Abstract

A person with Alzheimer’s disease (AD) experiences a gradual decline in brain function and a total loss of cognitive capacity. Currently, it is said that differentiating between AD stages is a difficult task because many of these stages overlap. Hence, a recently suggested strategy for Alzheimer’s disease classification using Competitive Swarm Coot Optimization in a the CSCOOT_CNN Convolutional Neural Network does well when trained via transfer learning. Input photographs are sourced from the ADNI collection before moving on to the pre-processing stage. One of the pre-processing modules uses a median filter to get rid of noise. After that, feature extraction is performed on the pre-processed image, specifically for convolutional neural network (CNN) feature extraction. The next step is to feed the collected features into the classification level, where the disease is labelled using convolutional neural networks (CNNs) with transfer learning. In this case, the CNNs are trained using hyper parameters from the visual geometry group 19 (VGG19) model. Furthermore, the suggested CSCOOT algorithm is used to fine-tune CNNs that incorporate transfer learning. In addition, a new method called CSCOOT, which combines the COOT optimizer with Competitive Swarm Optimizer (CSO), has been introduced. Therefore, with values of 0.926 for accuracy, 0.979 for specificity, and 0.909 for sensitivity, the suggested CSCOOT_CNN achieved better results.