Classification of ADHD by Using Textural Analysis of MR Images

Abstract

Attention deficit hyperactivity disorder (ADHD) is a mental disorder that affects millions of children. It is difficult to diagnose because symptoms of ADHD differ from person to person. The aim of this study is to develop an objective tool that can help physicians to diagnose this disease by minimizing the human effort. Brain MR images from 26 individuals from NP Istanbul Neuropsychiatry Hospital were collected. Two separated dataset is created in order to extract features from magnetic resonance images. These are gray level co-occurrence matrix dataset and Haralick texture features dataset. The most useful features are determined by Principal Component Analysis. K-nearest neighbor, Naive Bayes and Decision Tree algorithms are used for classification. Performance of the model is tested by cross validation and hold out methods, and evaluated by using sensivity and specifity values.