Prediction for human intelligence using morphometric characteristics of cortical surface: Partial least square analysis

Improved explanation of human intelligence using cortical features with second order moments and regression

“…Two sets of PLSR analysis were performed. The mean values of cortical measurements of the chosen ROIs with FSIQ in one set were analyzed, which is consistent with previous research [13]. We also analyzed two moment values computed from cortical measurements of chosen ROIs with FISQ in the other set.…”

“…45% of the explained variance was obtained by using two moments for the combined cortical feature. These results are on par with or better than those of related research, which reported explained variance values between 0.42 and 0.50 [9,13]. Choi et al reported 50% explained variance, but they did not factor in predictability considerations such as PRESS [9].…”

“…Using them all in the regression analysis would lead to an excessive number of independent variables. The same approach from our previous work was used [13]. The mean values were measured for 78 ROIs, and then simple Pearson's correlations with FSIQ values were computed to find out what ROIs are the most informative for each cortical feature.…”

“…Other cortical features, including sulcal depth, surface area, and mean curvature, were also used to find neuroanatomical correlates of human intelligence [11,12]. In our previous Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/cbm study, the cortical thickness, sulcal depth, mean curvature, and surface area were used to explain FSIQ for 78 subjects [13].…”

“…Second order moments are used to extract non-correlated measurements from ROIs in addition to the simple mean values. This method is used to analyze existing data to better explain the variance of FSIQ using various cortical features [13]. The combination of (1) using four cortical features to account for human intelligence, and (2) using additional second order moments besides simple mean values of cortical measurements to represent an ROI could not be found in existing literature to the best of our knowledge.…”