Early Diagnosis of Alzheimer’s Disease using Soft Computing Based Deep Learning Technique
Early diagnosis of Alzheimer’s disease is a significant clinical challenge but it has wherewithal to know the disease before the onset of its symptoms. It has potential to stop disease progression before it reveals symptoms besides enabling treatment avenues. Towards this end, in this paper, we proposed a framework known as Soft Computing Based Deep Learning Framework for Alzheimer’s Disease Prediction (SCDLF-ADP). The framework has provision for deep learning techniques such as Convolutional Neural Network (CNN), Probabilistic Auto Encoder (PAE) and the combination of these two known as CNN-PAE. The three models are used for empirical study. The dataset used for the experiments has structural Magnetic Resonance Imaging (MRI) images. It is known as Alzheimer’s Disease Neuroimaging Initiative (ANDI) dataset. The models learn clinically relevant and latent variables such as cerebellum, neocortex and brain stem for confidently detecting the Alzheimer’s disease. The performance of the models is compared in terms of cross-entropy, accuracy and F1 score. A prototype application is built using Python data science platform to evaluate the proposed framework. The experimental results revealed that CNN-PAE showed better performance over the other two baseline methods such as CNN and PAE.
Index Terms – Soft computing, Alzheimer’s disease prediction, CNN, Probabilistic Auto Encoder