[¹⁸F]FDG PET integrated with structural MRI for accurate brain age prediction

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Gerontology

Background:

• Brain aging is a complex process with structural and functional decline.
• Predicting brain age is crucial for understanding aging and neurodegenerative diseases.

Purpose of the Study:

• To develop a novel deep learning (DL) method for predicting brain age using structural and metabolic imaging data.
• To evaluate the performance of a dual-pathway DL model integrating [18F]FDG PET and structural MRI (sMRI).

Main Methods:

• A dual-pathway, 3D simple fully convolutional network (Dual-SFCNeXt) was developed to estimate brain age.
• The model utilized [18F]FDG PET and sMRI data from normal control (NC), mild cognitive impairment (MCI), and Alzheimer's disease (AD) subjects.
• Model accuracy was assessed using mean absolute error (MAE) and Pearson's correlation coefficient (r); brain age gap (BAG) was correlated with cognitive scores.

Main Results:

• The Dual-SFCNeXt model achieved high prediction accuracy (MAE = 2.37, r = 0.97) by integrating PET and MRI data.
• This multimodal approach outperformed single-modality DL models.
• Significantly higher BAG was observed in MCI and AD groups compared to NC, correlating with cognitive decline (MMSE, CDR-SB).

Conclusions:

• Integrating [18F]FDG PET with sMRI enhances brain age prediction accuracy.
• This multimodal approach offers a promising new avenue for brain age prediction studies.
• The findings highlight the potential of brain age gap as a biomarker for cognitive impairment.