An Ensemble CNN for Brain Age Estimation Based on Hippocampal Region Applicable to Alzheimer's Diagnosis

Alzheimer's disease is a progressive and insidious condition that causes both structural and functional changes in the brain. Researchers are increasingly interested in early detection of disease using medical imaging. Our study proposed a model to estimate brain age using a convolutional neural network (CNN). The proposed method estimates the age of the right and left hemispheres, focusing on the respective hippocampal region and ensemble the results. We used T1-weighted magnetic resonance images (MRI) from 764 participants in the ADNI dataset, aged 70 to 96 years, targeting regions associated with Alzheimer's disease in older adults. By concentrating on the right and left hippocampi as input data, we developed an ensemble deep-learning (DL) approach with a mean absolute error (MAE) of 3.21 for cognitively normal individuals (CN) and 5.42 for those with Alzheimer's Disease (AD). Our analysis of the brain age gap (BAG) revealed a significant difference <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathbf{P} &lt; \mathbf{0.05})$</tex> between the two groups. Furthermore, our model simplified computational complexity using fewer hyperparameters than previous studies. Overall, our model has the potential to aid in diagnosing Alzheimer's disease.