Alzheimer's Disease Detection Using Deep Learning: A Googlenet-Based Classification Framework

Alzheimer's Disease is a progressive neurodegenerative disease that significantly alters cognitive functions, making early detection imperative for the proper treatment and management of this condition. In this paper, we described a deep learningbased method of GoogLeNet-Based automated classification of Alzheimer's disease using neuroimaging data based on a deep hierarchical structure called GoogLeNet architecture. GoogLeNet is formed by inception modules that are known to extract and classify features efficiently through complex spatial patterns within brain MRI images. The model will therefore be trained and further fine-tuned on a dataset of diverse stages of Alzheimer's: normal, MCI, and AD-affected cases. This work shows the feasibility of deep convolutional neural networks in medical imaging for reliable early diagnosis of Alzheimer's, contributing to better clinical decision-making and patient care. The predicted GoogleNet's performance metrics include a 93.65 % accuracy rate and recall values of $0.7,0.80645$, 0.97461, and 0.4 for MildDemented, ModerateDemented, NonDemented, and VeryMildDemented, respectively. Experimental results show that the proposed GoogLeNet-based framework achieves high classification accuracy as compared to traditional machine learning methods.