Application and comparison of K-means and PCA based segmentation models for Alzheimer disease detection using MRI

Abstract

Abstract Machine learning algorithms can be used to detect Alzheimer disease with RMI-images. One of the challenges of these algorithms is to clearly extract image features that show small variants of brain cells changes, which reveal the condition of dementia at the intermediate stages of mild cognitive impairment. In this article, we explore the abilities of two approaches to diagnose Alzheimer’s disease with MRI. In the first approach, after noise reduction and correction of alterations was conducted by a non-linear filter size 3*3, a kmeans algorithm is used for segmentation of cells showing white and grey matters of brain images. A Convolutional neural network (CNN) model is trained to indicate variations on these cells and the presence of Alzheimer Disease. The second approach performs image feature reduction using principal component analysis (PCA) to extract white and grey matters, and the cerebrospinal fluid as the three relevant features for Alzheimer diagnosis. A multilayer perceptron algorithm is trained to classify Alzheimer cases. Experiments are conducted on both approaches to compare accuracy and processing time using a real dataset of 602 images from the Alzheimer’s disease Neuroimaging Initiative (ADNI) of cognitively normal and Alzheimer’s disease patients. Results show that the accuracy can be enhanced when PCA is used to extract relevant features on RMI images; and with relatively low processing time.

Keywords

• Alzheimer detection
• Image-segmentation
• Kmeans-clustering
• PCA
• Features-extraction