Detection and Classification of Alzheimer's disease from cognitive impairment with resting-state fMRI

Detection and classification of Alzheimer's disease (AD) are a demanding field of research in medicine throws light on innovative approach in detecting and classifying AD from cognitive impairment with resting-state functional magnetic resonance imaging (rsfMRI). The goal of this research is chiefly aimed to diagnose mild cognitive impairment (MCI) patients who essentially need support for medical intervention. A new concept is presented in classifying AD and MCI from rsfMRI using Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. The images are preprocessed using some advanced technique to eliminate noise and parameter variations, and the preprocessed images are used for extracting the raw features. The rsfMRI is applied for feature selection processes in order to reduce feature dimensions using principal component analysis (PCA). The proposed kernel-based PCA-support vector regression (SVR) includes t-distributed stochastic neighbor embedding (tSNE) and polynomial kernel-based tSNE that are separately handled by significantly merging correlated local and class features. The kernel PCA method analysis the new features explicitly based on nonlinear mapping function in the data points of high-dimensional search. The kernel PCA method is suitable to analysis the new feature and feature importance in AD classification. The proposed kernel SVR method has the advantage of effectively analyzing the high-dimensional data to provide linear relationship and suitable to apply in MCI and AD data. The PCA method is applied for feature reduction process due to its capacity to select the relevant features and effectively analyzing the individual features. The proposed kernel-SVR method has the advantage of selecting the relevant features and avoid overfitting problem in the classifier. The SVR uses reduced features that are obtained from different reduction methods for classification of AD and MCI, a polynomial kernel based. The results showed that the proposed kernel-based PCA-SVR showed better average accuracy values 98.53% for kernel PCA when compared the existing models hippocampal visual features of 79.15% and deep neural network of 80.21%

Automated summary

Detection and classification of Alzheimer's disease (AD) using resting-state functional magnetic resonance imaging (rsfMRI) is a demanding field of research. This study proposes a novel approach for classifying and detecting AD and mild cognitive impairment (MCI) using rsfMRI. The study preprocesses the images and extracts raw features, then applies PCA and SVR for feature selection and classification. The results show that the proposed method outperforms existing models in terms of classification accuracy.