Automated Detection of Alzheimer’s Disease using Transfer Learning Algorithm

Alzheimer's is one of the primary causes of dementia, in which cognitive functioning deteriorates over time in aging individuals. Though early stages show some improvement, subsequent stages turn out to be irreversible. Early intervention during the mild cognitive impairment stage offers potential reversibility.The inception V3 is pre-trained model so time for training will be less over other models. Structural biomarkers, such as brain atrophy and accumulation of plaques and tangles, are directly related to the pathology of AD and can be viewed and captured using sophisticated imaging technologies like MRI and PET scans. This project seeks to create a predictive model for the diagnosis of Alzheimer's disease based on an MRI dataset of about 1,332 images. 982 images are utilized for training, and 350 for validation. The validation set comprises 100 images of mild dementia, 52 of moderate dementia, 98 of normal, and 100 of very mild dementia. In addition, the half-diagonal images of the dataset are composed of 77 mild, 49 moderate, 98 normal, and 86 very mild. The research uses transfer learning from the Inception V3 model, which is traditionally used to identify neurological diseases. The project is hoping to classify Alzheimer's disease into four classes: Normal, Very Mild Dementia, Mild Dementia, and Moderate Dementia with the aid of this model.