An Accurate Prediction and Diagnosis of Alzheimer’s Disease using Deep Learning

Recent years have seen an increase in the amount of focus placed on the study of Alzheimer's disease (AD) categorization and prediction. Historically, one would handle such an issue by employing a combination of several types of machine learning algorithms, including regular machine learning algorithms and deep learning algorithms. However, academics are making ongoing efforts to seek for the possibilities of just employing the deep learning method. In order to accurately diagnose this disease, early diagnosis and detection are required. Due to the vast number of people who present themselves with the disease, manual diagnosis by medical professionals is fraught with the possibility of error and is extremely time consuming. The analysis and classification of AD have been approached using a variety of approaches; nonetheless, there is a pressing want for early analysis solutions that are more accurate. The model that was developed as a result of this research provides a deep-learning (DL)-based clarification for the analysis and classification of AD. This solution would make use of the DenseNet-169 &ResNet-50 CNN structures. The AD is categorized in the proposed model as not having dementia, having very mild-dementia, having mild-dementia, and having moderate-dementia. During both the train and the testing stages, the DenseNet-169 demonstrated superior performance. The accuracy values for train &test for DenseNet-169 are respectively 98.7% and 87.36%. In comparison, the accuracy for ResNet-50 were respectively 93.70% and 85.8%. The analysis and classification of AD can be done in real time using the model that was proposed.