الفهرس 
Chapter (1) IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Alzheimer’s disease (AD) is a degenerative brain disease and the most
common cause of dementia. The most common initial symptom is a gradually
worsening ability to remember. AD is a progressive disease, which means that it
gets worse over time. There is no cure, specific blood or imaging test for AD.
However, some drugs are available which may help slow the progression of AD
symptoms for a limited time. Diagnosis of the AD still a challenge and difficult,
especially in the early stages. The early detection will be key to slow and stop AD.
The objective of this thesis is to propose different approaches to build up a
Computer Aided Diagnosis (CAD) system for extracting the most effective and
significant features of AD from 3-D Magnetic Resonance Image (MRI) images.
Before the proposed approaches, there are two methods had been used to proof the
difficulty of the AD diagnosis. The first method is histogram plotting and entropy
calculation for normal and patient images. The second method is extracting
statistical, structural and textural features for both normal and patient images.
This thesis proposes three approaches to build up a CAD system for fast
and early diagnosis of the AD especially at the early stages where it is very difficult
to diagnose it. The first approach consists of three stages: feature selection and
reduction by discarding unnecessary features from the images, then feature
extraction by extracting the most significant features from the images, and finally
classification using Support Vector Machine (SVM) classifier. This proposed
approach compared with Principle Component Analysis (PCA) and Linear
Discriminate Analysis (LDA) techniques. The results indicate that the proposed
approach gives better classification performance as compared to the other
techniques. The first proposed approach gives high metric parameters values
(100%) with number of extracted features equal to 2000 features from more than
2 Million features. A simple Graphical User Interface (GUI) application based on
this proposed approach had been built to simplify using of the diagnosis of the AD
to the doctors. The accuracy of the diagnosed cases reaches to 100% within only 3
seconds.