Classification and Visualization of Alzheimer\u2019s Disease using Volumetric Convolutional Neural Network and Transfer Learning

Woo-Sung Kim,Kanghan Oh,Young-Chul Chung,Ko Woon Kim,Il-Seok Oh

doi:10.1038/s41598-019-54548-6

Abstract

Recently, deep-learning-based approaches have been proposed for the classification of neuroimaging data related to Alzheimer’s disease (AD), and significant progress has been made. However, end-to-end learning that is capable of maximizing the impact of deep learning has yet to receive much attention due to the endemic challenge of neuroimaging caused by the scarcity of data. Thus, this study presents an approach meant to encourage the end-to-end learning of a volumetric convolutional neural network (CNN) model for four binary classification tasks (AD vs. normal control (NC), progressive mild cognitive impairment (pMCI) vs. NC, stable mild cognitive impairment (sMCI) vs. NC and pMCI vs. sMCI) based on magnetic resonance imaging (MRI) and visualizes its outcomes in terms of the decision of the CNNs without any human intervention. In the proposed approach, we use convolutional autoencoder (CAE)-based unsupervised learning for the AD vs. NC classification task, and supervised transfer learning is applied to solve the pMCI vs. sMCI classification task. To detect the most important biomarkers related to AD and pMCI, a gradient-based visualization method that approximates the spatial influence of the CNN model’s decision was applied. To validate the contributions of this study, we conducted experiments on the ADNI database, and the results demonstrated that the proposed approach achieved the accuracies of 86.60% and 73.95% for the AD and pMCI classification tasks respectively, outperforming other network models. In the visualization results, the temporal and parietal lobes were identified as key regions for classification.

Highlights

Deep-learning-based approaches have been proposed for the classification of neuroimaging data related to Alzheimer’s disease (AD), and significant progress has been made
Since the progressive MCI (pMCI) vs. stable MCI (sMCI) classification task is more difficult and subtler than the classification of AD/NC, we employed a transfer learning procedure, for which the visual representations drawn from the classification task of AD/NC were transferred to the pMCI vs. sMCI learning model to improve the generalization capability of the pMCI vs. sMCI classification task
For the unsupervised learning-based models, we set up the following experiments: AD vs. NC, pMCI vs. NC, sMCI vs. NC, pMCI vs. AD and sMCI vs. AD

Summary

Introduction

Deep-learning-based approaches have been proposed for the classification of neuroimaging data related to Alzheimer’s disease (AD), and significant progress has been made. To slow the progression of dementia, timely treatment is crucial, which requires the early diagnosis of AD and its prodromal stage, mild cognitive impairment (MCI) To this end, a reliable diagnosis from brain imaging is necessary, and a robust diagnostic system aided by the analysis of neuroimaging data would allow for a more informative and reliable approach and could potentially increase diagnostic accuracy. DL models allow a system to use raw data as input, thereby allowing them to automatically discover highly discriminative features in the given training data set[12] This end-to-end learning design philosophy is the fundamental basis of DL. It is difficult to expect a reliable explanation of how the network reaches a classification decision

Objectives

Methods

Results

Discussion

Conclusion