Ensemble Deep Learning Framework for Alzheimer's Disease Early Identification and Categorization Using Brain MRI

In this paper, we present a novel ensemble deep learning framework for robust AD classification based on MRI scans. The pipeline uses ConvNeXt backbone variants to improve representation, semantic segmentation to highlight important brain regions, and hybrid feature extraction to combine handcrafted and deep attention-based features. When multiple models are aggregated into an ensemble, the predictions are accurate and reliable. Area Under the Curve (AUC), Intersection over Union (IoU), accuracy, precision, and sensitivity are used to evaluate the results of the tests, which are performed using the ADNI and OASIS datasets. The results show that the suggested methodology performs better in early-stage classification than existing singlemodel approaches, particularly when it comes to distinguishing Mild Cognitive Impairment (MCI) from Normal Controls (NC). In general, the framework appears to be a good tool for helping physicians diagnose neuro degenerative disorders.