Alzheimer Disease Classification Using Transfer Learning
Heta Acharya, Rutvik Mehta, Dheeraj Kumar Singh
Abstract
In recent years, transfer learning has gained huge popularity in solving problems from various fields including the medical image analysis. Clinical picture examination has altered medical services in the course of recent years, permitting specialists to discover disease earlier and improve patient recovery. Imaging has assumed a significant role in the finding of Alzheimer disease (AD). AD is a progressive neurological issue that gradually destroy memory and thinking skills of human. Initially, Computed Tomography scan (CT) and then Magnetic Resonance Imaging (MRI) were utilized to discover reasons of dementia in AD patients. This research aims to classify MRI of Alzheimer disease patients into multiple class by using VGG16, ResNet -50 and AlexNet as transfer learning models along with convolution neural networks. There are some stages of AD like mild cognitive impairment, mild Alzheimer’s, moderate Alzheimer’s and severe impairment. The proposed strategies show results with an accuracy of 95.70%, this represents a substantial improvement in accuracy over previous studies, demonstrating the efficacy of the proposed method.