Alzr-Net: A Novel Approach to Detect Alzheimer Disease

Alzheimer disease is the early stage of dementia that leads to loss of memory and other working skills mostly in elderly people. There is currently no specific treatment available for Alzheimer's disease, however, early detection of the disease can prevent the worsening of symptoms in patients. In this work, we used a transfer learning approach for the accurate detection of Alzheimer patients through MRI scans. We proposed a customized transfer learning approach, named as Alzr-Net, which is based on the customized Inception v3 to examine the effectiveness of Alzr-Net for the detection of Alzheimer diseases. We performed extensive experimentation using the other pretrained models and Alzr-Net and compared the performance of both type of models. The proposed Alzr-Net obtained an accuracy, precision, recall, and Fl-score of 94.38%, 97.24%, 95.49%, and 96.36% respectively. We also compared Alzr-Net with other modern techniques used for Alzheimer detection, which signified the performance of the proposed model. The results of the above-mentioned performance metrics illustrated that the Alzr-Net is an effective technique to be employed for the detection of Alzheimer patients, and this system is reliable to implement in real-time environments.