Vascular Clog Loss Classification: An Advanced Alzheimer's Research Using ConvNets

Abstract

As of today, no cure has been found that can fully treat Alzheimer's disease. One in three people above the age of 60 die from Alzheimer's disease or either dementia. Alzheimer's disease (AD) kills more people than Breast Cancer and Prostate Cancer combined. It is predicted that, by the year 2050, the total number of people being affected by Alzheimer’s will be around 14 million worldwide. Diagnosis of Alzheimer's disease at early stage would help in facilitating family planning and cost control and further reduce additional cost that might be involved in long-term care. The purpose of this study is to classify restricted and normal flowing cranial vessels in the brain using convolution neural networks (AlexNet, ResNet 50, 101, 152). Convolution neural networks have proven to have a robust architecture that can be used for extracting high-level features and further assist in image recognition and analysis. Using the AlexNet technique, we have been able to successfully classify a video (of blood flowing in a blood vessel) as stalled or flowing with an astonishing accuracy of 98.59%.