HD-CNN: Early-stage Alzheimer Detection system using Hybrid Deep Convolutional Neural Network

Abstract

According to recent predictions, Alzheimer disease (AD), which is presently the sixth greatest cause of impermanence in the USA, may come in third place among all causes of death for seniors, just after cancer and heart disease. It is obvious that it is crucial to identify this illness early and stop it from spreading. Numerous medical tests are necessary for the diagnosis of Alzheimer disease (AD), which generates enormous amounts of multivariate heterogeneous data. The varied nature of medical testing makes it difficult and taxing to manually compare, evaluate, and analyze this data. In this research we proposed an early-stage detection AD using hybrid deep learning algorithms. The various feature extraction and selection methods are used for extraction of potential features. The RESNET-101 and VGGNET are the deep learning frameworks that we use for classification. The YOLOv8 is used for data preprocessing as well as object detection. The RESNET-101 obtains higher 99.35% accuracy with 100 epoch size and 15 hidden layers which is higher than all experiments. In comparative analysis our model evaluation has done with VGGNET and ShallowNet, As a result our system outperforms higher result than both.