Dominant hippocampus segmentation with brain atrophy analysis-based AD subtype classification using KLW-RU-Net and T1FL

Abstract

A progressive neurodegenerative disorder that leads to cognitive decline and memory loss is known as Alzheimer's Disease (AD). Researchers propose an integrated framework centered on brain atrophy analysis and hippocampus segmentation to predict AD and its subtypes. This approach augments the accurate prediction of AD subtype classification, thus addressing the lack of emphasis in prevailing works. Primarily, the T1-weighted brain Magnetic Resonance Imaging (MRI) images from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database undergo preprocessing. After that, for spatiotemporal evaluation, region segmentation, tau accumulation analysis, and Feature Extraction (FE), the image is processed. By employing the Fisher-Kolmogorov (FK) model, the essential parameters are extracted in spatio-temporal evaluation. In the meantime, for tissue segmentation, the Knowledge Partitioned Clustering (KPC) is utilized, and for hippocampus segmentation, the Kullback-Leibler Within-layer Regularized UNet (KLW-RU-Net) is employed. Moreover, by binarizing the image, the tau accumulation is analyzed; in addition, Principal Component Analysis (PCA) is utilized for FE. After PCA, the Adaptive Step-sized Gauss Rabbits Optimization (ASGRO) is employed for feature selection. Subsequently, to classify AD, the Dilated Derivative Sigmoid-Weighted DenseNet (DD-SWDN) is employed. Additionally, by employing Type-1 Fuzzy Logic (T1FL), the subtype of AD is classified, thus achieving higher accuracy for AD subtype classification. Therefore, as per these outcomes, the proposed work performs better than the other conventional approaches.