Structural magnetic resonance imaging (sMRI) is a vital tool for diagnosing neurological brain diseases. However, sMRI scans often show significant structural changes only in limited brain regions due to localised atrophy, making the identification of discriminative features a key challenge. Importantly, the human brain exhibits inherent bilateral symmetry, and deviations from this symmetry—such as asymmetric atrophy—are strong indicators of early Alzheimer’s disease (AD). Patch-based methods help capture local brain changes for early AD diagnosis, but they often struggle with fixed-size limitations, potentially missing subtle asymmetries or broader contextual cues. To address these limitations, we propose a novel augmented reality (AR)-enhanced patch-level explainable deep learning (ARE-PaLED) system. It includes an adaptive multi-scale patch extraction network (AMPEN) to adjust patch sizes based on anatomical characteristics and spatial context, as well as an informative patch selection algorithm (IPSA) to identify discriminative patches, including those reflecting asymmetry patterns associated with AD; additionally, an AR module is proposed for future immersive explainability, complementing the patch-level interpretation framework. Evaluated on 1862 subjects from the ADNI and AIBL datasets, the framework achieved an accuracy of 92.5% (AD vs. NC) and 85.9% (AD vs. MCI). The proposed ARE-PaLED demonstrates potential as an interpretable and immersive diagnostic aid for sMRI-based AD diagnosis, supporting the interpretation of model predictions for AD diagnosis.
6 March 2026 in International Journal of Computational Intelligence Systems
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by structural brain changes, often analysed using longitudinal 3D structural Magnetic resonance imaging (sMRI) data. Existing deep learning-based disease progression models face limitations in capturing long-term structural changes by handling class imbalance, and generating realistic future sMRI scans. However, very few works consider predicting future brain MRI images within four years. To address these challenges, a novel Adaptive Regression and ROI-enhanced Generative Adversarial Network (ARE-GAN) is proposed for AD progression prediction with the generation of real future MRI images. This framework integrates an improved generator conditioned on baseline (BL) MRI, an ROI-based attention mechanism to focus on critical brain regions, and an adaptive regression model for precise disease stage estimation. A dedicated feature fusion module is introduced to integrate extracted features from 3D sMRI data and clinical BL information. A discriminator refines the generated sMRI images by distinguishing them from real future MRIs by ensuring realistic spatial fidelity. Additionally, a Self-attention integrated DenseNet (SA-DenseNet) is used as the classification model to categorise the generated MRIs into AD, stable-MCI (s-MCI), progressive-MCI (p-MCI), and normal cognition (NC). From the experimental analysis, the proposed ARE-GAN model attains accuracy of 99.25% based on the ADNI dataset. The proposed ARE-GAN model framework increases the overall accuracy by 12.78%, 8.76%, 7.88% and 11.51% better than 3D CNN, ResNet50V2, 3D ResNet with a CBAM and EfficientNet, respectively.