Evaluation of fetal brain development in growth restriction subtypes using brain MRI volume measurement

Abstract

Background

This retrospective study aims to explore the value of brain MRI volume measurements in evaluating fetal brain development in fetal growth restriction (FGR) fetuses, by comparing 2 subtypes of FGR with appropriate gestational age (AGA) fetuses.

Methods

A total of 169 fetal brain MRI with suspected abnormal development, identified through ultrasound screening at this hospital between 2021 and 2025 were analyzed,11 cases were excluded. 98 cases were FGR fetuses (43 early-onset subtypes and 55 late-onset subtypes), and 60 were AGA fetuses, gestational age ranging from 26 to 37 weeks. Three-dimensional reconstruction and image segmentation were performed on fetal intracranial tissues including cerebrum, cerebellum and brainstem. Changes in brain volume at different gestational weeks were analyzed to assess the development of fetal brain anatomical structures.

Results

In both groups, the Pearson correlation coefficients for cerebrum, brainstem, and cerebellum volume with head circumference and gestational age were greater than 0.8, indicating a strong correlation. Across all gestational age subgroups (26–31 weeks), cerebral, brainstem, and cerebellar volumes were all significantly smaller in early-onset FGR fetuses than in AGA fetuses (independent-samples t-test; e.g., at 26–27 weeks: cerebrum 87.3 ± 9.2 cm³ vs. 113.2 ± 10.3 cm³, brainstem 1.7 ± 0.2 cm³ vs. 2.7 ± 0.3 cm³, cerebellum 3.3 ± 0.6 cm³ vs. 4.6 ± 1.0 cm³; all p < 0.05; early-onset FGR n = 43, AGA n = 30, analysed across three gestational age windows of 10 fetuses each). For late-onset FGR fetuses (32–37 weeks; n = 55 vs. AGA n = 30, analysed across three gestational age windows of 10 fetuses each), cerebral volume was significantly reduced compared with AGA fetuses across all gestational age subgroups (e.g., at 32–33 weeks: 160.1 ± 18.7 cm³ vs. 194.1 ± 18.2 cm³; p < 0.05), while there was no statistically significant difference in brain stem volume between late-onset FGR and AGA fetuses at 34 weeks(4.6 ± 0.7cm³ vs. 5.0 ± 0.5cm³; BH-FDR adjusted p = 0.228) or later(5.2 ± 0.7cm³ vs. 5.5 ± 0.6cm³; BH-FDR adjusted p = 0.206), nor in cerebellum volume at 36 weeks or later(14.0 ± 1.6cm³ vs. 15.5 ± 2.2cm³; BH-FDR adjusted p = 0.110).

Discussion

MRI measurement of fetal brain volume can serve as a predictor of brain development in 2 subtypes of FGR. In early-onset FGR, volumes of the brainstem, cerebellum and cerebrum consistently showed smaller growth prior to 32 weeks. Late-onset FGR was characterized by persistently smaller cerebrum volume. Brainstem volume was no longer significantly different from AGA levels by 34 weeks, and cerebellum volume approached AGA levels after 36 weeks.

Background

Fetal growth restriction (FGR), characterized by low body weight due to pathological factors that impair fetal growth potential, affects 5–10% of pregnancies and approximately 30 million fetuses worldwide each year [1]. Studies have found that FGR is most commonly caused by placental insufficiency, resulting in significant perinatal morbidity and mortality [2]. Despite a universal consensus on its definition has not been reached, there is global agreement on its classification into early-onset and late-onset subtypes according to the timing of occurrence [3].

Early-onset FGR refers to intrauterine growth restriction that occurs earlier than 32 weeks of gestation, while late-onset subtype occurs after this gestational week. This classification is based on studies identifying 32 weeks of gestation as the cutoff for differentiating between early- and late-onset FGR fetuses, as it helps maximize the differences in underlying etiology, Doppler parameters, complications, and pregnancy outcomes [4]. Early-onset FGR is more prone to placental lesions and generally has a poorer prognosis compared to the late-onset subtype [5].

Ultrasound screening (US) is the primary method for prenatal evaluation of fetal development, allowing real-time observation of fetal structure, growth parameters and placental condition. It can also detect deformities, monitor growth and development, offering the advantages of being non-invasive, safe, and repeatable, which benefits both eugenics and clinical decision-making. In recent years, fetal magnetic resonance imaging (MRI) has been widely used in major prenatal diagnostic centres. It is a crucial tool for evaluating fetal brain development and complements the US technique. MRI has been proved as an effective means in evaluating FGR fetal brain injury or structural defects, calculating brain segmentation volumes, and assessing brain gyrus development [6–8].

Currently, there are few studies on the application of brain segmentation volume calculation to assess FGR fetal brain development, and the findings remain controversial. This study compares the brain segmentation volumes of fetuses with 2 subtypes of FGR to those of appropriate for gestational age (AGA) fetuses, and evaluates potential differences in fetal brain development. We propose the following hypothesis: Compared with AGA fetuses, fetuses with early-onset FGR have significant differences in brain volume (including the cerebral cortex, cerebellum, and brainstem), while compared with the AGA fetuses, fetuses with late-onset FGR may not have significant differences in brain volume (including the cerebral cortex, cerebellum, and brainstem) as the gestational weeks increase.

Materials and methods

Participants selection

This retrospective study included pregnant women who underwent MRI examinations at our hospital due to suspected fetal development abnormalities identified through ultrasound screening (US) between January 2021 and March 2025. Among these, 43 cases were clinically diagnosed with early-onset FGR, and 55 cases with late-onset FGR. The diagnostic criteria were based on the 2020 guidelines from the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) [9]. Fetal weight was estimated using the Hadlock formula based on biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC), and femur length (FL). Gestational age was confirmed by a combination of the last menstrual period and first-trimester early ultrasound scanning for all enrolled pregnant women.Early-onset FGR is defined by any of the following criteria: an estimated fetal weight (EFW) or abdominal circumference below the 3rd percentile for the corresponding gestational age by ultrasound before 32 weeks of pregnancy; loss of end-diastolic umbilical artery blood flow; or EFW or abdominal circumference below the 10th percentile for the corresponding gestational age, combined with a uterine artery or umbilical artery pulsation index above the 95th percentile for the corresponding gestational age. Late-onset FGR is diagnosed after 32 weeks of gestation if any of the following conditions are met: EFW or abdominal circumference is below the 3rd percentile for the corresponding gestational age, or two of the following criteria are satisfied: EFW or abdominal circumference is below the 10th percentile for the corresponding gestational age; EFW or abdominal circumference decreases by more than two quartiles; cerebral placenta rate is below the 5th percentile for the corresponding gestational age, or umbilical artery pulsatile index is above the 95th percentile for the corresponding gestational age. In this study, early-onset FGR cases were limited to MRI scans performed at 26–31 weeks of gestation, while late-onset FGR cases were restricted to 32–37 weeks of gestation. Each FGR subgroup was separately compared with AGA fetuses within the identical gestational age window to eliminate the confounding effect of gestational age. Sixty cases of appropriate for gestational age (AGA) fetuses are included, all of which are single-fetal pregnancies between 26 and 37 weeks of gestation. Ten AGA fetuses were consecutively enrolled in each gestational window in chronological order of their MRI examination dates: all AGA-eligible candidates during the study period were sorted by examination date, and the first 10 within each window were included. The AGA group comprised fetuses who underwent MRI for maternal or obstetric indications unrelated to suspected fetal brain pathology. The most common indications included maternal anxiety with inconclusive ultrasound, evaluation of extra-cranial findings (e.g., suspected lung mass or abdominal anomaly), or placental localization. Importantly, all fetuses in the AGA group were confirmed to have normal brain anatomy on MRI, normal biometric growth parameters (EFW > 10th percentile).

Written informed consent was obtained from all individual participants prior to enrolment. All pregnant women voluntarily underwent MRI examinations and signed informed consent for regular prenatal care. We retrieved basic information about pregnant women and fetal MRI images from the PACS system. The exclusion criteria include: (1) missing or incomplete clinical history, (2) poor brain image quality in all MRI columns due to motion artifacts or non-standard acquisition of scanning planes in fetal head images, (3) presence of severe central nervous system complications in addition to FGR, or (4) MRI examinations conducted at 32 weeks and beyond for early-onset fetal growth restriction.

Fetal brain MRI scan

The scan was performed using a 1.5T MRI system (Philips) equipped with a phased array coil. Pregnant women were positioned supine and required to cooperate with breath-holding when necessary. MRI scanning parameters are shown as following (Table 1). The films were jointly analyzed by two radiologists with prenatal diagnostic qualifications; the fetal head circumference was measured at the level of thalamus in the cross-section of the fetal brain (Fig. 1).

Table 1.

MRI scanning parameters

Sequences	Scanning parameters
SSH_TSE sequence	Transverse, sagittaland coronal planes with a slice thickness of 4 mm and a slice spacing of -1 mm; FOV: 400 mm; Matrix: 256 × 256 mm; acquisition time: 25 s; TR 8000 ms, TE 120 ms
BTFE_BH sequence	Transverse, sagittaland coronal planes with a slice thickness of 4 mm and a slice spacing of -1 mm; FOV: 400 mm; Matrix: 256 × 256 mm; acquisition time: 35 s; TR 4 ms, TE 2 ms
T1WI sequence	Transverse, sagittaland coronal planes with a slice thickness of 4 mm and a slice spacing of -1 mm; FOV: 400 mm; Matrix: 256 × 256 mm
DWI sequence	Transverse, sagittaland coronal planes with a slice thickness of 4 mm and a slice spacing of -1 mm; TR3000-4000ms, TE72.7-76.3ms; FOV: 400 mm; Matrix: 256 × 256 mm, Number of excitations: 2, b value selection: 0 and 800 s/mm²

Fig. 1.

Fetal MRI head circumference measurement standard level. Gestational week 31, SSH_TSE sequence, axial plane, matrix 1200 × 1200.The HC was measured as the outer perimeter of the fetal calvarium (dashed line). Key anatomical landmarks include the bilateral frontal bones anteriorly, parietal bones laterally, and occipital bone posteriorly

Image segmentation and volume calculation

SSH sequence of brain MRI images were independently segmented by two radiologists with prenatal MRI diagnostic credentials using ITK-SNAP. The manual segmentation method was used for this process. The segmented fetal craniocerebral anatomy includes cerebrum, brainstem, and cerebellum, and images of different anatomical structures are annotated using different colours (Fig. 2). After segmentation, the software is used to measure and calculate the volume of each anatomical structure of the fetal brain corresponding to the pixel number of different anatomical tissues (unit: cm³). In the study, if the difference between the measurement data of the two surveyors is ≤ 10%, the average value is taken directly; if the gap is > 10%, a third physician who conducts prenatal MRI diagnosis will review the data, and the final value is determined as the average value of the two measurements.

Fig. 2.

Fetal MRI images of different anatomical structures. A-C, Gestational week 27. D-F, Gestational week 30. G-I, Gestational week 33. J-L, Gestational week, 36. Brainstem is marked blue; cerebellum is marked green, and cerebrum is marked red. A 3-dimensional image is generated

Statistical analysis

Categorical data were expressed as percentages, and continuous data were presented as mean ± standard deviation (SD) or median (interquartile range, IQR), as appropriate. Comparisons between the FGR and AGA groups were performed using Fisher’s exact test for proportions and independent two-sample t-test for continuous variables. To control the false positive rate induced by repeated pairwise comparisons across multiple gestational windows and brain structures, the Benjamini–Hochberg false discovery rate (BH-FDR) correction was applied for all multiple testing, with an adjusted significance threshold of p < 0.05, and adjusted p-values were reported accordingly. Inter-rater reliability for scores assigned by different evaluators to the same set of subjects was assessed using the intraclass correlation coefficient (ICC). The relationship between two continuous variables, gestational age and brain volume, was visualized using a scatter plot. Correlations between fetal brain anatomical structures and both head circumference and gestational age were examined using Pearson’s correlation analysis. Correlation strength was interpreted as follows: 0.8-1.0 for very strong, 0.6–0.8 for strong, 0.4–0.6 for moderate, and 0.2–0.4 for weak. A two-sided p-value < 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS software (version 22.0).

Results

According to the inclusion and exclusion criteria, a total of 109 cases with FGR and 60 cases with AGA were enrolled in this study. Among them, 9 cases in the FGR group were excluded owing to severe central nervous system complications, and an additional 2 cases were excluded because of insufficient quality of brain MRI images for reliable evaluation. Including 60 cases in the normal group (AGA group) and 98 cases in the experimental group (FGR group). The basic information of pregnant women, data of fetal head circumference and sex are shown in Table 2. The evaluation of measurement data consistency between the two prenatal diagnostic physicians revealed a two-way random-effects, single-rater intraclass correlation coefficient (ICC 2,1) of 0.87 (95% CI: 0.71–0.92, p < 0.001), indicating strong inter-rater agreement.

Table 2.

Maternal, fetal, biometry and doppler characteristic

Characteristics	FGR Group	AGA Group
	(n = 98)	(n = 60)	P value
Maternal
Age at conception, mean (SD), y	29.31 ± 3.96	29.78 ± 4.31	0.47
Height, mean (SD), cm	157.70 ± 4.50	157.90 ± 4.53	0.74
Weight, mean (SD), kg	60.77 ± 7.23	66.89 ± 9.11	0.08
Gestational age, median (P25-P75)	32(29–34)	32(29–35)
Hypertensive disorders of pregnancy, No.(%)	11(11.2)	2(3.3)	0.32
Gestational diabetes mellitus, No.(%)	18(18.7)	8(13.3)	0.48
Smoking, No.(%)	3(3.1)	1(1.7)	0.99
Alcohol, No.(%)	4(4.1)	2(3.3)	0.99
Fetal
Sex, No.(%)
Male	31(31.6)	23(38.3)	0.39
Female	67(68.7)	37(61.7)
Biometry and dopplers*
EFW or AC below the 3rd percentile, No.(%)	47(48.0)	NA
Loss of end-diastolic umbilical artery blood flow, No.(%)	0(0)	NA
EFW or AC below the 10th percentile with a uterine artery or umbilical artery pulsation index above the 95th percentile, No.(%)	20(20.4)	NA
Others, No.(%)	31(31.6)	NA

Abbreviations: EFW Estimated fetal weight, AC Abdominal circumference; *The diagnostic criteria are based on the 2020 guidelines from the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG); NA Not applicable

After segmentation of each fetal brain MRI image, the volumes of cerebrum, brainstem, and cerebellum were calculated at different gestational weeks in both FGR and AGA groups (Table 3). Differences in cerebrum, brainstem, and cerebellum volumes between early-onset FGR fetuses and AGA fetuses were statistically significant. Statistical analyses revealed significant differences in cerebrum volumes between late-onset FGR fetuses and AGA fetuses, whereas no significant differences were indicated in brainstem volumes between FGR fetuses and AGA fetuses at 34 weeks of gestation or later, nor in cerebellum volumes between FGR fetuses and AGA fetuses at 36 weeks or later (Table 3).

Table 3.

Statistical differences in brain volumes between FGR fetus and AGA fetus

	Gestational week		AGA Group		FGR Group	P value (BH-FDR adjusted)
		n	mean x ± SD	n	mean x ± SD
Head circumference (mean x ± SD, cm)	26–27	10	25.8 ± 0.8	11	23.1 ± 0.6	< 0.001
	28–29	10	26.7 ± 0.6	18	25.0 ± 0.6	< 0.001
	30–31	10	28.8 ± 0.6	14	26.6 ± 0.9	< 0.001
	32–33	10	29.7 ± 0.6	27	27.9 ± 0.5	< 0.001
	34–35	10	31.1 ± 0.9	18	29.4 ± 0.9	< 0.001
	36–37	10	31.9 ± 0.8	10	29.9 ± 1.0	< 0.001
Brainstem volume (mean x ± SD, cm3)	26–27	10	2.7 ± 0.3	11	1.7 ± 0.2	< 0.001
	28–29	10	2.8 ± 0.4	18	2.0 ± 0.3	< 0.001
	30–31	10	3.0 ± 0.5	14	2.4 ± 0.3	0.006
	32–33	10	4.1 ± 0.5	27	3.6 ± 0.3	0.034
	34–35	10	5.0 ± 0.5	18	4.6 ± 0.7	0.228*
	36–37	10	5.5 ± 0.6	10	5.2 ± 0.7	0.206*
Cerebellum volume (mean x ± SD, cm3)	26–27	10	4.6 ± 1.0	11	3.3 ± 0.6	0.003
	28–29	10	5.6 ± 0.9	18	4.9 ± 0.5	0.013
	30–31	10	8.5 ± 1.2	14	6.5 ± 0.9	< 0.001
	32–33	10	10.4 ± 1.1	27	7.2 ± 1.4	< 0.001
	34–35	10	13.2 ± 2.0	18	10.5 ± 2.0	0.005
	36–37	10	15.5 ± 2.2	10	14.0 ± 1.6	0.110*
Cerebrum volume(mean x ± SD, cm3)	26–27	10	113.2 ± 10.3	11	87.3 ± 9.2	< 0.001
	28–29	10	131.6 ± 8.8	18	112.5 ± 23.4	0.024
	30–31	10	158.6 ± 10.9	14	139.9 ± 17.1	0.008
	32–33	10	194.1 ± 18.2	27	160.1 ± 18.7	< 0.001
	34–35	10	232.9 ± 18.3	18	197.6 ± 17.0	< 0.001
	36–37	10	258.9 ± 27.2	10	226.4 ± 13.8	0.005

The FGR group in this table is divided into two independent subgroups: early-onset FGR (26–31 weeks) and late-onset FGR (32–37 weeks), which were matched and compared with AGA fetuses at corresponding gestational ages, rather than a mixed unified FGR cohort. P values were adjusted using the Benjamini-Hochberg false discovery rate (FDR) correction for multiple comparisons across all 24 tests (4 structures × 6 gestational age windows)

* indicates BH-FDR adjusted p > 0.05 (not statistically significant)

In both the FGR and AGA groups, Pearson correlation analyses indicated a very strong correlation, with coefficients above 0.8, between cerebrum, brainstem, and cerebellum volumes, as well as head circumference and gestational week (Table 4, Fig. 3).

Table 4.

Correlation between brain volumes and gestational week and head circumference in FGR and AGA fetuses (Pearson correlation coefficients)

	FGR group		AGA group
Volumes	Gestational week	Head circumference	Gestational week	Head circumference
Brainstem	0.917	0.867	0.909	0.880
Cerebellum	0.888	0.844	0.952	0.942
Cerebrum	0.936	0.910	0.964	0.959

Fig. 3.

Scatter plots of FGR and AGA fetal brain volumes including cerebrum, cerebellum and brainstem (cm3, weeks). The blue, green, and red fitted lines represent the AGA group, the early-onset FGR group, and the late-onset FGR group

Discussion

Significance of the volumes of FGR fetuses and AGA fetuses on MRI

This study used MRI data of fetal brains to obtain the volumes of fetal cerebrum, cerebellum and brainstem for both FGR and AGA fetuses through ITK-SNAP, the manual segmentation method was used. A comparative analysis and correlation analysis were conducted to determine the relationship with head circumference and gestational week. The differences in the volumes of cerebrum, brainstem, and cerebellum between early-onset FGR fetuses and AGA fetuses was statistically significant. The differences in cerebral volumes between late-onset FGR fetuses and AGA fetuses were statistically significant, but there was no statistically significant differences in brainstem volumes at 34 weeks or later nor cerebellum volumes at 36 weeks or later. In both early- and late-onset FGR fetuses, cerebrum, brainstem, and cerebellum volumes are strongly correlated with head circumference and gestational age.

In a prospective observational study, researchers use 3D-MRI imaging of 35 FGR fetuses at a median gestational age of 30 weeks. Compared with 79 AGA fetuses, cerebral and cerebellar volumes in the FGR group are significantly smaller than those in the AGA group, and there is no significant difference in the brainstem volumes between the two groups [7]. A cohort study analyses MRI scan data from 26 FGR fetuses due to diagnosis of placental insufficiency, and measures the volumes of the supratentorial brain, left and right hemispheres, and cerebellum using semi-automatic methods, and the absolute volumes and percentiles of all brain structures in the FGR group are smaller compared to those of 66 fetuses in the control group [10]. A prospective observational case-control study found that 14 FGR fetuses at 35 weeks of gestation had smaller brain volumes compared to 26 non-FGR fetuses [11]. Controversially, a study comparing fetal MRI scans between 20 FGR fetuses and 19 AGA fetuses at 20-36 weeks of gestation showed no significant difference in brain volumes [12]. Another study examined fetal MRI scans from 40 cases of placental-derived FGR and compared them with 78 cases of non-placental lesions, and there was no significant difference in fetal brain volumes between the two groups [13]. Conversely, there is research reported larger cerebellum volumes in FGR fetuses compared to AGA fetuses at 37 weeks of gestation [14]. L. Meijerink et al. conducted a systematic review of all the research related to FGR from 1985 to 2023. Their findings indicate that MRI can provide valuable insights into fetal brain development, especially for those at risk of growth restriction and brain development abnormalities. In the group of FGR, overall brain and cerebellar volumes were smaller, with lower diffusion coefficient (ADC) values in the anterior brain, frontal lobe, parietal lobe, thalamus, subcallosal area of the corpus callosum, basal ganglia, pons, and cerebellum [15]. A recently published prospective study established gestational age-specific normative fetal brain volume curves based on fetal MRI measurement, which provides standardized volumetric reference values across gestation for fetal brain research [16]. With this unified normative reference dataset available, future follow-up research of our team can further conduct quantitative inter-study comparison by calibrating measured brain structural volumes against the normative range proposed by Bouachba et al., helping to unify the quantitative evaluation standard of fetal brain volume in FGR-related studies.

In our view, divergent conclusions in existing studies can primarily be attributed to several key factors: first, recent revisions in the diagnostic criteria for FGR; second, the inherent limitations of ultrasound-based fetal weight estimation, particularly for FGR cases near the 10th percentile; and third, small sample sizes in many studies, coupled with potential selection bias in subject recruitment.

The clinical feasibility of FGR subtypes on MRI scan

FGR can be categorized into two subtypes: early onset and late onset. Early-onset FGR refers to fetuses diagnosed before 32 weeks of gestation, while late-onset FGR is diagnosed after 32 weeks. Early-onset FGR is primarily associated with placental insufficiency and tends to have a relatively poor prognosis. In the present study, before 32 weeks, significant differences were observed in the volumes of the cerebrum, brainstem, and cerebellum between early-onset FGR fetuses and AGA fetuses. However, as gestational age progressed, the differences in brainstem volumes between late-onset FGR fetuses and AGA fetuses were not statistically significant at 34 weeks or later, nor were the cerebellum volumes at 36 weeks or later between the two groups. It is important to note that, given the limited sample size per gestational window, the absence of statistically significant differences in brainstem and cerebellum volumes at later gestational ages may reflect insufficient statistical power rather than a true biological convergence, and thus these findings should be interpreted with caution. Our findings were consistent with recent studies, indicating that in late-onset FGR fetuses, fetal brain development has been consistently smaller as gestational age increases, while the development gaps in the brainstem and cerebellum tend to diminish compared to AGA fetuses. This interesting finding suggests that in clinical practice, extending gestational age may be associated with favorable brain development in late-onset FGR fetuses; however, the absence of neonatal neurodevelopmental follow-up data in the present study precludes any definitive interpretation in terms of clinical brain-sparing efficacy. This was consistent with the results of a recent cohort study on the neurodevelopment of late-onset FGR fetuses [17]. The brain-sparing phenomenon in FGR was originally characterized by Doppler ultrasound studies documenting a progressive sequence of hemodynamic changes—from umbilical artery abnormalities through middle cerebral artery redistribution to ductus venosus deterioration—reflecting worsening placental insufficiency [18]. In a study on the patterns of brain sparing in a cohort of late-onset FGR fetuses, it was found that ACA + MCA vessel abnormalities outnumbered PCA + VA abnormalities [19]. Together, the observation is interpreted as possibly reflecting a differential or smaller volumes adaptation in late-onset FGR, rather than a uniform “catch-up” process, highlighting the need for subtype-specific monitoring strategies.

In contrast, early-onset FGR fetuses constantly show relatively smaller development in the cerebrum, brainstem and cerebellum throughout gestation.An association has been reported between perinatal outcomes and FGR subtypes, suggesting that late-onset FGR fetuses typically experience milder clinical outcomes due to less severe blood flow abnormalities, while early-onset FGR fetuses tend to develop more severe clinical outcomes that may be linked to significant uterine placental vascular insufficiency, maternal hypertension, or genetic disorders [20, 21]. Based on the results of these related studies, the pathogenesis mechanisms of 2 subtypes of FGR were found to vary. More reliable research are needed to clarify the pathogenesis mechanisms across FGR subtypes and to explore the prenatal and postnatal neurodevelopmental outcomes and their correlations in FGR fetuses.Recently, functional MRI imaging techniques, such as SWI [22] and T2*[23] Sequences have been increasingly applied to evaluate the hemodynamics and cerebral oxygen levels in FGR fetuses, allowing for non-invasive evaluation of fetal brain tissue oxygenation, microbleeding and venous blood flow distribution. However, the systematic review related to FGR by L. Meijerink et al. [15] found no significant difference in T2*. But these methods provide new imaging biomarkers for monitoring cerebral blood perfusion and hypoxia-induced injury in FGR fetuses.

Clinical implications

From a clinical perspective, MRI-based brain volume quantification holds several potential applications in the management of FGR. First, the differential volumetric patterns between early-onset and late-onset FGR observed in this study—persistently smaller cerebrum, brainstem, and cerebellum in early-onset FGR versus predominantly smaller cerebrum with a tendency toward convergence of brainstem and cerebellum volumes in late-onset FGR—may aid in subtype-specific risk stratification and inform the frequency and modality of antenatal surveillance. Second, the strong correlation between brain volumes and head circumference as well as gestational age suggests that MRI volumetry could serve as a complementary tool to ultrasound for monitoring fetal brain growth trajectories, particularly in cases where ultrasound visualization is limited by maternal habitus or fetal position. Third, quantifying the degree of brain volume deviation may assist in prenatal counseling by providing objective, structure-specific information regarding the extent of cerebral involvement, which may help clinicians and families make more informed decisions regarding delivery timing and perinatal management. However, it should be emphasized that brain volume measurements alone are currently insufficient to independently guide clinical decisions; rather, they should be integrated with Doppler hemodynamic parameters, fetal heart rate monitoring, and overall obstetric assessment within a multimodal evaluation framework [9]. Prospective studies with longitudinal follow-up are needed to establish whether the volumetric parameters reported here translate into improved neonatal neurodevelopmental outcomes [15].

Strengths and promising prospects

Currently, there are relatively few studies on the precise segmentation of FGR brain tissues, primarily due to the special challenges associated with fetal MRI imaging. Traditional MRI scans often face significant challenges in obtaining high-resolution, motion-free images due to the fetus’s autonomous movements and the influence of maternal respiratory movements, further complicating subsequent segmentation and analysis of brain tissues. Additionally, either manual or semi-automatic segmentation of fetal brain tissues are time-consuming, making it difficult to implement such methods in clinical practice [24, 25].

Despite these challenges, advancements in medical imaging technology offer promising prospects for the future of fetal MRI brain tissue segmentation research. First, the use of high-field strength MRI scanners and rapid imaging sequences can significantly enhance image resolution and reduce motion artifacts, thereby providing high-quality raw data for subsequent segmentation. Second, the optimization of deep learning and AI-assisted segmentation algorithms is expected to enable automatic segmentation of fetal brain tissues, greatly reducing the time of manual intervention and improving the accuracy and repeatability of segmentation. Finally, the establishment of a multicenter, large-sample database and the development of a standardized segmentation protocol will further facilitate the clinical application of this technology in the early diagnosis, prognosis evaluation, and intervention strategy development for FGR.

Limitations and future perspectives

This research has several limitations. Firstly, as a retrospective study, MRI brain volume assessment in FGR fetuses is limited to specific gestational weeks without dynamic observations, which precludes accurate evaluation of brain development in early-onset FGR beyond 32 weeks. In our cross-sectional design, early- and late-onset FGR cases were enrolled in distinct gestational windows and compared separately with age-matched AGA controls, making direct comparison of their intracranial volume trajectories unfeasible. Additionally, due to the limited sample size of this single-center study, we grouped fetuses into two-week gestational windows instead of the more precise single-week stratification used in previous studies, which limits the granularity of our analysis. Future follow-up research will address this gap.

Secondly, fetal brain volumes were evaluated using 2D-MRI data from 1.5T equipment with manual segmentation, which is less precise than automated 3D volumetric acquisition. Future studies will therefore adopt dedicated 3D MRI sequences or advanced scanning protocols to improve measurement accuracy.

Thirdly, only the volumes of the cerebrum, brainstem, and cerebellum were analyzed; finer brain structures (e.g., cerebral lobes, basal ganglia, thalamus) were not evaluated. This is attributed to low tissue contrast and limited spatial resolution at early gestational ages, which hinders the reliable delineation of small adjacent neuroanatomical structures.

Fourthly, manual rather than automated (e.g., deep learning-based) segmentation was employed. While manual segmentation serves as a reference standard in some contexts, it is time-consuming, operator-dependent, and not scalable for large datasets. Future work will therefore focus on curating multi-center annotated datasets and developing deep learning models tailored to fetal neuroanatomy to enhance measurement efficiency, accuracy, and objectivity.

Lastly, selection bias may exist: the AGA group cannot be regarded as a strictly healthy normative population, as these fetuses underwent MRI for clinical indications (e.g., maternal anxiety, extracranial abnormalities), which impacts the representativeness of brain volume references. It is also worth noting that both FGR classification and AGA definition relied on ultrasound-derived estimated fetal weight (EFW), which carries a documented inter-operator and BMI-related measurement imprecision of approximately 10-15%, potentially introducing a degree of misclassification that may attenuate observed inter-group differences; future studies could consider incorporating MRI-based fetal body volume measurements—recently shown to provide a more accurate and BMI-independent surrogate for fetal weight [26] to refine cohort classification and reduce this uncertainty. Combined with the single-center design and limited sample size, these factors reduce the external validity of our findings. We plan to validate and expand our findings through a multi-center cohort study.

In summary, future studies will optimize FGR fetal MRI protocols, expand the clinical validation cohort, conduct multi-center research to enlarge the sample size and adopt single-week gestational age grouping as suggested, and combine multimodal imaging with long-term neurodevelopmental follow-up to clarify the clinical value of these parameters in the early diagnosis and prognosis of FGR.

Conclusion

MRI measurement of fetal brain volume can serve as a predictor of brain development in 2 subtypes of FGR. In the early-onset FGR, volumes of the brainstem, cerebellum and cerebrum consistently showed smaller volumes prior to 32 weeks. Late-onset FGR was characterized by persistently smaller cerebrum volume. Brainstem volume was no longer significantly different from AGA levels by 34 weeks, and cerebellum volume approached AGA levels after 36 weeks.

Abbreviations

FGR

Fetal growth restriction

AGA

Appropriate gestational age

US

Ultrasound screening

MRI

Magnetic resonance imaging

EFW

Estimated fetal weight

Authors’ contributions

CF X and S P investigated and wrote the manuscript. CF X, S P and CY Z conducted statistical analyses and designed this study. CF X, W T and CY Z participated in data collection, acquisition, analyses and interpretation and assisted in completing the draft. All contributors reviewed the manuscript and agreed to publish it.

Funding

The Chongqing young and middle-aged medical high-end talent project (YXGD202472) supported this study.

Data availability

The data used or tested in this study can be obtained from the corresponding authors upon request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki developed by the World Medical Association in 2000. The Clinical Research Ethics Committee of Chongqing Health Center for Women and Children (Women and Children’s Hospital of Chongqing Medical University) has approved this study (No: 2025057). Written informed consent was obtained from all individual participants prior to enrolment. After the benefits of the study were explained, the participants agreed to take part. Participants were informed that they had right to withdraw from the study at any time. Confidentiality is ensured by excluding questions that contain identifying information and by storing completed questionnaires and results in secure areas.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.