Predicting conversion from MCI to AD by integrating rs-fMRI and structural MRI

Area of Science:

• Neuroimaging
• Neurology
• Radiology

Background:

• Alzheimer's disease (AD) diagnosis relies on identifying early pathological changes.
• Structural MRI (sMRI) and resting-state functional MRI (rs-fMRI) show potential for AD detection.
• Integrating sMRI and rs-fMRI for early AD detection remains underexplored.

Purpose of the Study:

• To evaluate the diagnostic performance of sMRI and rs-fMRI, individually and combined.
• To classify patients with mild cognitive impairment (MCI) who convert to probable AD (MCI-C) from those who do not (MCI-NC).
• To identify optimal neuroimaging features for early AD detection.

Main Methods:

• Utilized cortical and subcortical measurements from sMRI (e.g., cortical thickness).
• Extracted graph theory measures from rs-fMRI functional connectivity.
• Trained a support vector machine classifier using selected sMRI and rs-fMRI features.

Main Results:

• Single-modality classification achieved 89% accuracy for sMRI and 93% for rs-fMRI.
• The combined multi-modality approach (sMRI + rs-fMRI) reached 97% accuracy.
• The algorithm effectively identified MCI converters using a minimal set of optimal features.

Conclusions:

• Integration of sMRI and rs-fMRI significantly enhances the accuracy of early AD detection.
• This multi-modal approach offers a promising tool for identifying individuals at high risk of AD progression.
• This study represents the first investigation into combining sMRI and rs-fMRI for early AD identification.