Sparse Optimized Bayesian Learning Algorithm (Sobla) Based Image Processing Methods in Alzheimer’s Disease for Cognitive Impairment

Abstract

Alzheimer’s disease (AD) is a chronic mental illness that gradually limits recall, cognitive capacity and daily pursuit. This is why an prompt detection is vital for good clinical management. The traditional diagnostic methods hinge mainly on cognitive evaluations, which often fail to detect the disease in its early stages. Afterwards, MRI has become an important tool for detecting structural brain abnormalities associated with AD and Cognitive Impairment (CI). However, the presence of noise, intensity variations, and motion artifacts in MRI scans can reduce the accuracy of automated diagnostic systems. To overcome these challenges, this study presents an improved MRI-based framework for the detection of AD and CI. The proposed method employs Mean Contrast Limited Adaptive Histogram Equalization (CLAHE) for enhancing image quality and Adaptive Median Filter (AMF) for filtering the noise and reduces the distortions. Then, the Wavelet Transform (WT) is applied to extract significant image features for analysis. To classify, a Sparse Optimized Bayesian Learning Algorithm (SOBLA) combined with Genetic Algorithm (GA)-based parameter optimization is proposed to improve predictive performance. Experimental results on ADNI dataset show that the proposed approach can achieve higher diagnostic accuracy and reliability than existing methods, and thus is suitable for computer aided diagnosis of Alzheimer ’s disease.