Early Diagnosis of Alzheimer’s Disease Using Adaptive Neuro K-Means Clustering Technique

Early diagnosis of Alzheimer’s Disease (AD) plays a vital role in initiating timely treatment and improving patient outcomes. This study introduces an efficient diagnostic framework utilizing Adaptive Neuro K-Means Clustering integrated with hybrid segmentation and classification techniques for brain MRI analysis. Initially, the MRI image is pre-processed through grayscale conversion, noise addition, and denoising using median filtering. Skull removal is performed to extract brain tissues, followed by segmentation using a hybrid AMSOM-FKM (Adaptive Self-Organizing Map with Fuzzy K-Means) method, which enhances regional differentiation of affected tissues. Significant features are extracted using Gray-Level Co-occurrence Matrix (GLCM) and dimensionality reduction is achieved using Principal Component Analysis (PCA). These features are classified into Mild, Moderate, and Severe AD stages using four classifiers: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, Naive Bayes and Random Forest. Among these, to calculate the accuracy with a notable precision, recall, and F1-score. Confusion matrices and performance metrics validate the robustness of the system. This proposed method demonstrates promising potential for automated and accurate Alzheimer’s stage classification, aiding neurologists in early-stage AD detection. The integration of adaptive clustering with advanced classification significantly enhances the diagnostic reliability of brain MRI analysis.

Keywords: Dataset, Machine Learning, Image Processing Techniques, segmentation and accuracy.