The chronic brain disease known as Alzheimer’s disease (AD) primarily affects short-term memory while advancing through its neurodegenerative stages. The initial symptoms of the disease develop gradually before the condition deteriorates thus early detection becomes vital. A Machine learning approach powers the Disease Diagnostic System (DDS) that analyzes T2 weighted Magnetic Resonance Imaging (MRI) scans from Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The paper examines the Hippocampus and amygdala located in the left hemisphere of the human brain as Region of Interest (ROI) which is extracted from small cohort MRI scans. The µ ± 3σ normalization method applies to segmented ROI while first-order histogram features extract Skewness and Kurtosis values. The Region of Interest receives two-dimensional wavelet features which include the Max norm of the original image and the Diagonal Detail Coefficient. The extracted textural markers serve as inputs to build a Mamdani FIS which defines minimum rules for AD stage diagnosis. The proposed classification method delivers accuracy levels of 96.13% for AD vs NC diagnosis and 94.73% for MCI vs NC diagnosis and 93.11% for AD vs MCI diagnosis. Multiple feature extraction through biomarker texture analysis enables better decision generation within the expert system framework.