Alzheimer's Imaging Consortium

Area of Science:

• Neuroimaging
• Radiology
• Biomedical Engineering

Background:

• Arterial spin labelling (ASL) MRI enables non-invasive measurement of cerebral blood flow (CBF).
• Regional hypoperfusion on ASL MRI is linked to Alzheimer's disease (AD) progression before atrophy becomes apparent.
• This study investigates ASL's utility in diagnosing AD and tracking its progression.

Purpose of the Study:

• To assess the diagnostic accuracy of ASL MRI for Alzheimer's disease (AD) and mild cognitive impairment (MCI).
• To evaluate ASL MRI's capability in predicting MCI conversion to AD.
• To compare the performance of ASL MRI with other imaging modalities for AD assessment.

Main Methods:

• Utilized a subset of Alzheimer's Disease Neuroimaging Initiative (ADNI-3) data with diagnostic labels (cognitively normal, MCI, AD).
• Processed ASL images using the qasl pipeline to generate calibrated CBF maps.
• Trained support vector machine (SVM) classifiers using ROI-based CBF, covariates, and scanner type, with feature selection driven by ANOVA.

Main Results:

• Classifiers achieved high accuracy in distinguishing AD from cognitively normal (CN) or MCI (AUCs of 0.957 and 0.936, respectively), with high specificity (0.98).
• Classification between CN and MCI was less performant (AUC 0.702).
• ASL MRI demonstrated strong predictive power for MCI conversion to AD (AUC > 0.9 for all timeframes).

Conclusions:

• ASL MRI measures can effectively predict clinical diagnosis and disease progression in Alzheimer's disease.
• The performance of ASL MRI in this study surpassed that of PET and previous ASL classifiers.
• ASL MRI represents a promising non-invasive tool for AD diagnosis and progression monitoring.