MRI study of the cortical folding process in the premature newborn brain

Introduction In the adult brain, the cortical sulci morphology is highly variable among individuals [1], whereas the primary folds (called “sulcal roots”) are assumed to develop in a stable way during intra-uterine life [2]. So far, several hypotheses have been conducted on the factors that underlie this folding process, like the mechanical tension from white matter fibers [3], but quantitative measurements of the sulci formation are still lacking in support of such theories. As MRI is now used clinically for the noninvasive evaluation of the preterm newborn [4], we investigated the emergence of cortical folding through the application of post-processing tools dedicated to the immature brain. Materials and Methods MR acquisition 35 premature newborns (mean gestational age GA: 30.9±2.1weeks, range: 26.7w-35.9w) were studied by MRI on a 1.5T scanner (Philips Medical Systems). Coronal slices covering the whole brain were imaged by a fast spin echo sequence yielding a T2 contrast between cerebral tissues, with a spatial resolution of 0.7x0.7x1.5mm or 0.8x0.8x1.2mm (Figure 1a). Data post-processing Cortical surface reconstruction The cortical surface was reconstructed through a sequence of image post-processing tools adapted from a sequence developed for the adult brain [5-6], as described previously for foetal images [7]. It included a bias correction and semi-automatic segmentations of the brain and of the interface between cortex and white matter, which used mathematical morphology and a manual optimization of the threshold parameters for each newborn. A smooth triangle based mesh of this detected surface was computed, and its curvature was estimated locally. Cortical sulci identification The sulci were highlighted through a curvature threshold and a detection of connected components, and labelled manually according to post-mortem atlases [8-9] and prenatal images [10]. Their area (A) was calculated as a marker of their maturation. Evaluation of the preterm structural maturation Sulcation index For each newborn, we computed a sulcation index, defined as the ratio between the areas of external sulci (except the sylvian fissure) and the closed surface of the whole brain, and we assessed the age-related changes of this index over the newborn group. Interindividual variability Then we compared the folding process in newborns with similar sulcation index, by realigning their images with an iterative normalization procedure based on the creation of template images with SPM2 software [11] (Figure 1b). Inter-hemispherical asymmetries Finally, we evaluated the differences in maturation between the right (r) and left (l) sulci by testing the nullity of asymmetry indices (A(r)-A(l))/(A(r)+A(l)) over the group using one-paired Student t tests. Results Cortical surface reconstruction From high quality images, cortical surfaces could be reconstructed semi-automatically for all newborns (Figure 2). In the older ones (>31w old), some manual corrections were required around the central sulcus and at the level of the sylvian fissure, because the increasing gyration and myelination implied partial volume effects and decreasing contrast in the image. Imaging more isotropic voxels (0.8x0.8x1.2mm) enabled to prevent in part these corrections, while having no influence on the sulcus quantification. We optimized the reconstruction quality for the brain external surface, where most sulci are localized (Figure 3a), as the central grey nuclei and the ventricles complicated the segmentation of the internal surface, where the sulci appeared roughly connected (Figure 3b). Cortical gyrification We observed an increasing sulcation index with gestational age (correlation coefficient: 0.9, p<0.001), but the onset of the folding process differed between individuals. Several sulci were identified in all newborns (Figure 3a-b): the sylvian fissure (SF), the callosal and cingulum sulci (CaS, CiS), the calcarine and parieto-occipital fissures (CF, POF) in the internal surface, and the central sulcus (CS) in the external surface. In the temporal lobe, we observed the appearance of the superior and inferior temporal sulci (STS/ITS) from 26.7/27.3w GA to 30.5/31w GA respectively, and of the collateral sulcus (CoS) from 27.9w GA to 30.6w GA. In the frontal lobe, the pre-central sulcus (preCS) seemed to form in two parts between the superior and inferior frontal sulci (SFS/IFS), between 27.3w GA and 32.4w GA; the uncinate sulcus (UnS) was present from 30.3-31w GA on. Finally, the post-central sulcus (postCS) development was concomitant to the parietal and parieto-occipital sulci (POS) folding, from 27.9/27.3w GA to 31/32.4w GA respectively. Inter-individual variability Thus some inter-individual variability was observed in the folding process of newborns of equivalent age. Besides, the sulci morphology and localization in the external surface also differed between newborns with similar sulcation index (Figure 3c-e), particularly in the frontal and parietal lobes at the level of the preand post-central sulci, as shown by the sulci superposition resulting from images realignement. Inter-hemispherical asymmetries Finally, the area of the superior temporal sulcus was significantly larger in the right hemisphere (t=-3.49, p<0.001).