MR Image Feature Analysis for Alzheimer’s Disease Detection Using Machine Learning Approaches
Journal Title: Information Dynamics and Applications - Year 2023, Vol 2, Issue 3
Abstract
Alzheimer’s disease (AD), a progressive neurological disorder, predominantly impacts cognitive functions, manifesting as memory loss and deteriorating thinking abilities. Recognized as the primary form of dementia, this affliction subtly commences within brain cells and gradually aggravates over time. In 2023, dementia's financial burden for elderly adults aged 65 and older was projected to reach \$345 billion, encompassing health care, long-term care, and hospice services. Alarmingly, Alzheimer's disease claims one in three seniors, outnumbering combined fatalities from breast and prostate cancer. Currently, the diagnostic landscape for Alzheimer's lacks definitive tests, and diagnoses based purely on biological definitions have been observed to possess low predictive accuracy. In the presented study, a diagnostic methodology has been proposed using machine learning models that harness image features derived from brain MRI scans. Specifically, nine salient image features, grounded in color, texture, shape, and orientation, were extracted for the study. Four classifiers — Naïve-Bayes, Logistic regression, XGBoost, and AdaBoost — were employed, as the challenge presented a binary classification scenario. A grid search parameter optimization technique was employed to fine-tune model configurations, ensuring optimal predictive outcomes. Conducted experiments utilizing the Kaggle dataset, and for each model, parameters were rigorously optimized. The XGBoost classifier demonstrated superior performance, achieving a test accuracy of 92%, while Naïve Bayes, Logistic Regression, and AdaBoost registered accuracies of 63%, 70%, and 72%, respectively. Relative to contemporary methods, the proposed diagnostic approach exhibits commendable accuracy in predicting AD. If AI-based predictive diagnostics for AD are realized using the strategies delineated in this study, significant benefits may be anticipated for healthcare practitioners.
Authors and Affiliations
D. S. A. Aashiqur Reza, Sadia Afrin, Md. Ahsan Ullah, Sourav Kumar Kha, Sadia Chowdhury Toma, Raju Roy, Lasker Ershad Ali
K-Means Clustering Algorithm Based on Improved Differential Evolution
The traditional K-means clustering algorithm has unstable clustering results and low efficiency due to the random selection of initial cluster centres. To address the limitations, an improved K-means clustering algorithm...
Enhancing Pneumonia Diagnosis with Transfer Learning: A Deep Learning Approach
The significant impact of pneumonia on public health, particularly among vulnerable populations, underscores the critical need for early detection and treatment. This research leverages the National Institutes of Health...
Critical Factors Influencing Cloud Security Posture of Enterprises: An Empirical Analysis
This study examines the aspects that can impact an organization's cloud security posture and the consequences for their cloud adoption strategies. Based on a thorough examination of existing literature, a conceptual fram...
An Optimized Algorithm for Peak to Average Power Ratio Reduction in Orthogonal Frequency Division Multiplexing Communication Systems: An Integrated Approach
The impact of the peak to Average Power Ratio (PAPR) on the efficiency of an Orthogonal Frequency Division Multiplexing (OFDM) communication system is significantly mitigated through an innovative Reconfigurable Integrat...
Enhanced Method for Monitoring Internet Abnormal Traffic Based on the Improved BiLSTM Network Algorithm
The complexity and variability of Internet traffic data present significant challenges in feature extraction and selection, often resulting in ineffective abnormal traffic monitoring. To address these challenges, an impr...