Efficient Multi-Class Alzheimer's Detection Using Fuzzy and Decision Tree Algorithms

Abstract

Alzheimer's disease (AD) possesses advanced neurological state which reasons brain atrophy. Early identification and action to postpone the onset of dementia may be beneficial. This research offers a novel method for classification for AD. Initially, an input image is preprocessed using a decision tree and fuzzy algorithm. The following step was utilizing an adaptive fuzzy-based atom search optimizer to segment the filtered images into parts of the brain that were linked to cerebrospinal fluid, grey matter and white matter. Registration of the GM with filtered images was performed using an improved affine transformation after image segmentation. Subsequently, the features were extracted utilizing hybrid wavelet Walsh and enhanced Zernike techniques. Adaptive rain optimization was then used to choose the features. Lastly, a capsule autoencoder and hybrid equilibrium optimizer (HEOCAE). MATLAB was the implementation platform utilized in this investigation. The strategy provided accuracy (99.60%), sensitivity (97.31%), precision (97.45%), specificity (98.64%), area AUC (98.29%), and receiver operating characteristic curve (ROC) (0.93).