Discrete Margin-Infused Laplacian Boost Technique for Detecting Early Alzheimer's Disease

Abstract

Alzheimer's Disease (AD) is a progressive neurodegenerative disorder that mostly affects older adults. It is distinguished by a slow decline in cognitive functions, including impaired reasoning, reduced problem-solving abilities, depression, and irritability. The disease results from the collection of abnormal protein deposits in the brain, leading to the death of neurons and the lessening of brain tissue over time. Recent ML techniques provide an empirical evidence-based approach for detecting the Alzheimer's disease status of patients. However, current techniques face challenges in accurately early diagnosing the disease with minimal time and error. To address these issues, a Discrete Laplacian Margin-Infused Emphasis Boost classifier (DILMIE-boost) technique is introduced for early diagnosing Alzheimer's disease with higher accuracy in less time. It comprises five processes. Initially, brain MRI images are gathered from a database. Discrete Laplacian Filter Pre-processing is carried out to remove noisy pixels from input MRI images. The Discrete Laplacian operator is a finite-dimensional graph for improving image quality. After that, Fuzzy Piecewise Linear Mumford–Shah functional algorithm-based Segmentation is carried out to partition the enhanced image into a number of sub-divisions based on triangular fuzzy membership functions. For every segmented region, RoI features are extracted for efficient classification to perform disease diagnosis. With the extracted features, the Margin-Infused Emphasis Boost classifier utilizes Margin-infused relaxed classification as a weak learner to classify the MRI images into multiple classes for performing disease diagnosis. The performance outcome of the quantitative study demonstrates that DILMIE-boost offers improved accuracy in disease prediction while also reducing time when compared to existing approaches.