Development and validation of an automatic classification algorithm for the diagnosis of Alzheimer's disease using a...

Area of Science:

• Neuroimaging
• Artificial Intelligence in Medicine
• Biomedical Data Analysis

Background:

• Alzheimer's disease (AD) and mild cognitive impairment (MCI) are progressive neurodegenerative conditions.
• Accurate and early diagnosis is crucial for effective management and treatment.
• Current diagnostic methods often rely on clinical assessments and neuroimaging, with a need for improved automated classification tools.

Purpose of the Study:

• To develop and validate an automatic classification algorithm for diagnosing Alzheimer's disease (AD) or mild cognitive impairment (MCI).
• To compare the performance of a high-performance interpretable network algorithm (TabNet) with XGBoost for AD/MCI classification.
• To investigate the utility of brain MRI volumes and radiomics features for diagnostic classification.

Main Methods:

• TabNet and XGBoost algorithms were trained on segmented brain regions (102 regions) using MRI data.
• Classification was performed for Alzheimer's disease (AD), mild cognitive impairment (MCI), and cognitively normal (CN) groups.
• Diagnostic performance was evaluated using area under the curve (AUC) metrics and validated internally and externally.

Main Results:

• TabNet achieved an AUC of 0.951 for AD vs. CN classification using volume features, comparable to XGBoost (0.953).
• Both algorithms demonstrated similar performance in classifying MCI and showed comparable results in external validation.
• The addition of radiomics features did not enhance TabNet's diagnostic performance.
• TabNet and XGBoost identified key AD-related regions, including the hippocampus and entorhinal cortex.

Conclusions:

• TabNet is a high-performance, interpretable algorithm for Alzheimer's disease detection using 3D T1-weighted brain MRI.
• The algorithm provides accurate classification and detailed interpretation of diagnostically relevant brain regions.
• This deep learning approach aids in the accurate detection of Alzheimer's disease.