Generating MRI-Derived CSF Proxy-Markers to Predict and Visualize Alzheimer's Disease Progression

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Biomarker Discovery

Background:

• Early Alzheimer's disease (AD) detection is vital for clinical trials and patient benefit.
• Current methods often involve invasive procedures like lumbar punctures or radiopharmaceutical injections.
• Noninvasive neuroimaging offers a safer alternative for detecting AD biomarkers and understanding disease progression.

Purpose of the Study:

• To develop a noninvasive method for predicting Alzheimer's disease (AD) risk using structural MRI (sMRI).
• To train neural networks to generate sMRI-based representations as proxies for cerebrospinal fluid (CSF) biomarker status.
• To assess the clinical utility of this approach for AD screening, diagnosis, and progression prediction.

Main Methods:

• Utilizing neural networks to analyze structural MRI (sMRI) data.
• Generating latent sMRI representations as proxies for cerebrospinal fluid (CSF) biomarker status.
• Applying these models for classification, prediction, and risk staging in Alzheimer's disease.

Main Results:

• Neural networks successfully generated predictive sMRI representations for AD biomarkers.
• Key brain regions including the amygdala, hippocampus, and cingulate gyrus showed high prognostic value for AD risk.
• The approach demonstrated potential for screening, diagnosis, and predicting AD progression.

Conclusions:

• AI-driven analysis of sMRI can serve as a noninvasive proxy for AD biomarkers.
• This method aids in early AD detection, patient stratification for clinical trials, and predicting disease progression.
• Noninvasive neuroimaging offers a promising avenue for improving Alzheimer's disease management and research.