A 3D CNN Model Framework for Early Identification and Classification of Alzheimer’s in Brain MRI Images

Abstract

Alzheimer’s disease stands as the most prevalent and debilitating form of dementia among elderly individuals. Despite its prevalence, there exists a notable challenge in promptly and accurately identifying elderly individuals afflicted with Alzheimer’s disease. The primary objective of this study was to enhance the prediction and diagnosis accuracy of Alzheimer’s disease by integrating the image classification approach with the expertise of radiologists. The focus was on achieving accurate diagnosis at the initial stage or Mild Cognitive Impairment (MCI) stage, where interventions are potentially more effective. In this research, a Directed Acyclic Graph (DAG)-based 3D Convolutional Neural Network (CNN) was employed to diagnose Alzheimer’s disease, Mild Cognitive Impairment, and normal control subjects more precisely and efficiently. The methodology involved extracting pertinent features from brain MRI images, specifically focusing on the Left and Right hippocampus regions, cerebral spinal fluid, white matter, and grey matter. Prior analysis, the MRI images underwent preprocessing using a 3D slicer and SPM12. The datasets utilized in this study were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The integration of the DAG-based 3D-CNN approach with expert radiological knowledge yielded promising results. The model demonstrated enhanced accuracy in diagnosing Alzheimer’s disease, Mild Cognitive Impairment, and normal control subjects. By leveraging key features extracted from brain MRI images, the method showcased improved performance in identifying individuals at the early stages of cognitive impairment. Combining advanced image classification techniques with the expertise of radiologists holds significant promise in improving the accuracy and timeliness of Alzheimer’s disease diagnosis. The findings of this study underscore the potential of utilizing innovative approaches to enhance early detection and intervention strategies for Alzheimer’s disease, ultimately contributing to more effective therapeutic outcomes and improved patient care.