PREDICTION FOR HUMAN INTELLIGENCE USING MORPHOMETRIC CHARACTERISTICS OF CORTICAL SURFACE: PARTIAL LEAST SQUARE ANALYSIS J.-J. YANG, a U. YOON, b H. J. YUN, a K. IM, c,d Y. Y. CHOI, e K. H. LEE, f H. PARK, g M. G. HOUGH h AND J.-M. LEE a * a Department of Biomedical Engineering, Hanyang University, Seoul, South Korea b Department of Biomedical Engineering, Catholic University of Daegu, Gyeongsan-si, South Korea c Division of Newborn Medicine, Children’s Hospital Boston, Harvard Medical School, Boston, MA, USA d Center for Fetal Neonatal Neuroimaging and Developmental Science, Children’s Hospital Boston, Harvard Medical School, Boston, MA, USA e Bioimaging Research Center, Gwangju Institute of Science and Technology, Gwangju, South Korea f Department of Marine Life Science, College of Natural Sciences, Chosun University, Gwangju, South Korea g School of Electronic and Electrical Engineering, Sungkyunkwan University, South Korea h Functional Magnetic Resonance Imaging of the Brain Centre, University of Oxford, Oxford OX3 9DU, United Kingdom Abstract—A number of imaging studies have reported neuroanatomical correlates of human intelligence with vari- ous morphological characteristics of the cerebral cortex. However, it is not yet clear whether these morphological properties of the cerebral cortex account for human intelli- gence. We assumed that the complex structure of the cere- bral cortex could be explained effectively considering cortical thickness, surface area, sulcal depth and absolute mean curvature together. In 78 young healthy adults (age range: 17–27, male/female: 39/39), we used the full-scale intelligence quotient (FSIQ) and the cortical measurements calculated in native space from each subject to determine how much combining various cortical measures explained human intelligence. Since each cortical measure is thought to be not independent but highly inter-related, we applied partial least square (PLS) regression, which is one of the most promising multivariate analysis approaches, to over- come multicollinearity among cortical measures. Our results showed that 30% of FSIQ was explained by the first latent variable extracted from PLS regression analysis. Although it is difficult to relate the first derived latent variable with specific anatomy, we found that cortical thickness measures had a substantial impact on the PLS model supporting the most significant factor accounting for FSIQ. Our results pre- sented here strongly suggest that the new predictor combin- ing different morphometric properties of complex cortical structure is well suited for predicting human intelligence. Ó 2013 IBRO. Published by Elsevier Ltd. All rights reserved. Key words: cortical thickness, sulcal depth, curvature, sur- face area, partial least square regression, human intelligence. INTRODUCTION With the advent of modern brain imaging technology, researchers have vigorously examined how brain structure and function are related to intelligence with various neuroimaging techniques and intelligence measures (Jung and Haier, 2007; Haier, 2009; Luders et al., 2009). Especially, neuroanatomical correlates of intelligence have made considerable progress based on structural magnetic resonance imaging (MRI). The cerebral cortex holds two thirds of the brain’s neurons and thus appears to be a promising candidate for determining the primary neuroanatomical correlates of intelligence (Luders et al., 2006a). Measure of cortical thickness has been used for studies of neuroanatomical correlates of its local variations with human intelligence in prefrontal and temporal cortical regions (Shaw et al., 2006; Narr et al., 2007; Karama et al., 2009). Narr et al. (2007) showed prominent correlation between FSIQ and cortical thickness in a number of regions with adults. Shaw et al. (2006) revealed that individuals with a superior IQ had a generally thicker cortex primarily in frontal areas during their late childhood to early adulthood (between 8.6 and 29 years of age) than subjects with a lower IQ. Karama et al. (2009) also reported that significant positive associations were evident between the cognitive ability factor and cortical thickness in most multimodal association areas. Measures of cortical shape such as cortical convolution, surface area, sulcal depth and absolute mean curvature also reported the neuroanatomical correlates of human intelligence (Im et al., 2006; Luders et al., 2008). Luders et al. (2008) showed a relationship between cortical convolution, which is defined as a point-specific 0306-4522/13 $36.00 Ó 2013 IBRO. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuroscience.2013.04.051 * Corresponding author. Address: Department of Biomedical Engi- neering, Hanyang University, Sanhakgisulkwan 319, Wangsimni-ro, Seongdong-gu, Seoul 133-791, South Korea. Tel: +82-2-2220-0685; fax: +82-2-2296-5943. E-mail address: ljm@hanyang.ac.kr (J.-M. Lee). Abbreviations: AAL, automated anatomical labeling; BAs, Brodmann areas; FSIQ, full-scale intelligence quotient; ICA, independent component analysis; IQ, intelligence quotient; PCA, principal component analysis; P-FIT, Parieto-Frontal Integration Theory; PLS, partial least square; PRESS, Predicted Residual Estimated Sum of Squares; RESS, Residual Estimated Sum of Squares; RMSE, root mean squared error of the residual sum; RMSEp, root mean squared error of the predicted residual sum; WAIS, Wechsler Adult Intelligence Scale Revised. Neuroscience 246 (2013) 351–361 351