Detection of Alzheimer's Disease Using Graph-Regularized Convolutional Neural Network Based on Structural Similarity Learning of Brain Magnetic Resonance Images

Objective: This paper presents an Alzheimer's disease (AD) detection method based on learning structural similarity between Magnetic Resonance Images (MRIs) and representing the similarity as a graph. Methods: We construct the similarity graph using embedded features of the input image (i.e., Non-Demented (ND), Very Mild Demented (VMD), Mild Demented (MD), and Moderated Demented (MDTD)). We experiment and compare different dimension-reduction and clustering algorithms to construct the best similarity graph to capture the similarity between the same class images using the cosine distance as a similarity measure. We utilize the similarity graph to present (sample) the training data to a convolutional neural network (CNN). We use the similarity graph as a regularizer in the loss function of a CNN model to minimize the distance between the input images and their k-nearest neighbours in the similarity graph while minimizing the categorical cross-entropy loss between the training image predictions and the actual image class labels. Results: We conduct a baseline experiment with a standard, none-tuned VGG-19 CNN model with different settings and compare the results to identify the best settings to classify (ND vs VMD vs MD vs MDTD). Conclusion: Our method achieves an adequate performance on the testing dataset (accuracy = 0.71, area under receiver operating characteristics curve = 0.86, F1 measure = 0.71). Significance: The classification results show that utilizing a structural similarity graph with none tuned standard CNN model can achieve adequate prediction accuracy and motivate fine-tuning several pre-trained models for better classification performance.

Automated summary

The paper introduces an AD detection method based on learning structural similarity between MRI images and representing the similarity as a graph. The method utilizes CNN to extract features and classify image categories, achieving satisfactory performance on the testing dataset.