ADHD-200 Classification Based on Social Network Method

Attention Deficit Hyperactivity Disorder (ADHD) is one of the most common diseases in school aged children. In this study, we proposed a method based on social network to extract the features of the ADHD-200 resting state fMRI data between ADHD conditioned and control subjects. And the classification is done by using the support vector machine. The innovation of this paper lies in that: firstly, in the social network, the edge is defined by correlation between two voxels, where the threshold is proposed based on the optimal properties of small world; secondly, in the procedure of feature extraction, besides the traditional network features, we also exploit the new features such as assortative mixing and synchronization. We obtain an average accuracy of 63.75%, which is better than the average best imaging-based diagnostic performance 61.54% achieved in the ADHD-200 global competition. Compared with the proposed method, the result of the method based on traditional features is 61.04% , which verified that the proposed method based on new features is better than traditional one.