A Novel Approach to Memory Disorder Diagnose Using Deep Learning Integration and Adaptive Artificial Approaches

Alzheimer’s illness is an untreatable brain condition that gradually erodes cognitive function, early identification can greatly lessen effects. In light of the lack of skilled doctors, the computerized identification of the illness becomes increasingly crucial since it lessens the workload on doctors and improves the accuracy in making the diagnostic. Using segmentation MRI scans, the condition must be precisely diagnosed by a thorough examination of particular brain problem tissues. The collection of pictures taken for dementia detected using MRI scans is divided into four categories: mildly informal (896 pictures), moderately informal (64 photos), not informal (3200 pictures), as well as extremely mildly informal (2240 pictures). The information available is seriously skewed. To solve this problem, we applied the adaptive artificial exceeding method. This method was used, and the information collected was matched. In order to test the effectiveness of the designs, a combination of VGG16 as well as Efficient Net was utilised to identify Alzheimer’s disease using both unbalanced and equal information. This method was used, and the information collected was normalised. The suggested approach merged numerous algorithms' projections to create a team approach that extracted intricate and subtle trends within the information at hand. Both algorithms' input and output were merged to create a combined approach, which was subsequently transferred to additional levels to create a more reliable prediction. In this project, we suggested a combination of EfficientNet-B2 and VGG-16 to accurately detect the illness at its earliest stages. Using two databases that are accessible to the public, studies were conducted. According to the findings of the study, the suggested technique obtained 97.85% efficiency and 99.64% (Area Under Curve) for multiple classes information.