The impact of small for gestational age on visual-motor integration in preterm infants: insights from early brain magnetic resonance imaging

Introduction

Prematurity and small for gestational age (SGA) are significant risk factors for neurodevelopmental delays (1). Emerging evidences indicate that preterm infants who are SGA or have very low birth weights are more susceptible to developmental impairments (2,3). However, distinguishing the individual effects of preterm and SGA on cognitive developmental delay remains challenging.

Several factors contribute to the increased risk of SGA and its association with neurodevelopment delays. Iron deficiency, for instance, can lead to inadequate myelination during central nervous system (CNS) development (4). Gestational diabetes may affect neuronal oxidative stress and inflammatory responses (5), while insufficient placental perfusion can impede axon development in the periventricular white matter germinal layer (6). Previous studies have highlighted a connection between SGA and visual functions, including visual perception and visual-motor integration (VMI) (2,7,8). VMI refers to the ability to coordinate visual perception with motor skills. Core components include visual perception, eye-hand coordination (EHC), spatial awareness, and fine motor control. Impairments in VMI have been shown to impact academic performance (9). However, the neural mechanisms underlying VMI developmental differences in the early postnatal period remain unclear.

Over the past few decades, significant advancements in neuroimaging techniques used in early life have greatly enhanced our ability to predict neurological outcomes and investigate neural mechanisms. Beyond the peripheral optic nerve pathways, CNS structures are crucial for VMI. These structures transmit information from lateral geniculate nucleus to the cortex, include the optic radiation (OR), inferior-fronto-occipital fasciculus (IFOF), middle temporal gyrus (MTG), superior temporal gyrus (STG), fusiform gyrus (FG), and corpus callosum (CC) (10,11). The functional integrity of these structures is essential for effective visual processing and is influenced by factors such as myelination maturity, regional volume and fiber density (FD). These characteristics are associated with the birth weight and gestational age (GA) of the fetus, as well as genetic factors during early development (12,13).

Magnetic resonance imaging (MRI) enables precise segmentation of both primary brain regions and white matter fiber bundles, even in infancy (14). Particularly, high angular resolution diffusion imaging can explore and characterize multiple microstructural components within a voxel (14).

To date, few studies have examined the microstructure differences of the visual processing network between SGA and appropriate for gestational age (AGA) infants during early infancy. By utilizing diffusion MRI, we aimed to identify the microstructural alterations associated with VMI deficits in SGA preterm infants, potentially leading to more accurate earlier prognoses and better interventions. We present this article in accordance with the STROBE reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0014/rc).

Methods

Participants

Participants were enrolled from a longitudinal investigation conducted at the High-Risk Infant Follow-Up Clinic in Children’s Hospital, Zhejiang University School of Medicine (China) from 2019 to 2021. The study focused on neonates born before 37 weeks of gestation. The recruitment was aimed at those with viable diffusion MRI and T1 and T2 data within 3 months after birth. VMI evaluations were done at 3, 6 and 18 months of corrected age of expected date of confinement (CA). SGA infants were identified by having a birth weight below the 10th percentile for their GA and sex, whereas the birth weights of AGA infants were between the 10th and 90th percentiles for their GA and sex (by Fenton growth chart-2). Infants with neurological conditions unrelated to prematurity that could independently explain VMI impairment were excluded from the study. These conditions included congenital abnormalities, active seizure disorders, cerebral palsy, hydrocephalus, sensorineural hearing loss, loss of peripheral vision, or retinopathy of prematurity. According to previous literature, the incidence of SGA is estimated to be around 10–20% (15,16). The sample size is adequate, as confirmed by power analysis based on the risk factors of the case and the matched control.

The demographic profile of the participants was enriched by evaluating their socioeconomic status (SES), which was quantified using an adapted version of the Hollingshead Four-Factor Index (17). This index provides a structured method to assess SES based on the occupational prestige and education level of the parents. Detailed medical records pertaining to the perinatal phase were accessible for every participant. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of the Children’s Hospital, Zhejiang University School of Medicine (No. 2019-IRB-027). Written informed consent to participate in the study was obtained from the patients’ parents/legal guardians.

Assessments of VMI ability

Assessments of VMI abilities were conducted for all participants using the Peabody Developmental Motor Scales, Second Edition (PDMS-2), at CAs of 3 and 6 months. These evaluations included skills such as eye tracking, fixation, grasping followed rods and hanging toys, and building blocks. At 18 months CA, further assessment of EHC was performed to reflect the VMI capabilities, utilizing the Griffiths Mental Development Scales. This included tasks like tracking and grasping objects, manipulating small items, construction activities and drawing. However, these tools are not directly interchangeable. To address this, we used age-standardized percentiles from each instrument and focused on the conceptual continuity between early visuomotor precursors and later VMI abilities, rather than direct cross-age score comparisons. Evaluations were conducted by two pediatricians who were blinded to the participants’ medical histories. Raw scores from these subtests were converted into age-equivalent scores or percentages to accurately reflect the children’s VMI abilities.

Brain MRI

Image acquisition

Infants were scanned after receiving 50 mg/kg of enema or oral chloral hydrate at 40–50 weeks of CA. The imaging was conducted using a Philips 3.0T Achieva system with standard eight-channel head coils. Two sequences were used in this study: (I) 3-D sagittal T2-weighted sequences [echo time (TE) =278 ms, repetition time (TR) =2,200 ms, acquisition matrix =224×204, voxel size =0.8×0.8×0.8 mm³, field of view (FOV) =180×161×140 mm³]; and (II) diffusion MRI images collected using echo-planar image (EPI) sequence with 32 non-colinear diffusion encoding directions for b value =800 and 1,500 s/mm² each, in addition to one non-weighted image (TE =115 ms, TR =9,652 ms, voxel size =1.5×1.5×2 mm³, flipangle =90°, FOV =180×180×120 mm³, acquisition matrix =120×118, bandwidth =1,341 Hz/pixel, number of volumes =60, with 60 slices). An extra b0 image with the opposite phase-encoding direction was captured for eddy current correction.

Brain region segmentation and volume calculation

The T2-weighted images were preprocessed, which included brain extraction (18), creation of a brain mask, as well as motion, drift and bias correction (19). Then, the whole brain of each subject was segmented into 126 brain regions using Johns Hopkins University (JHU) neonatal atlas, followed by manual checking, and the volume of each region was extracted. Insufficient automated segmentations were manually corrected.

Diffusion weighted imaging (DWI) preprocessing

All diffusion MRI data underwent susceptibility-induced distortion correction, followed by eddy current correction and head motion correction using “topup” and “eddy” in FSL (20), along with bias correction. Fractional anisotropy (FA) and axial diffusivity/radial diffusivity (AD/RD)/apparent diffusion coefficient (ADC) maps were generated from the diffusion tensor using the weighted linear least squares method (21). The JHU neonatal atlas was nonlinearly registered to each subject’s DWI space using multi-channel information (b0, DWI, and FA) with Advanced Normalization Tools (22,23). Then, the JHU-neonate parcellation map, which included 63 regions of interest (ROIs), was transformed to the individual native space (14). Registration of all subjects was checked.

Quantification of ROIs of brain DWI properties

In the present analyses, we primarily focused on FA due to its significance in evaluating myelination levels within white matter. We then incorporated additional metrics—AD, RD and ADC—for a more comprehensive evaluation. Decreased AD, RD and ADC values, akin to increased FA readings, suggest augmented myelin content, and improved axonal fiber organization. These changes also suggest reduced extracellular water, diminished contribution from crossing fibers, and minimized partial-volume effects. ORs, IFOF, MTG, STG, FG, CC (body), CC (genu) and CC (splenium) were chosen as ROIs in the visual perceptual network.

Fixel-based analysis

First, the group average response functions for single-fiber white matter, gray matter and cerebrospinal fluid were derived. Subsequently, fiber orientation distributions (FODs) (1-mm isotropic resolution) were estimated using the multi-shell multi-tissue constrained spherical deconvolution (MSMT-CSD) method (24). A study-specific FOD template was first constructed using the FOD images from all subjects. Subsequently, each individual’s FOD image was nonlinearly registered to this population template. Fixels were then defined on the template via FOD segmentation, and individual fixels were mapped onto the corresponding template fixels. Finally, FD and fiber cross-section (FC) were computed for each fixel. The JHU neonatal atlas was further mapped to the study-specific template for ROI-based analysis, the sum of FD and mean FC in each voxel were extracted to obtain the ROI averages.

Statistical analysis

For the demographic information, continuous demographic variables such as scan age, GA, birth weight, SES were analyzed using Student’s t-test after normal distribution test to evaluate discrepancies between the SGA and AGA preterm groups. The Chi-squared test was administered for the binary categorical variable sex. VMI scores, EHC percentiles and seven brain MRI characteristics were analyzed using analysis of covariance (ANCOVA) after Levene’s test with sex, GA at birth, scan age and SES as covariates. Bonferroni correction was used for multiple comparisons of VMI abilities and brain MRI characteristics, which divides the nominal significance level (α =0.05) by the number of comparisons, yielding a new significance threshold of α =0.05/n. Multiple linear regression was performed to evaluate the intergroup differences in VMI between the SGA and AGA groups are mediated by MRI characteristics (diffusion metrics/volume/fixel-based microstructural metrics) of eight VMI-related brain regions after controlling for sex, GA at birth, scan age and SES (25). To control the false discovery rate (FDR) for multiple linear regression, we applied the BenjaminiHochberg procedure with an FDR threshold of q =0.05. The level of significance for all the tests was established at P<0.05. These analyses were performed using SPSS software, version 23.0 (IBM Corporation, Armonk, NY, USA).

Results

Demographic and clinical variables

Seventy-nine preterm infants within 3 months of age were enrolled in this study. A total of fourteen infants (two with severe brain lesions on MRI, one with hereditary disease, one with intracranial infection, and 10 with poor imaging quality) who did not meet the inclusion criteria were excluded from this study. In addition, twelve infants who could not complete the VMI ability assessment quit the experiment. A total of 53 infants (including 20 SGA infants and 33 AGA infants) completed the entire follow-up, as shown in the schematic in Figure 1. No data were missed. The demographic and clinical information is presented in Table 1. No significant differences were found between the two groups in terms of sex, GA, scan age, SES and clinical profile during neonatal period (Table 1).

Figure 1 Flow diagram illustrating the case selection and workflow. The study was divided into three parts: (I) VMI comparison at different ages between SGA and AGA preterm groups; (II) brain MRI metrics analysis; and (III) MRI features of VMI-related ROIs filtered between the SGA and AGA preterm groups. Two analytic methods of image processing were used in this study: (I) for ROI-based analysis, segmentation of the diffusion MRI data was achieved by transforming individual subjects to the JHU-neonatal atlas utilizing volume analysis and multi-channel contrasts of b0, DWI, and FA; (II) for FBA, an FOD template was generated using all subjects in this study, and FD and FC were calculated. Statistical analysis was performed to identify the significant group differences. AD, axial diffusivity; ADC, apparent diffusion coefficient; AGA, appropriate for gestational age; DWI, diffusion weighted imaging; FA, fractional anisotropy; FBA, fiber based analysis; FC, fiber cross-section; FD, fiber density; FOD, fiber orientation distribution; JHU, Johns Hopkins University; M, months; MRI, magnetic resonance imaging; RD, radial diffusivity; ROIs, regions of interest; SGA, small for gestational age; VMI, visual-motor integration; WM, white matter.

The development trajectories of VMI abilities are different between SGA and AGA preterm infants

SGA preterm group had significantly lower VMI scores compared with the AGA group at a CA of 6 months (22.45±5.2 vs. 25.45±4.0, P=0.03) and lower EHC percentiles at a CA of 18 months (36.8%±14.1% vs. 51.2%±19.8%, P<0.001). Both were correlated with SGA effect after controlling for sex, GA at birth, scan age and SES [VMI: β =−2.81, R²=0.22, 95% confidence interval (CI): −5.46, −0.17; EHC: β =−14.37, R²=0.20, 95% CI: −24.81, −3.93]. There were no significant differences in VMI scores between the groups at a CA of 3 months (18.50±4.5 vs. 18.79±4.8, P=0.84) (Figure 2). As shown in Figure 2, although the mean scores for gross motor skills assessed by the PDMS-2 were slightly higher in the AGA group at both 6 and 18 months of CA, the differences were not statistically significant. From 3 to 18 months of CA, the percentile of VMI/EHC in the SGA group showed a significant decline (P<0.001, adjusted), while no similar changes were observed in the AGA group. Although there was a slight increase in the percentile of gross motor abilities with age in both groups, the difference was not statistically significant. Notably, all infants received immediate rehabilitation or clinical interventions when developmental delays were identified.

Figure 2 Visual-motor integration and gross motor abilities in SGA and AGA infants. (A) The SGA preterm group exhibited significantly lower VMI scores compared with the AGA group at a CA of 6 months (*, P<0.05) and 18 months (***, P<0.001). No significant difference was observed in VMI scores between the two groups at 3 months of CA. There were no significant differences in gross motor abilities between the two groups at any of these three stages. (B) The development trajectories of VMI/EHC abilities and gross motor abilities between the SGA and AGA groups are illustrated. In the SGA group, the percentages of VMI/EHC abilities were significantly lower at 18 months CA compared to 3 months and 6 months CA (***, P<0.001). AGA, appropriate for gestational age; CA, corrected age of expected date of confinement; EHC, eye and hand coordination; GMDS, Griffiths Mental Development Scales; M, months; PDMS, Peabody Developmental Motor Scales; SGA, small for gestational age; VMI, visual-motor integration.

The volume of FG in early infancy is lower in the SGA preterm group

In line with previous reports of reduced volumes of brain regions in SGA preterm infants, we first calculated the volumes of ROIs within the visual perceptual network. The overall volume of the right and left FGs was significantly lower in the SGA preterm group compared with the AGA group (15,519±2,709 vs. 17,012±1,854 mm³, P=0.03, adjusted), after adjusting for GA at birth, sex, SES scores and post-menstrual age at the time of scanning. No significant differences were observed in other regions of the visual perceptual network between the two groups (Table 2, Figure 3A,3B). Next, we investigated the relationship between reduced FG volume and delayed development of VMI abilities in SGA preterm infants by multiple linear regression. The results showed that the volume of the FG partially mediated the intergroups difference of EHC percentiles at 18 months of CA (β =−13.67; R²=0.21; 95% CI: −24.78, −2.57).

Figure 3 Comparison of the volumes of visual perception-related regions between SGA and AGA preterm infants. (A) Locations and signs of visual perception-related regions are illustrated. The volume of the FG (rectangular box) was smaller in the SGA preterm group than in the AGA preterm group. Colored bars from red to yellow represented the volume of ROIs, and blue bars from dark to light represent the discrepancies (T values) between the SGA and AGA preterm groups, respectively. (B) The results of linear regression revealed that the volumes of ROIs were not linearly associated with EHC percentage at 18 months of CA in the SGA and AGA preterm groups. AGA, appropriate for gestational age; CA, corrected age of expected date of confinement; CC, corpus callosum; CCB, corpus callosum (body); CCG, corpus callosum (genu); CCS, corpus callosum (splenium); EHC, eye-hand coordination; FG, fusiform gyrus; IFOF, inferior-fronto-occipital fasciculus; MTG, middle temporal gyrus; OR, optic radiation; ROIs, regions of interest; SGA, small for gestational age; STG, superior temporal gyrus; VOL, volume.

DWI metrics of the visual perceptual network in early infancy are associated with VMI abilities in the SGA preterm group

Compared with the AGA preterm group, infants in the SGA preterm group exhibited decreased FA values in the FG (0.1125±0.01466 vs. 0.1222±0.01532, P=0.03, adjusted) and CC (splenium) (0.2619±0.05202 vs. 0.2860±0.04034, P=0.04, adjusted) (Table 2, Figure 4A), which were correlated with axonal loss or disrupted myelination in immature brain regions (26). No significant differences were observed for RD, AD and ADC within the visual perceptual network (Table 2, Figure 4A). Multiple linear regression was then used to analyze the relationships between DWI metrics and VMI ability discrepancies between the SGA and AGA preterm groups. Though correlations were detected between EHC percentiles and RD [in the OR, IFOF, FG, CC (body) and CC (genu)], FA (in the OR) in SGA group, as well as FA and ADC [in the CC (splenium)] in the AGA group (Figure 4B), only FA in the FG and CC (splenium) partially mediated the intergroups difference of EHC percentiles at 18 months of CA [FG: β =−14.31, R²=0.20, 95% CI: −25.45, −3.17; CC (splenium): β =−12.36, R²=0.23, 95% CI: −23.23, −1.50].

Figure 4 Discrepancies of the diffusion metrics in the visual perception-related ROIs and the relationship between diffusion metrics and VMI ability at a CA of 18 months in the SGA and AGA preterm groups. (A) Signal differences in AD, RD, FA and ADC between the SGA and AGA preterm groups are shown in visual perception-related regions (red rectangular box). Colored bars from red to yellow represent the enrichment of AD, RD, FA and ADC, and blue bars from dark to light represent the discrepancies (T values) between the SGA and AGA preterm groups. (B) The four rows from top to bottom represent the relationships between EHC percentages and AD, RD, FA, ADC metrics in eight representative ROIs. The RDs of the OR, IFOF, FG, CCB and CCG were negatively correlated with EHC percentage in the SGA preterm group. Additionally, the FA of the OR was positively correlated with EHC percentage in the same group (red bars). In the AGA preterm group, FA was positively correlated with EHC percentage in the CCS whereas ADC of the CCS was negatively correlated with EHC percentage (green bars). There was no linear association between diffusion metrics and EHC percentage in other ROIs in the SGA and AGA preterm groups. AD, axial diffusivity; ADC, apparent diffusion coefficient; AGA, appropriate for gestational age; CA, corrected age of expected date of confinement; CC, corpus callosum; CCB, corpus callosum (body); CCG, corpus callosum (genu); CCS, corpus callosum (splenium); EHC, eye-hand coordination; FA, fractional anisotropy; FG, fusiform gyrus; IFOF, inferior-fronto-occipital fasciculus; MTG, middle temporal gyrus; OR, optic radiation; RD, radial diffusivity; ROIs, regions of interest; SGA, small for gestational age; STG, superior temporal gyrus; VMI, visual-motor integration.

Fixel-based (FD/FC) metrics of the visual perceptual network in early infancy are associated with VMI abilities in the SGA preterm group

Interestingly, FD was significantly lower in the SGA preterm group compared with the AGA preterm group in the OR (0.4101±0.07076 vs. 0.4430±0.06772, P=0.045, adjusted), IFOF (0.3416±0.03766 vs. 0.3741±0.04284, P=0.003, adjusted), FG (0.1103±0.02174 vs. 0.1273±0.02445, P=0.007, adjusted) and CC (splenium) (0.1273±0.02445 vs. 0.4153±0.03101, P=0.01, adjusted) at 40–53 weeks of CA (Table 2, Figure 5A). FD in the OR, IFOF, FG, CC (body) and CC (genu) was found positively related with EHC percentiles at 18 months of CA in SGA group (Figure 5B). Moreover, FCs of the OR (0.9992±0.1390 vs. 1.0800±0.1357, P=0.046, adjusted) and CC (genu) (0.9063±0.1043 vs. 0.9767±0.1096, P=0.03, adjusted) were significantly lower in the SGA preterm group compared with the AGA preterm group (Table 2, Figure 5A). Multiple linear regression was then used to analyze the relationship between FD/FC and VMI ability discrepancies between the SGA and AGA preterm groups. Only FD in the OR and CC (splenium) [OR: β =−14.03, R²=0.20, 95% CI: −25.05, −3.01; CC (splenium): β =−13.70, R²=0.21, 95% CI: −25.01, −2.39] and FC in the OR and CC (genu) [OR: β =−13.68, R²=0.21, 95% CI: −24.68, −2.69; CC (genu): β =−13.89, R²=0.20, 95% CI: −24.94, −2.83] partially mediated the intergroups difference of EHC percentiles at 18 months of CA in the SGA group.

Figure 5 Discrepancies of fiber density and fiber cross-section in the visual perception network between the SGA and AGA preterm groups. (A) FBA based analysis showing with a discrepancy between the FD and FC. The FD values in the OR, IFOF, FG and CCS (red rectangular box) were decreased in the SGA preterm group compared with the AGA preterm group. Moreover, the FCs of the OR and CCG were decreased in the SGA preterm group compared with the AGA preterm group (red rectangular box). Colored bars from red to yellow represent the enrichment of FD and FC, whereas blue bars from dark to light represent the discrepancies (T values) between the SGA and AGA preterm groups. (B) The results of linear regression analysis revealed that the FDs of the OR, IFOF, FG, CCB and CCG were positively correlated with EHC percentages in the SGA preterm group, whereas the FCs and FDs in other regions were not linearly associated with EHC percentage in the SGA or AGA preterm group. AGA, appropriate for gestational age; CC, corpus callosum; CCB, corpus callosum (body); CCG, corpus callosum (genu); CCS, corpus callosum (splenium); EHC, eye-hand coordination; FBA, fiber based analysis; FC, fiber cross-section; FD, fiber density; FG, fusiform gyrus; IFOF, inferior-fronto-occipital fasciculus; MTG, middle temporal gyrus; OR, optic radiation; SGA, small for gestational age; STG, superior temporal gyrus.

Discussion

This study explored the impact of SGA in conjunction with preterm birth on the development of VMI. The findings highlighted several key points: (I) at 3 months of age, an early postnatal period, SGA preterm infants exhibited subtle deficits in VMI abilities compared with AGA preterm infants. These discrepancies became more evident at 6 and 18 months of age. (II) Within 3 months of age, the volume, diffusion metrics and fixel-based (FD/FC) metrics of brain regions involved in visual processing were significantly less mature in the SGA preterm group than in the AGA preterm group. (III) Some of the early MRI features significantly correlated with VMI outcomes at 18 months of CA, which can be utilized to predict the development of VMI in SGA preterm infants. Also, these findings offer a potential neurological foundation for the reduced VMI function observed in SGA preterm infants later in life.

Differences in neurodevelopmental trajectory between SGA and AGA preterm infants

Although both groups were preterm, there were significant differences in neurodevelopmental outcomes between SGA and AGA infants at different periods of infancy. A retrospective cohort study on preterm SGA infants revealed that at 40 weeks post-conceptional age, late preterm SGA infants born at 33–36-week GA exhibited a mild delay in neural conduction in the auditory brainstem compared with AGA preterm infants. By 56 weeks, these SGA infants showed moderately faster neural conduction in the caudal brainstem regions (3). Lee et al. reported that in a community-based cohort in rural Bangladesh, SGA infants were at higher risk for fine motor and language delays at 24 months of age (27). Additionally, Kono et al. found that preterm SGA infants were at a higher risk for cerebral palsy and developmental delays compared with AGA preterm infants at 3 years of age (28).

As early as in 2013, differences in visual acuity revealed that SGA preterm infants lagged behind AGA preterm infants in visual processing functions at 1 year of age (2). In our study, we found that VMI development in SGA preterm infants progressed more slowly at an earlier age of (6 months of CA) and continued to lag until 18 months of CA. Østgård et al. also found that VMI deficits in individuals born with SGA persist into late adolescence (19–20 years of age) (29).

The regions of the CNS related to visual processing are at high risk during the first and second trimesters (30-32). SGA combined with preterm delivery often occurs during this period, which is primarily caused by genetics, epigenetics, exposure to harmful environmental factors such as infections, and inadequate nutritional factors (33). These insults can also lead to early developmental delay of the visual perceptual network (31). Without timely intervention, this dysfunction can persist throughout childhood and even into adulthood (34).

Differences in the microstructure of the visual processing network between SGA and AGA preterm infants

The neural mechanisms underlying VMI developmental differences in the early postnatal period remain unclear. Previous studies have reported smaller thalami, smaller cortical surface areas in the STG, insula, and medial occipital lobe, as well as thinner ORs and related cortical regions in children born prematurely (9,35,36). However, few studies have examined the microstructural differences in the visual processing network between SGA and AGA infants during early infancy. Eduard et al. identified distinct microstructural patterns in the fetal brains of term AGA and SGA fetuses, particularly in regions such as the mesencephalon, which houses nuclei related to visual perception (37). Other studies focusing on differences in visual perception networks between SGA and AGA infants have verified these discrepancies at 12 months of CA or even into adulthood (11,38).

The current study not only supports prior findings, but also further showing differential effects of SGA on the volumes of visual processing loops at early infancy, especially in the FG, a key structure for high-level visual functions, such as face perception, object recognition, and reading (39).

In addition to the volume differences in visual processing pathways between SGA and AGA preterm infants, visual processing abilities are also closely associated with the maturation and integrity of CNS conduction pathways. Using neuroimaging segmentation techniques, previous studies have shown that regions such as the ORs, IFOF, MTG, STG, FG and CC (segmented into the genu, splenium and body) are involved in visual processing conduction (11,40-42). In the present study, diffusion MRI-based imagine was employed to assess the myelin development of the visual perception network during the early infancy period. Group differences in FA values in the FG and CC (splenium) likely reflected a combination of microstructural alterations, including myelin, axonal density and fiber organization, in agreement with previous studies on motor development patterns in SGA preterm infants (43). The reduced FD in the OR, IFOF, FG and CC (splenium), along with decreased FC in the OR and CC (genu), indicate compromised microstructural integrity-as reflected by fixel-based metrics-in SGA preterm infants compared with AGA preterm infants during the neonatal period. The FG is a part of the more medial visual network, known as the largest component of the human ventral temporal cortex and a higher-order visual region involved in complex integration, may have higher metabolic demands and greater synaptic density (39,44). This could make its neurons and glia more vulnerable to the chronic hypoxia and oxidative stress associated with SGA. The OR and IFOF are white matter tracts that are important for connecting visual association areas with primary motor frontal eye fields and facilitate the transmission of basic visual information. The IFOF also links the posterior temporal, orbito-frontal, and occipital regions (45). The CC, the largest connective structure in the white matter; comprises of over 190 million axons that transfer information between the two cerebral hemispheres and contains both homotopic and heterotopic interhemispheric connections (46).

The distinct patterns observed in volumes, diffusion-based (FA/RD/AD) and fixel-based (FD/FC) metrics, localized in the FG, OR, IFOF and CC (genu and splenium), highlight unique developmental paradigms within the visual perceptual network between SGA and AGA preterm infants.

Relationships between early MRI metrics and VMI development in SGA preterm infants

In this study, we evaluate features from diffusion MRI images to identify patterns associated with delays in VMI development. We found that FA, FD and FC values, along with the volumes of the aforementioned ROIs in the visual perception pathway, are novel important metrics linked to VMI abilities in SGA preterm infants. Linear regression analysis revealed that at 18 months CA, the VMI abilities discrepancy between SGA and AGA preterm infants were partially mediated by volume, diffusion and fixel based metrics in the OR, FG, CC (splenium) and CC (genu). These results suggest that VMI developmental delay in SGA infants may stem from structural alterations in specific gray and white matter regions of the visual networks.

Moreover, the associations between volume, FA, FD, FC, and VMI ability at 18 months CA indicated that the delays in the development of MRI characteristics in areas related to visual processing may continue into later infancy, which may serve as biomarkers for forecasting VMI development based on imaging data (47).

Limitations

Due to the limitations of current image segmentation techniques for infant brains, this study was unable to isolate the lateral geniculate nucleus from the thalamus, an important component in the visual perception network. Also, chloral hydrate was used in this study for sedation according to the routine scan protocol at our local hospital, though it is considered to be safe (48). Additionally, to avoid excessive examination, the number of term-born SGA and AGA infants was insufficient to be included in the analysis. The sample size in this study was small and the scan age was relatively wide due to low MRI compliance. In future studies, we aim to establish a model using a larger scale of SGA cohort, controlling for scan age in a more concentrated distribution and longitudinal studies are needed to validate these findings and explore potential interventions.

Conclusions

This study confirms that the SGA status in preterm infants significantly influences developmental delays in visual perception, especially in tasks requiring VMI. MRI metrics of the OR, IFOF, FG and CC served as sensitive biomarkers for early identification. These findings support the integration of MRI into routine follow-up to facilitate timely intervention and improve VMI development of preterm infants with SGA.

Acknowledgments

We thank the team of Professor Dan Wu, from Zhejiang University, for their great work and support to the neuroimage analysis.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0014/rc

Data Sharing Statement: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0014/dss

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0014/prf

Funding: This study was supported by the National Natural Science Foundation of China (No. 82271738, to Y.L.), Zhejiang Provincial Basic Public Welfare Research Projects (No. LTGY24H260003, to C.J.), and Zhejiang Provincial Natural Science Foundation of China (No. LY24H180002, to H.Z.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0014/coif). Y.L. reports funding from the National Natural Science Foundation of China (No. 82271738); H.Z. reports funding from the Zhejiang Provincial Natural Science Foundation of China (No. LY24H180002); and C.J. reports funding from the Projects of Zhejiang Provincial Basic Public Welfare Research (No. LTGY24H260003). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of the Children’s Hospital, Zhejiang University School of Medicine (No. 2019-IRB-027). Written informed consent to participate in the study was obtained from the patients’ parents/legal guardians.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Kim DY, Park HK, Kim NS, et al. Neonatal diffusion tensor brain imaging predicts later motor outcome in preterm neonates with white matter abnormalities. Ital J Pediatr 2016;42:104. [Crossref] [PubMed]
2. Pinello L, Manea S, Visonà Dalla Pozza L, et al. Visual, motor, and psychomotor development in small-for-gestational-age preterm infants. J AAPOS 2013;17:352-6. [Crossref] [PubMed]
3. Jiang ZD, Wang C, Jiang JK. Postnatal functional integrity of the brainstem auditory pathway in late preterm infants born of small-for-gestation age: how different from those born of appropriate-for-gestation. Eur J Pediatr 2024;183:3041-51. [Crossref] [PubMed]
4. Oh C, Keats EC, Bhutta ZA. Vitamin and Mineral Supplementation During Pregnancy on Maternal, Birth, Child Health and Development Outcomes in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis. Nutrients 2020;12:491. [Crossref] [PubMed]
5. Chen Z, Zhao Y, Yang Y, et al. Leptin withdrawal after birth: a neglected factor account for cognitive deficit in offspring of GDM mother. Med Hypotheses 2011;77:125-7. [Crossref] [PubMed]
6. Huang W, Bhaduri A, Velmeshev D, et al. Origins and Proliferative States of Human Oligodendrocyte Precursor Cells. Cell 2020;182:594-608.e11. [Crossref] [PubMed]
7. Lindqvist S, Vik T, Indredavik MS, et al. Visual acuity, contrast sensitivity, peripheral vision and refraction in low birthweight teenagers. Acta Ophthalmol Scand 2007;85:157-64. [Crossref] [PubMed]
8. Birch EE, Carlson SE, Hoffman DR, et al. The DIAMOND (DHA Intake And Measurement Of Neural Development) Study: a double-masked, randomized controlled clinical trial of the maturation of infant visual acuity as a function of the dietary level of docosahexaenoic acid. Am J Clin Nutr 2010;91:848-59. [Crossref] [PubMed]
9. Sripada K, Løhaugen GC, Eikenes L, et al. Visual-motor deficits relate to altered gray and white matter in young adults born preterm with very low birth weight. Neuroimage 2015;109:493-504. [Crossref] [PubMed]
10. Petri S, Tinelli F. Visual impairment and periventricular leukomalacia in children: A systematic review. Res Dev Disabil 2023;135:104439. [Crossref] [PubMed]
11. Maldonado IL, Descoteaux M, Rheault F, et al. Multimodal study of multilevel pulvino-temporal connections: a new piece in the puzzle of lexical retrieval networks. Brain 2024;147:2245-57. [Crossref] [PubMed]
12. Sanz-Cortes M, Ratta GA, Figueras F, et al. Automatic quantitative MRI texture analysis in small-for-gestational-age fetuses discriminates abnormal neonatal neurobehavior. PLoS One 2013;8:e69595. [Crossref] [PubMed]
13. Han X, Maharjan S, Chen J, et al. High-resolution diffusion magnetic resonance imaging and spatial-transcriptomic in developing mouse brain. Neuroimage 2024;297:120734. [Crossref] [PubMed]
14. Liu T, Gao F, Zheng W, et al. Diffusion MRI of the infant brain reveals unique asymmetry patterns during the first-half-year of development. Neuroimage 2021;242:118465. [Crossref] [PubMed]
15. Deng Y, Yang F, Mu D. First-year growth of 834 preterm infants in a Chinese population: a single-center study. BMC Pediatr 2019;19:403. [Crossref] [PubMed]
16. Monachesi C, Correani A, Antognoli L, et al. Incidence and two-year neurodevelopmental outcomes of small-for-gestational-age preterm infants: how do they relate to using different neonatal anthropometric charts? Minerva Pediatr (Torino) 2023. [Epub ahead of print]. doi: 10.23736/S2724-5276.23.07446-3.
17. Vohr BR, Garcia Coll CT. Neurodevelopmental and school performance of very low-birth-weight infants: a seven-year longitudinal study. Pediatrics 1985;76:345-50.
18. Smith SM. Fast robust automated brain extraction. Hum Brain Mapp 2002;17:143-55. [Crossref] [PubMed]
19. Tustison NJ, Avants BB, Cook PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging 2010;29:1310-20. [Crossref] [PubMed]
20. Andersson JLR, Sotiropoulos SN. An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage 2016;125:1063-78. [Crossref] [PubMed]
21. Basser PJ, Mattiello J, LeBihan D. MR diffusion tensor spectroscopy and imaging. Biophys J 1994;66:259-67. [Crossref] [PubMed]
22. Oishi K, Mori S, Donohue PK, et al. Multi-contrast human neonatal brain atlas: application to normal neonate development analysis. Neuroimage 2011;56:8-20. [Crossref] [PubMed]
23. Djamanakova A, Faria AV, Hsu J, et al. Diffeomorphic brain mapping based on T1-weighted images: improvement of registration accuracy by multichannel mapping. J Magn Reson Imaging 2013;37:76-84. [Crossref] [PubMed]
24. Jeurissen B, Tournier JD, Dhollander T, et al. Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data. Neuroimage 2014;103:411-26. [Crossref] [PubMed]
25. Liu X, Xia X, Hu F, et al. The mediation role of sleep quality in the relationship between cognitive decline and depression. BMC Geriatr 2022;22:178. [Crossref] [PubMed]
26. Alexander AL, Lee JE, Lazar M, et al. Diffusion tensor imaging of the brain. Neurotherapeutics 2007;4:316-29. [Crossref] [PubMed]
27. Lee AC, Cherkerzian S, Tofail F, et al. Perinatal inflammation, fetal growth restriction, and long-term neurodevelopmental impairment in Bangladesh. Pediatr Res 2024;96:1777-87. [Crossref] [PubMed]
28. Kono Y, Yonemoto N, Nakanishi H, et al. A Retrospective Cohort Study on Mortality and Neurodevelopmental Outcomes of Preterm Very Low Birth Weight Infants Born to Mothers with Hypertensive Disorders of Pregnancy. Am J Perinatol 2022;39:1465-77. [Crossref] [PubMed]
29. Østgård HF, Skranes J, Martinussen M, et al. Neuropsychological deficits in young adults born small-for-gestational age (SGA) at term. J Int Neuropsychol Soc 2014;20:313-23. [Crossref] [PubMed]
30. McMahon E, Wintermark P, Lahav A. Auditory brain development in premature infants: the importance of early experience. Ann N Y Acad Sci 2012;1252:17-24. [Crossref] [PubMed]
31. Birch EE, O'Connor AR. Preterm birth and visual development. Semin Neonatol 2001;6:487-97. [Crossref] [PubMed]
32. Neuhauss SC. Behavioral genetic approaches to visual system development and function in zebrafish. J Neurobiol 2003;54:148-60. [Crossref] [PubMed]
33. Finken MJJ, van der Steen M, Smeets CCJ, et al. Children Born Small for Gestational Age: Differential Diagnosis, Molecular Genetic Evaluation, and Implications. Endocr Rev 2018;39:851-94. [Crossref] [PubMed]
34. Burstein O, Zevin Z, Geva R. Preterm Birth and the Development of Visual Attention During the First 2 Years of Life: A Systematic Review and Meta-analysis. JAMA Netw Open 2021;4:e213687. [Crossref] [PubMed]
35. Jacobson LK, Dutton GN. Periventricular leukomalacia: an important cause of visual and ocular motility dysfunction in children. Surv Ophthalmol 2000;45:1-13. [Crossref] [PubMed]
36. Khurana R, Shyamsundar K, Taank P, et al. Periventricular leukomalacia: an ophthalmic perspective. Med J Armed Forces India 2021;77:147-53. [Crossref] [PubMed]
37. Sanz-Cortés M, Figueras F, Bonet-Carne E, et al. Fetal brain MRI texture analysis identifies different microstructural patterns in adequate and small for gestational age fetuses at term. Fetal Diagn Ther 2013;33:122-9. [Crossref] [PubMed]
38. Padilla N, Falcón C, Sanz-Cortés M, et al. Differential effects of intrauterine growth restriction on brain structure and development in preterm infants: a magnetic resonance imaging study. Brain Res 2011;1382:98-108. [Crossref] [PubMed]
39. Weiner KS, Zilles K. The anatomical and functional specialization of the fusiform gyrus. Neuropsychologia 2016;83:48-62. [Crossref] [PubMed]
40. Ingvaldsen SH, Jørgensen AP, Grøtting A, et al. Visual outcomes and their association with grey and white matter microstructure in adults born preterm with very low birth weight. Sci Rep 2024;14:2624. [Crossref] [PubMed]
41. Urbanski M, Thiebaut de Schotten M, Rodrigo S, et al. Brain networks of spatial awareness: evidence from diffusion tensor imaging tractography. J Neurol Neurosurg Psychiatry 2008;79:598-601. [Crossref] [PubMed]
42. Thomas C, Humphreys K, Jung KJ, et al. The anatomy of the callosal and visual-association pathways in high-functioning autism: a DTI tractography study. Cortex 2011;47:863-73. [Crossref] [PubMed]
43. Padilla N, Junqué C, Figueras F, et al. Differential vulnerability of gray matter and white matter to intrauterine growth restriction in preterm infants at 12 months corrected age. Brain Res 2014;1545:1-11. [Crossref] [PubMed]
44. Uddin LQ, Yeo BTT, Spreng RN. Towards a Universal Taxonomy of Macro-scale Functional Human Brain Networks. Brain Topogr 2019;32:926-42. [Crossref] [PubMed]
45. Ashtari M. Anatomy and functional role of the inferior longitudinal fasciculus: a search that has just begun. Dev Med Child Neurol 2012;54:6-7. [Crossref] [PubMed]
46. Paul LK, Brown WS, Adolphs R, et al. Agenesis of the corpus callosum: genetic, developmental and functional aspects of connectivity. Nat Rev Neurosci 2007;8:287-99. [Crossref] [PubMed]
47. Mostapha M, Styner M. Role of deep learning in infant brain MRI analysis. Magn Reson Imaging 2019;64:171-89. [Crossref] [PubMed]
48. Finnemore A, Toulmin H, Merchant N, et al. Chloral hydrate sedation for magnetic resonance imaging in newborn infants. Paediatr Anaesth 2014;24:190-5. [Crossref] [PubMed]