Early Diagnosis of Alzheimer’s Disease using Unsupervised Clustering | ||
| International Journal of Intelligent Computing and Information Sciences | ||
| Article 15, Volume 20, Issue 2, December 2020, Pages 112-124 PDF (650.9 K) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/ijicis.2021.51180.1044 | ||
| Authors | ||
| Yasmeen Farouk* 1; Sherine Rady2 | ||
| 1Information Systems Departement, Faculty of Computer and Information Sciences, Ain Shams University, Cairo, Egypt. | ||
| 2Ain Shams University | ||
| Abstract | ||
| Alzheimer's disease (AD) is a progressive brain disorder and a very common form of dementia. Neuroimaging techniques, such as Magnetic Resonance Imaging (MRI), produce detailed 3-dimensional images of the brain showing insights for amyloid deposits and inflammatory alterations as disease markers. The early diagnosis of AD using MRI provides a good chance for patients to prevent further brain deterioration by stopping the loss of nerve cells. This paper explores the use of unsupervised clustering approaches for the early diagnosis of AD. Though it is very common to use classification techniques for identifying medical diseases, the lack or the inaccuracies of labeled data can generate a problem. In this work, the k-means and k-medoids are compared while employing the Voxel Based Morphometry (VBM) features extracted from the MRI images. The effect of choosing certain local regions of interest (ROIs) for the analysis is also compared to the global whole-brain analysis. The results show that the proposed approach can perform an early diagnosis of AD with an accuracy of 76%. | ||
| Keywords | ||
| Unsupervised Learning; Clustering; Regions of Interest (ROI); Alzheimer’s disease; Magnetic Resonance Imaging (MRI) | ||
|
Statistics Article View: 706 PDF Download: 783 |
||
