Classification of Alzheimer's Disease using Attention UNet-MobileNet Model

Alzheimer's disease (ALD) is a degenerative and incurable neurological condition that impacts many individuals globally. Early identification and precise diagnosis of ALD are essential for efficient therapy and control of the condition. This research presents a transfer learning-based method for diagnosing ALD using magnetic resonance image (MRI) data. This research introduced a deep learning (DL) model known as MobileNet-v2 to accurately classify ALD using MRI images. The classification task involves the use of MIRIAD, a dataset consisting of MRI images specifically related to ALD. The collection comprises 708 MRI pictures pertaining to ALD and its different phases. The MRI pictures undergo preprocessing and are then split into training and testing sets, with 70% allocated for training and 30% for testing. Following the preprocessing stage, the segmentation procedure is carried out with the Attention UNet architecture. The segmented images are inputted into the MobileNet-v2 model for ALD classification. The classification is conducted as a binary classification, categorizing the data as either the presence or absence of ALD. The research model's performances are evaluated based on metrics including accuracy, specificity, sensitivity, f1-scores, and precision. The test results were compared to the existing models for validation. The research model has demonstrated superior performance when compared to all other models.