A Long Short-Term Memory Deep Learning Network for MRI Based Alzheimer’s Disease Dementia Classification
MRI data has been widely used for early detection and diagnosis of Alzheimer’s disease. This work outlines a deep learning based Long Short-Term Memory (LSTM) algorithm combined with Boruta algorithm-based feature selection approach was used to classify the Alzheimer disease MRI dataset as demented or non-demented. The wrapper based all relevant algorithm Boruta is used to select the important features from MRI data set. LSTM is a type of recurrent neural network with layered architecture to classify the datasets. The advantage of using LSTM is that it creates memory components that are for both short and long terms compared to traditional RNNs. The feature selection approach identified measures such as CDR (Clinical Dementia Rating) and MMSE (Mini mental status Examination) as the top ranking features in the Open Access Series of Imaging Studies (OASIS) MRI data set. The evaluation of LSTMbased methodology with Boruta feature selection using OASIS MRI and Alzheimer’s MRI dataset suggest that proposed methodology was able to achieve an accuracy of 94% on test dataset, exhibiting a significant increase compare to the other state-of-the-art systems.