Leveraging Bi-Focal Perspectives and Granular Feature Integration for Accurate Reliable Early Alzheimer's Detection
Shravan Venkatraman, Pandiyaraju V, Abeshek A, Pavan Kumar S, Aravintakshan S A
TL;DR
The paper tackles early Alzheimer’s disease detection from MRI, a task hindered by subtle biomarkers. It introduces a Bi-Focal Perspective CNN with Granular Feature Integration (BFPCNN) that fuses multi-scale features and focuses on key biomarkers via attention mechanisms. The authors demonstrate near-state-of-the-art performance on the OASIS dataset, reporting validation metrics around 99% across accuracy, precision, recall, and F1, and validate robustness against class imbalance through augmentation. This approach offers a fast, accurate, and potentially interpretable framework to aid clinical decision-making in AD diagnosis.
Abstract
Being the most commonly known neurodegeneration, Alzheimer's Disease (AD) is annually diagnosed in millions of patients. The present medical scenario still finds the exact diagnosis and classification of AD through neuroimaging data as a challenging task. Traditional CNNs can extract a good amount of low-level information in an image while failing to extract high-level minuscule particles, which is a significant challenge in detecting AD from MRI scans. To overcome this, we propose a novel Granular Feature Integration method to combine information extraction at different scales along with an efficient information flow, enabling the model to capture both broad and fine-grained features simultaneously. We also propose a Bi-Focal Perspective mechanism to highlight the subtle neurofibrillary tangles and amyloid plaques in the MRI scans, ensuring that critical pathological markers are accurately identified. Our model achieved an F1-Score of 99.31%, precision of 99.24%, and recall of 99.51%. These scores prove that our model is significantly better than the state-of-the-art (SOTA) CNNs in existence.