The purpose of this study was to distinguish axon-related and myelin-related developmental processes on the neonatal white matter.This study was approved by the local Institutional Review Board. Before the MRI scan, parents of neonates were informed about the goal and risks involved in the MR scan. Written informed consents were obtained from parents of neonates. The inclusion criteria were as follows: age at scan less than 4 weeks, successful MRI data acquisition, and complete clinical information. Subjects who were confirmed or suspected to have cerebral infection, congenital malformation, metabolic disorder, neonatal hypoxic-ischemic encephalopathy, small for gestational age, intracranial hemorrhage, neonatal punctate white matter injury, periventricular leukomalacia, or cortical infection were excluded. During the data processing, artifact-corrupted datasets were also excluded by using the homemade software of an automated method. The neonates were all sedated with oral chloral hydrate before MRI scan. Diffusion kurtosis imaging (DKI) by single short echo planar imaging sequence was performed in a 3T scanner (Signa HDxt, General Electric Medical System, Milwaukee, WI, USA) with an 8-channel RF head coil. DKI was carried out with the following variables: b values = 500, 1000, 2000, 2500 s/mm²; 18 gradient directions; TR = 8000 ms; TE = 109.915 ± 7.825 ms; 20 axial slices with thickness = 4 mm without gap; field of view = 180 × 180 mm²; acquisition matrix size = 128 × 128. Each DKI scan took 11 minutes 33 seconds. Tensors were estimated by using constrained weighted linear least squares after artifacts rejection ^[1,2]. Diffusion tensor imaging (DTI) metrics (fractional anisotropy, FA; axial diffusivity, D_∥; radial diffusivity, D_⊥) and white matter tract integrity (WMTI) metrics (intra-axonal axial diffusivity,D_a,∥; extra-axonal axial diffusivity, D_e,∥; extra-axonal radial diffusivity, D_e,⊥) were derived from DKI. Linear and nonlinear registrations were used to register FA images of all neonates to the neonatal FA template from Johns Hopkins University ^[3]. The other parameters were normalized to the template space by using deformation parameters of FA images. Inter-group differences of metrics were analyzed by using voxel-based analysis (VBA). All tests were taken to be significant at p < 0.05 after family-wise error rate (FWE) correction with threshold-free cluster enhancement (TFCE). To demonstrate the spatial distribution of the possible developmental processes on neonatal white matter, voxel-wise change patterns were obtained according to significant changes of WMTI and DTI metrics separately. Percentages of voxels (voxels%) for different change patterns were calculated in various structures: voxels%=(number of voxels with a change pattern in a region/number of voxels in the region)×100%.

After applying the inclusion and exclusion criteria, datasets of 41 neonates were available, including 19 preterm neonates with postmenstrual age (PMA) from 32.71 to 38.71 weeks (8 males and 11 females) and 22 full-term neonates with PMA from 39.43 to 44.29 weeks (11 males and 11 females).

Increased intra-axonal axial diffusivity and decreased extra-axonal diffusivities were observed on full-term neonates compared with preterm neonates. The axon-related changes in genu corpus callosum were not detected by DTI diffusivities (Figure 1). Furthermore, 5 change patterns were found for WMTI metrics, while 2 patterns for DTI metrics. The spatial distribution of developmental processes was demonstrated by using these change patterns (Figure 2). Main parts of posterior limb of internal capsule and splenium corpus callosum started myelination during the neonatal period. About half of genu corpus callosum (46.67%) was undergoing axon growth. Superior corona radiata (79.79%), inferior fronto-occipital fasciculus (79.92%), and external capsule (81.80%) were mainly in the process of glial cell proliferation.

DTI provided diffusivities by integrating information from different compartments. DTI metrics may be not specific enough to distinguish axon-related and myelin-related processes ^[4]. WMTI metrics of DKI were specific to intra-axonal and extra-axonal spaces ^[5]. DKI provided more metrics than conventional DTI. It is foreseeable to determine more development processes by using these WMTI metrics. Axon growth leads to the increase of the axoplasmic flow ^[6]. The gilial cell proliferation and myelination mainly lead to the structural changes in the extra-axonal space ^[7]. Axon growth is a long lasting process beginning in the premyelination period and continuing into the myelination period. The development processes revealed in this study were in agreement with the postmortem and conventional imaging studies ^[8,9,10]. Myelination starts from the posterior limb of internal capsule in the telencephalon ^[7]. Projection fibers are in a development state with higher maturation degree at birth ^[9]. And association fibers hold a lower maturation degree.WMTI metrics derived from DKI could provide more detailed developmental processes on neonatal white matter.