A novel discriminant feature selection–based mutual information extraction from MR brain images for Alzheimer's stages detection and prediction

Abstract

Alzheimer's disease (AD), a neurodegenerative disorder, is a form of dementia. Quick or early diagnosis of AD is essential, but most of the available studies have focused on clinical or survey-based data, leading to data inconsistency as many people feel hesitant or hide information due to disease and societal stigma. Nowadays, current computer-aided support and techniques are mainly based on feature extraction, but due to redundant or similarly extracted features, any specific model is not producing best performance. This paper proposes a novel mutual relationship-based feature selection model with high-altitude acute response like features for the brain-predefined feature areas using magnetic resonance imaging. The mutual relationship between features is used to avoid similar features or create discrimination between different features due to which accuracy gets affected. In this work, four machine learning classifiers, namely decision tree, support vector machine (SVM), k-nearest neighbor, and Naive Bayes are used for developing predictive model and algorithms. To avoid overfitting in model selection, we have used a 10-fold cross-validation for model training and testing. We collected data set from Alzheimer's disease Neuroimaging Initiative repository. The performance of supervised classifiers was estimated with area under the receiver operating characteristic curve, accuracy, precision, recall, and F1 score assessment measures. The results indicate SVM to be the best performing classifier for prediction (SVM recording the highest average area under curve score of 0.936, accuracy of 94.1%, precision of 96.9%, recall of 96.6%, and F1-score of 96.8%) with 37 discriminant features.

Prediction of Alzheimer's disease stages based on ResNet-Self-attention architecture with Bayesian optimization and best features selection

“…Additionally, AI-Atroshi et al ( 2022 ) utilized convolutional layers with freeze elements from ImageNet, achieving 99.27% accuracy on ADNI's MRI data collection for both binary and ternary classification. Authors in Shankar et al ( 2022 ) employed a ResNet-18 architecture using a transfer learning concept and obtained an accuracy of 83.3% on Kaggle datasets. Authors in Sharma et al ( 2022 ) utilized a CNN-based pre-trained network named ResNet-50 and achieved 91.78% accuracy.…”

“…The deep belief network (DBN) was utilized by AI-Atroshi et al ( 2022 ) to extract feature vectors from detected speech samples, which has an output accuracy of 90.2%. Shankar et al ( 2022 ) used HAAR-based object identification techniques because they are more suitable with discriminant attributes and generated 37 spatial pieces of information from seven characteristics that produced 94.1% accuracy on the dataset taken from ADNI. To aid in the initial diagnosis of AD (FDN-ADNet), Sharma et al ( 2022 ) used a DL network for all-level feature extraction from extracted sagittal plane slices of 3D MRI scans and a fuzzy hyperplane-oriented FLS-TWSVM for the classification of the retrieved features, which generated 97.29% accuracy on the publicly available ADNI dataset.…”

An intelligent hierarchical residual attention learning‐based conjoined twin neural network for Alzheimer's stage detection and prediction

“…The limitations of this model are that they have not covered all stages of Alzheimer's, and there is no feature discrimination, due to which there is less accuracy than 90% reported. Shankar et al 48 presented a novel shared correlation–based feature selection process. They used a discriminative feature selection‐initiated supervised machine learning model.…”

“…It can automatically extract efficient features for classifying Alzheimer's stages. 37,44,45 A study 47 48 proposed an idea with a feature selection model that is based on reciprocal relationships and has acute high-altitude response-like characteristics for brain regions that are predefined. In the machine learning model, feature extraction is manual, which will take too much time for Alzheimer's detection, and these models are also overfitted.…”

Predictive models in Alzheimer's disease: an evaluation based on data mining techniques

“…Ten-fold cross-validation was applied to prevent overfitting. The results revealed that SVM stands out as the most effective classifier for prediction, with an area under the curve of 0.936, an accuracy of 96.9%, a recall of 96.6%, and an F1 score of 96.8% [10]. On the other hand, another study aims to create a hybrid data mining model that integrates text mining with structured data to improve the diagnosis of dementia.…”