Alzheimer and Mild Cognitive disease Recognition Using Automated Deep Learning Techniques

Alzheimer disease is becoming a primary reason for death, so early detection of this disease is significant. This disease is dangerous as the patient's average survival rate lies within eleven years; therefore, earlier diagnosis can save lives and improve the mortality rate. The precise cause of Alzheimer disease is still unknown; however, earlier diagnosis can control the conversion of mild cognitive impairment (MCI) into Alzheimer disease (AD). With time the condition worsens and no proper cure of disease found yet. The aim is to reduce the disease's growth, reduce the signs, address behaviour problems and improve the lifestyle. However, some short-lived treatments can reduce the risk of conversion into AD at an early stage. In this paper, a new deep learning technique is proposed; which predict and classify the disease more precisely by using structural MRI images. Convolutional neural network (CNN) were applied on slices of MRI images of ADNI dataset. CNN performance was evaluated, and it gives higher accuracy in classifying the AD, CN and MCI. The proposed technique gives remarkable results compared to the existing techniques because the model is generated using the Bayesian optimization algorithm and network morphism. Our proposed method's testing accuracy gives 99.3%, which noticeably shows our technique's effectiveness.