Hippocampus Region's Volume-Based Alzheimer's Stages Detection Using a Deep Learning Model

Abstract

Alzheimer's disease (AD) is a kind of Dementia. It affects the brain functions, thinking ability and creates memory loss. Each stage of AD is worsening the symptoms and also affects the patient's daily activities. The current diagnosis AD detection process is not providing an accurate report for the early stages of AD. Therefore, accurate Alzheimer detection is an open challenge for the researchers. In this research, a deep learning model is introduced to improve Alzheimer's stage diagnosis. The Hybrid deep model is designed to detect abnormal changes in brain (Magnetic Resonance Imaging) MRI scans. It utilizes the multi-class log loss (MCLL) function as the objective function to reduce the error rate in detecting AD stages. The MCLL approach computes the variations in actual and detected AD stages of each biomarkers features of input MRI images to identify loss rate. It helps to reduce the loss rate in AD stages detection. Moreover, the Hybrid deep learning model for Alzheimer stages detection (HDLMASD)system analysis the images deeply to provide accurate biomarkers detection with the help of all essential biomarkers processing steps. The efficiency of the Alzheimer stages diagnosis system is evaluated with various evaluation metrics. Finally, a comparative analysis is made with multiple present diagnosis systems to verify the performance of the AD detection system. The performance analysis proves that the AD stages diagnosis approach accuracy rate is improved up to 0.47% and the prediction error rate reduced up to 0.49% than comparison approaches.