Alzheimer Disease Detection of 3D-CNN with SE-Net Model using SVM Classifier

Abstract

Alzheimer disease is a fatal progressive neurological brain disorder. Earlier detection of Alzheimer's disease can help with proper treatment and prevent brain tissue damage. In this work we proposed two methods. First, proposed connected median filter using PSO feature extraction from MRI images and Analysis of Alzheimer’s diseases state by using 3D-CNN based SE-Net. In the first phase, the algorithm first normalizes and removes skull from the MRI images. Connected median filter using Particle Swarm Optimization algorithm is used to partition the image into white matter (WM), grey matter (GM) and black holes (BH). The relevant diagnostic features are extracted from the segmented image component. The classifier is trained by the training data to predict the test data. The features are defined to construct classification model by using Support Vector Machine with Squeeze- Excitation block. Here, database contains total of 1000 images which are resized into 350 × 350 without loss of information. Deep Learning demands large number of images and its strength was increased as per requirement by augmentation technique. In the first phase of the method takes 1000 images of different features are selected to train SVM classifier and the accuracy obtained is 98.37%  and contribution of this work is classification of images into categories such as Alzheimer (AD) and normal. First phase of work emphasized program specific applications to extract features.  In the second phase the CNN multiple layers which are studied from lower level to the higher-level image characteristics.