Corpus Callosum shape analysis: a comparative study of group differences associated with dyslexia, gender and handedness

This paper investigates the extent to which the shape of one of the brain structures, the Corpus Callosum, is correlated to (or moreover, can predict) the gender, handedness or presence of dyslexia. A second goal is to apply a common shape analysis framework to several datasets (totaling 132 CC shapes) previously employed in the literature, as an attempt to mitigate the partly contradicting results that have been reported. As the main result, a statistical shape analysis reveals a significant difference in CC length between the group of normal and the group of dyslexic subjects. Moreover, a classification system based on the length of the posterior midbody region can accurately classify normal/dyslexic subjects about 80% of the time in the two different datasets we have available: a 54 (26 normal and 28 dyslexic) and a 28 (12 normal and 16 dyslexic) CC shape set. As far as we know, this is the first attempt to automatically predict dyslexia based on the characteristics of the Corpus Callosum extracted from MR images.