Figure 5

VTC sulci in chimpanzees can be predicted using sulcal definitions from human VTC. (A) Mean performance (as measured by the dice coefficient) to predict the CoS, OTS, and MFS within species. The three sulci were defined on the FreeSurfer average surface (human template), as well as an average chimpanzee surface that was generated from an independent set of 30 chimpanzee brains (chimp30 template). Using cortex-based alignment, each sulcus was projected to each individual human and chimpanzee surface, which we refer to as the predicted sulcus. The alignment between the predicted and actual sulci (which were manually defined in each individual) was then quantified using the dice coefficient (“Materials and methods”). Green bars illustrate performance when using sulci defined on a chimpanzee template to predict sulci in individual chimpanzee cortical surfaces. Blue bars illustrate performance when using sulci defined on a human template to predict sulci in individual human cortical surfaces. In both species, prediction performance is highest for the CoS. Between species, there is better prediction performance in humans compared to chimpanzees. (B) Same as in (A) except sulci are predicted between species. Specifically, green bars illustrate performance when using sulci defined on a human template to predict sulci in individual chimpanzee cortical surfaces, while blue bars illustrate performance when using sulci defined on a chimpanzee template to predict sulci in individual human cortical surfaces. Prediction performance is highest for the CoS. Additionally, prediction performance in humans is higher when using a human compared to chimpanzee template for all sulci, while the species template used to predict VTC sulci does not affect performance when predicting the CoS across hemispheres and the right MFS in individual chimpanzee surfaces. (C) To give the reader a sense as to what higher (~ 0.8) vs. lower (~ 0.5) dice coefficients look like in terms of the correspondence between the predicted MFS on the cortical surface, three examples are included. From left to right: (1) predicting the MFS (yellow outline) in an individual chimpanzee using a chimpanzee template, (2) predicting the MFS in an individual human using a human template, and (3) predicting the MFS in an individual chimpanzee using a human template. The same individual chimpanzee cortical surface is included in the first and third column. In the former case, the chimpanzee template well-predicted the MFS in the individual chimp (leftmost column), while in the latter case, the MFS prediction from the human template is shifted medially, resulting in a lower dice coefficient (right column).