Optimized VGG features with SpikeGoogle-Deep CNN for Alzheimer’s disease detection

Abstract

Generally, neurodegenerative syndrome referred to as Alzheimer’s syndrome affects neuron cells in a discriminating form. Gradually, the number of patients is rising, and thus, Alzheimer’s is referred to as a universal complicated issue that might be a reason for mortality in numerous scenarios. Moreover, a speedy and precise recognition and classification of Alzheimer’s disease encompass attained massive attentiveness from researchers because of studies of a deep approach. Nonetheless, effective detection of Alzheimer’s disease with accurate biomarkers is highly complicated. This paper introduces a deep learning approach for detection of Alzheimer’s disease. Primarily, input subsequently delineates calculation of filter extents of feature map, subsequent to convolution operation. Images are pre-processed by adopting process of realignment, normalization and smoothing. Subsequently, feature extraction is performed by adopting VGG-16 that is trained by gold rush optimizer algorithm. Finally, Alzheimer’s disease detection is performed by proposed hybrid deep learning model named SpikeGoogle-Deep CNN model that is developed by hybridizing SpikeGoogle and Deep Convolutional Neural Network (Deep CNN) architectures. Ultimately, experimentation is done and it states that proposed technique achieved a better accuracy of 0.953, sensitivity of 0.972 and specificity of 0.936.