Medical Image Retrieval for Alzheimer’s Disease Using Structural MRI Measures

The aim of the paper is to study medical image retrieval for Alzheimer’s Disease (AD) on the bases of structural MRI measures. The main goal of the strategy used in this paper is to improve the retrieval performance while using smaller number of features. The feature vector consists of the measurements of cortical and subcortical brain structures, including volumes of the brain structures and cortical thickness. The feature subset selection is additionally applied using the Correlation-based Feature Selection method to exclude irrelevant, redundant or possibly noisy data and to consider the most relevant and discriminative features. Six different scenarios for the image representation are studied: volumetric features, cortical thickness features, all imaging features, selected volumetric features, selected cortical thickness feature and selected imaging features. Euclidean distance is used as a similarity measurement. The dataset used for evaluation of the retrieval performance is provided by the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Experimental results show that the strategy used in this research outperforms the traditional one despite its simplicity and small number of features used for representation. Additionally, the performed analysis demonstrated that the selected features are highly stable through the leave-one-out strategy. Moreover, they are stressed in the literature as significant biomarkers for Alzheimer’s Disease, which makes the strategy used in this research even more reasonable.