Early Diagnosis of Alzheimer’s Disease with Transfer Learning Techniques Via ResNet50 and FSBi-LSTM

Abstract

Alzheimer's Disease (AD) is a neurological disorder marked by cognitive deterioration and neurological impairment that affects cognition, memory, and behavioral patterns. Alzheimer's is an incurable disease that predominantly impacts individuals over the age of 40. A patient's MRI (Magnetic Resonance Imaging) scan and cognitive assessment are manually analyzed to diagnose Alzheimer's disease. Recently, Artificial Intelligence (AI), particularly through Deep Learning, has pioneered innovative techniques for automated medical image identification. We devised a deep learning methodology for Alzheimer's disease identification utilizing Magnetic Resonance Imaging (MRI) data. The suggested method, termed Res+FSBILSTM, employs ResNet50 as a pre-trained model for feature extraction from MRIs, thereafter identifying Alzheimer's disease through a Fully-Stack Bidirectional Long-Short Term Memory deep learning model. The experimental results demonstrate that the suggested method surpasses state-of-the-art techniques across all evaluation metrics, rendering it a viable tool for medical professionals in identifying Alzheimer's disease using brain radiological images. Ultimately, we achieved results with an accuracy of 99.6%, an F1-score of 97.7%, an area under the curve of 99%, a recall of 97.3%, and a precision of 99.6%.