Alzheimer Disease Classification Using Custom CNN and a Hybrid Model of DcGAN with ResNet101

Alzheimer's Disease (AD) significantly impacts daily life, and manifest symptoms like memory loss, difficulty in problem-solving, confusion, language challenges, and changes in mood or personality. As the disease develops, these symptoms interrupt everyday tasks, impair relationships, and ultimately undermine independence. The suggested technique compares the outcomes of custom CNN model with a hybrid model of Deep Convolutional GAN (DcGAN) and Resnet101 model in a multi-stage manner. 1,125 labeled samples covering various stages of AD are included in the dataset, which was gathered from the Alzheimer's Disease Neuroimaging Initiative (ADNI) available on Kaggle. To tackle data shortage problem, synthetic images are generated with the DcGAN. The custom CNN framework is subsequently adjusted for AD classification by integrating these images with the actual ADNI dataset. The model is assessed using performance measures including accuracy, recall, precision, F1-score and validation accuracy. 90.6% accuracy is shown overall and demonstrates the model's ability to solve issue with data shortage by highlighting its competence in both generating along with classification tasks. This paper concludes by highlighting the prospects for furthering clinical processes and research on AD in the upcoming future.