Efficient Alzheimer’s disease segmentation on MRI brain image and classification using machine learning

In the recent past, Computer Aided Diagnosis (CAD) technologies plays vital role in computer vision in the field of medical diagnosis. Few nervous disorders and describing compulsive area are analyzed, and structural human organ brain is undergone for research with the support of magnetic resonance imaging (MRI) imaging. Standard classification machine learning algorithms are being used in analysis for the automated diagnosis of Alzheimer’s disease (AD), and machine learning-based methods have become a common alternative for AD diagnosis. State-of-the-art approaches that take diagnosis into account have been shown to be more accurate than conventional evaluation. Obtaining information from various approaches, on the other hand, is time-consuming and costly, and certain methodologies could have energy adverse effects. Our research is based on computerized automated Magnetic Resonance Imaging (MRI). In this research, automatic object segmentation and classification of Alzheimer disease (AD) is proposed. The collected MRI brain images contain various noises such as salt and pepper noise, speckle noise and Gaussian noise. The 2D Adaptive Median Filter (AMF) is proposed to filter all types of noises. The filtered image is further enhanced to improve the quality of image. The significant Edge Preservation-Contrast Limited Adaptive Histogram Equalization (EP-CLAHE) algorithm is used to improve the quality of image in terms of contrast and brightness. The AD region from MRI brain image is further clustered using Efficient Fuzzy based C Means Clustering (EFCMC). The clustered AD region is further segmented using Adaptive Otsu Thresholding (AOT). The significant features are calculated using Gray Smooth Co-occurrence Vector (GSCV). The 1 Dimensional Iterative Convolutional Neural Network (1D I ) is used to classify the AD stages. The experimental results show that the proposed methodology performance is better than conventional methodologies.