In recent years, the efficacy of deep learning models in accurately estimating brain age using structural magnetic resonance imaging (MRI) images has been extensively utilized. This study employs a self-attention-based convolutional neural network (CNN) to extract features from preprocessed MRI slices. While CNNs have shown remarkable performance in brain age prediction, they often fail to capture global dependencies within images. A self-attention mechanism is integrated into the CNN architecture to discern long-range relationships within the MRI images and enhance feature extraction robustness. Subsequently, a single hidden layered random vector functional link (RVFL) network is employed to predict the age of healthy individuals. The discrepancy between true age and predicted age, termed the brain age gap, serves as a biomarker for the early diagnosis of neurological disorders. Furthermore, the proposed age estimation framework is evaluated using an Alzheimer’s and Parkinson-affected dataset, demonstrating its versatility and potential for clinical applications.