A Graph Matching-Based Sulcal Pattern Analysis: Application to the Study of Twin Brains

Kiho Im¹, Rudolph Pienaar¹, Jong-Min Lee², Joon-Kyung Seong³, Yu Yong Choi², Kun Ho Lee⁴, P. Ellen Grant¹

We present a novel quantitative method for automatically comparing and analyzing sulcal patterns using a sulcal pit-based graph. We computed similarity between graphs by determining the optimal match using a spectral method which exploits geometric features of nodes as well as their relationships. Our method was applied to twin data and showed that the similarity of the sulcal pattern in twin pairs was significantly higher than in unrelated pairs. This method can be applied to various genetic developmental disorders, to provide a quantitative and reliable comparison of gyral folding based on sulcal anatomy.