 특히 초기 진단에서, 알츠하이머병의 진단은 여전히 힘들다. 그러므로
비-외과적 방식의 새로운 데이터 추출과 조기 진단 방법이 필요하다. 이 부분에서 Computer-Aided diagnosis (CAD) 방식의 필요성이 생긴다. 더 좋은 분류 체계를 찾기 위함과 알츠하이머병의 연구 중에서 뇌 영상 특징의 분석을 위한 기계학습 접근법들을 비교한 연구 조사가 있다.
이 문제를 해결하기 위해서, 철저하게 통합된 연구를 이행한다. 질병을 연구하기 위한 최선의 구조를 찾기 위하여, 서로 다른 기계학습 접근법을 각자 그리고 묶어서 효율성을 평가한다.
본 연구에서는, 많은 교사와 비교사 기계학습 접근법은 다양한
실험으로 결합된다. 그 중에서 SVM, MKL, GMKL, SOM, PCA 을 포함한 분류 방식은 특징 선택 기계학습 접근법으로 쓰인다.
이러한 연구들의 신기함은 앞에 언급된 방식(최근에 MR 이미지를 이용한 알츠하이머 병의 분류의 목적을 제공해주는데 사용됨)으로 보여진다. 특히, 특징 선택과 차원 축소 방법으로의 SOM 결합과, 분류 방법으로의 GMKL, 이 두 방법의 결합은 다른 특징 선택 과 분류 방법들에 비하여 높은 분류 정확성을 보여준다.
Alternative Title
알츠하이머병 진단을 위한 MRI 영상기반 SOM 과 GMKL의 상승 결합 구조
Alternative Author(s)
이상웅
Department
일반대학원 컴퓨터공학과
Advisor
이상웅
Awarded Date
2016-08
Table Of Contents
Acknowledgements........................................................i
TABLE Of CONTENT ............................................. iv
LIST OF FIGURES............................................ vi
LIST OF TABLES..................................................... vii
ABSTRACT .............................. x
Chapter 1................1
I. INTRODUCTION................................1
A. Alzheimer’s disease ........................1
B. Computer-aided Diagnosis (CAD) ....................3
C. Medical Image Computing (MIC)................................................................4
D. ADNI Clinical Dataset........................................5
E. Datasets and Materials......................................6
F. Thesis Contribution............................................7
Chapter 2........................................................9
II. BACKGROUND OF THE RELATED MACHINE LEARNING
THEORIES AND METHODS ...........................9
A. Kernel methods..............................................9
1. Support Vector Machine (SVM).............................................................9
2. Multiple Kernel SVM (MK-SVM)..........................................................14
B. Feature Selection and Dimensionality Reduction Methods.....................16
1. Principal Component Analysis (PCA) ................................................16
v
2. Self-Organizing Maps (SOMs) ..............................................................17
Chapter 3.........................................18
III. DEEP COMPARISON BETWEEN KERNEL CLASSIFIERS WHEN
APPLYING ON NEUROIMAGING DATA.........................................................18
A. Overview of the Experiment..............18
B. Experimental Results...........................................19
C. Conclusions ...............................................................22
Chapter 4.............................................................23
IV. APPLYING PCA AS A FEATURE SELECTION METHOD................23
A. Overview of the experiment........................23
B. Experimental Results ..........................24
C. Conclusions .................................................27
Chapter 5...............................................29
V. APPLYING SOM AS A FEATURE SELECTION METHOD...................29
A. Overview of the Experiment........................29
B. Experimental Results .................................30
Chapter 6...............................................37
VI. CONCLUSION...........................................37
BIBLIOGRAPHY........................................................38
Degree
Master
Publisher
조선대학교
Citation
말락 아흐람. (2016). Synergistic Structure of SOMs and GMKL Based on MR Images for Alzheimer’s disease Diagnosis.
Type
Dissertation
URI
https://oak.chosun.ac.kr/handle/2020.oak/12861
http://chosun.dcollection.net/common/orgView/200000265620
Appears in Collections:
General Graduate School > 3. Theses(Master)
Authorize & License
  • AuthorizeOpen
  • Embargo2016-08-25
Files in This Item:

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.