A comparative analysis of Alzheimer’s Disease by Pre-Trained CNN Classifiers using Two Palnes of MRI Images

Alzheimer’s Disease (AD) is a progressive neurological disorder that leads to significant deterioration in cognitive functions and memory. AD is accompanied by many symptoms like memory loss, mood swings, and asking the same thing many times. Currently, there are no specific medicines available to treat this disease in severe stages. The Alzheimer’s Association of India estimates that 7.4 percent of Indians aged 60 and above are living with dementia at this time. Therefore, it is essential to detect dementia in individuals within this age group at an early stage to provide optimal healthcare and support to those impacted by AD. Deep Learning (DL) has enhanced the precision of diagnostic processes, including clinical assessments and neuroimaging, by creating innovative techniques for the early detection of AD. This study exhibits the capability of pretrained CNN based methods in detecting and classifying various stages of AD in its early phase. Further, it has utilized Magnetic Resonance Image (MRI) data from the Kaggle AD dataset with three classes and Alzheimer’s Disease Neuroimaging Initiative (ADNI) data set with four classes. To evaluate the efficiency of the seven models, hold-out 80:20 validation is taken into consideration. The experimental findings reveal that, among the seven distinct Convolutional Neural Network (CNN) based pretrained models considered for this investigation, the pretrained Google Net with one inception block classifier has achieved the highest level of performance with two diverse datasets, achieving an accuracy of 88% and 92%, respectively