Pre-trained Convolutional Neural Networks and Otsu’s Multi-level Thresholding based Alzheimer’s Classification

Disorders affecting the brain and the nerves that supply the entire body are referred to as neurological diseases. Changes in the brain’s chemistry, structure, or electrical system can cause a variety of symptoms. Alzheimer’s disease (AD), a degenerative neurological condition that impairs cognitive function, is one example of this type of illness. The most prevalent type of dementia for which there is no effective treatment. Early detection and treatment, however, may help the patient’s lifestyle and reduce symptoms. Brain imaging techniques, including MRI, CT, PET, and others, are widely used to diagnose neurological problems. It has been discovered that the Artificial Intelligence based approaches, such as the deep learning algorithms and machine learning algorithms work well for assessing these photos and provide improved accuracy for identifying AD. In this paper, we classified AD and Control cases using four different Convolutional Neural Networks (CNN) versions. Additionally, in order to increase the accuracy of AD’s classification, we have segmented the images using Otsu’s thresholding technique. The outcomes demonstrate that, both before and after thresholding was implemented, the CNN AlexNet offers a significant improvement over the other deep learning models.