Feature integration of [18F]FDG PET brain imaging using deep learning for sensitive cognitive decline detection.

backgroundDistinguishing individuals with cognitive decline (CD), including early Alzheimer's disease, from cognitively normal (CN) individuals is essential for improving diagnostic accuracy and enabling timely intervention. Positron emission tomography (PET) captures metabolic brain alterations associated with CD, but its broader application is often limited by cost and radiation exposure. To enhance the clinical utility of PET while addressing data limitations, we propose a data-efficient framework that integrates complementary multi-scale PET representations at voxel-level and region-level.

methodsVoxel-level features were extracted using convolutional neural networks (CNN) or principal component analysis networks (PCANet) from [¹⁸F]FDG PET imaging. Region-level features were derived from standardized uptake value ratio measurements across predefined brain regions and processed using a deep neural network (DNN). These voxel- and region-level information are integrated through direct concatenation. For the final prediction, different machine learning models and ensemble technique were applied. The models were trained and validated using 5-fold cross-validation on PET scans from 252 participants in the Alzheimer's Disease Neuroimaging Initiative, comprising 118 CN and 134 CD subjects. Additional correlation analysis and disease classification comparison with the Mini-Mental State Examination (MMSE) were also performed.

resultsIn 5-fold cross-validation, CNN, PCANet, and DNN models achieved classification accuracies of 0.69 ± 0.04, 0.69 ± 0.06, and 0.82 ± 0.06, respectively. The integrated DNN-CNN model using direct concatenation yielded the highest accuracy (0.87 ± 0.05), with a 6.33% improvement in accuracy and reduced standard deviation relative to the DNN-only model. Overall, there were an increase of 14.22% in Recall (0.77 to 0.88) and an increase of 7.92% in F1-Score (0.82 to 0.88). Moreover, the predicted probability of CD showed a significant correlation with MMSE scores, and the model achieved higher accuracy, recall, and F1-score than MMSE-based classification.

conclusionCombining complementary voxel-level and region-level PET representations with deep learning improved classification performance over single-representation models, particularly by enhancing sensitivity to cognitive decline. These findings support the potential utility of multi-scale FDG-PET representations for machine learning-based cognitive decline detection.