Improved Alzheimer's Disease Detection by MRI Using Multimodal Machine Learning Algorithms.

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) is the most common form of adult-onset dementia.
• Traditional AD diagnosis relies on clinical criteria with limited accuracy (85%) and requires post-mortem confirmation.
• Advancements in Magnetic Resonance Imaging (MRI) and machine learning (ML) offer potential for earlier and more accurate AD detection.

Purpose of the Study:

• To develop and evaluate a machine learning framework for classifying Alzheimer's disease (AD) versus non-AD subjects.
• To leverage longitudinal brain MRI features for improved diagnostic accuracy.
• To compare the performance of six supervised learning classifiers for AD detection.

Main Methods:

• A supervised learning framework was implemented using longitudinal brain MRI data.
• Six distinct supervised classification algorithms were employed for AD subject categorization.
• Patient demographic information and pre-existing conditions were considered to enhance classifier performance.

Main Results:

• The gradient boosting algorithm demonstrated superior performance among the evaluated models.
• The proposed framework achieved a high accuracy of 97.58% in classifying AD subjects.
• The study highlights the potential of ML applied to MRI for precise AD diagnosis.

Conclusions:

• Machine learning models, particularly gradient boosting, show significant promise for accurate and early Alzheimer's disease diagnosis using MRI.
• Integrating demographic and clinical data can further enhance the predictive power of ML classifiers for AD.
• This approach offers a non-invasive method to aid in the definitive diagnosis of AD, potentially improving patient management.