Automated Multimodal Fusion Technique for the Classification of Human Brain on Alzheimer’s Disorder

Abstract

Alzheimer's Disorder (AD) may permanently impair memory cells, resulting in dementia. Researchers say that early Alzheimer's disease diagnosis is difficult. MRI is used to detect AD in clinical trials. It requires high discriminative MRI characteristics to accurately classify dementia stages. Due to the large extraction of features, improved deep CNN-based models have recently proven accurate. With fewer picture samples in the datasets, over-fitting issues arise, limiting the effectiveness of deep learning algorithms. This research article minimizes the overfitting error due to fusion techniques. This hybrid approach is used to classify Alzheimer's disease more accurately than other traditional approaches. Besides, the Convolutional Neural Network (CNN) provides more minute features of small changes in MRI scan images than any other algorithm. Therefore, the proposed algorithm provides great accuracy in the region of sagittal, coronal, and axial Mild Cognitive Impairments (MCI) in the brain segment classification. Moreover, this research article compares the proposed algorithm with previous research output that is used to help prove its superiority. The performance metrics uses Health Subject (HS), MCI, and Mini-Mental State Evaluation (MMSE) to evaluate the proposed research algorithm.