Automatic Internal Segmentation of Caudate Nucleus for Diagnosis of Attention-Deficit/Hyperactivity Disorder

Studies on volumetric brain Magnetic Resonance Imaging (MRI) showed neuroanatomical abnormalities in pediatric Attention Deficit Hyperactivity Disorder (ADHD). In particular, the diminished right caudate volume is one of the most replicated findings among ADHD samples in morphometric MRI studies. In this paper, we propose a fullyautomatic method for internal caudate nucleus segmentation based on machine learning. Moreover, the ratio between right caudate body volume and the bilateral caudate body volume is applied in a ADHD diagnostic test. We separately validate the automatic internal segmentation of caudate in head and body structures and the diagnostic test using real data from ADHD and control subjects. As a result, we show accurate internal caudate segmentation and similar performance among the proposed automatic diagnostic test and the manual annotation.