Biomarkers.

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• The advent of disease-modifying Alzheimer's drugs has increased demand for amyloid PET scans.
• Amyloid PET scans are essential for determining eligibility for Alzheimer's disease treatments.
• ¹⁸F-fluorodeoxyglucose (FDG)-PET imaging offers a potential alternative for screening.

Purpose of the Study:

• To evaluate an algorithm's capability in predicting amyloid accumulation from FDG-PET images.
• To assess the utility of FDG-PET as a screening tool for amyloid PET scans.
• To support the increased demand for Alzheimer's disease diagnostics.

Main Methods:

• Analysis of 194 subjects with cognitive disorders using FDG-PET, amyloid PET (¹¹C-PiB), and MRI.
• Application of a support-vector machine (SVM) algorithm to FDG-PET images standardized to a 120-region anatomical template.
• 10-fold cross-validation was employed to assess predictive accuracy.

Main Results:

• The algorithm achieved 81.5% accuracy, 78.5% sensitivity, 84.6% specificity, and an AUC of 0.846 during training.
• Validation results showed 85.9% accuracy, 88.4% sensitivity, 81.0% specificity, and an AUC of 0.918.
• The model demonstrated strong performance in predicting amyloid accumulation.

Conclusions:

• The developed algorithm demonstrates adequate performance for predicting amyloid accumulation from FDG-PET images.
• This approach shows promise for use in screening for amyloid PET scans.
• The findings support the use of FDG-PET as a viable tool in Alzheimer's disease diagnostics.