Magnetic Resonance Imaging (MRI) has emerged as a crucial tool for visualizing brain structures and functions due to its non-invasive nature, high spatial resolution, and ability to provide detailed anatomical and functional insights (Thompson et al., 2005).

Traditionally, most MRI scans are conducted with subjects in a supine position, a practice necessitated by the design constraints of conventional MRI scanners (Eşer et al., 2007). Although effective for many diagnostic purposes, this fixed, non-physiological posture may not fully represent how brain structures and functions operate, as human bodies are typically more dynamic and engage in various postures in daily life (Alperin et al., 2005a). Posture is known to significantly influence physiological and anatomical characteristics across the body. For example, shifting from lying down standing can alter spinal alignment, change lung capacity, and modify cardiovascular dynamics, among other effects (Rossberg and Peňaz, 1988; Fawcett et al., 2021). These observations underscore the importance of incorporating different postures in studies of body systems and their functions. However, traditional MRI systems are limited in their ability to assess posture-related structural differences, highlighting the need for innovative multi-positional MRI technologies.

To overcome these limitations, researchers have introduced various MRI systems capable for upright or weight-bearing imaging, including Fonar’s 0.6T system (Nicholson et al., 2023), Siemens’ 1.0T system (Gufler et al., 2004), and G-Scan’s 0.25T system (Lee et al., 2015b), et al. Although these systems operate at relatively low magnetic field strengths, which may reduce image resolution and signal-to-noise ratio (SNR) during scanning (Marques et al., 2019), they have proven the research value in the effects of specific postures on body structures (Shymon et al., 2014; Hansen et al., 2019). The 1.5T MRI system provides higher SNR and spatial resolution, allowing for clearer visualization of small lesions and anatomical structures, thus enhancing the quantitative analysis capabilities of the images and assisting clinicians in making more accurate diagnostic and treatment decisions (Magee et al., 2003; Arnold et al., 2023; Schmid et al., 1999). For instance, upright MRI has proven notably effective in detecting spinal stenosis (Fawcett et al., 2021) and in visualizing medial meniscus extrusion in the knee under weight-bearing conditions (Draper et al., 2011). Additionally, these systems offer significant benefits in the early diagnosis of pelvic organ prolapse (Shaikh et al., 2021), thereby facilitating more prompt and effective treatment. Despite significant technological advancements, most existing studies have focused on other body systems, leaving research on the impact of posture on brain morphology relatively underexplored. Regions of the brain, such as the pituitary gland, cerebellum, and choroid plexus, which are more susceptible to gravitational effects, are particularly affected. A previous study has shown that the choroid plexus undergoes morphological changes under the influence of gravity (Wang et al., 2025). These findings highlight the gap in our understanding of how posture affects brain structure and emphasize the pressing need for innovative MRI systems specifically designed for brain research.

While traditional MRI systems and some newly developed upright MRI systems have significantly advanced our understanding of postural effects on other body systems, research in the brain domain remains limited due to technical and methodological challenges. In prior studies, the evaluation of brain regions in the upright posture was typically done through visual assessment by two blinded clinicians, without image segmentation. This approach likely arose from the lack of automatic segmentation technology available in 2001, and the image quality at that time may not have met the requirements for accurate segmentation (Nakada and Tasaka, 2001). Additionally, there was relatively limited research on upright head imaging. To address these challenges, we have developed a cryogen-free 1.5T superconductive MRI system that can operate with an active magnetic field during scanner rotation (Wang et al., 2023). This system offers a unique opportunity to study the impact of different postures on brain morphology under conditions that closely mimic everyday physiological activities. This innovative design allows for high-resolution imaging in both supine and upright positions without compromising image quality.

The objective of this study is to investigate and validate whether significant volumetric changes occur in various brain regions between upright and supine positions in healthy individuals. By identifying and quantifying these differences, we aim to elucidate the potential clinical implications of postural effects on brain structure. These findings could offer valuable insights for advancing clinical brain research, especially in conditions where postural influences may play a significant role.

All imaging was conducted using a state-of-the-art, rotatable, cryogen-free superconductive 1.5T MRI scanner (XGY-Spin MRI-R001, XGY, Ningbo, China), which served as the primary imaging platform (Figure 1). This prospective study was carried out between August to September 2024, with ethical approval granted by the institutional ethics committees of both the University of Nottingham Ningbo China, and Ningbo No. 2 Hospital. Written informed consent was obtained from all participants, ensuring compliance with established guidelines for human research. Additionally, participant confidentiality was meticulously maintained by anonymizing data prior to analysis.

The inclusion criteria for eligible participants were aged 20–35 years, without any history of neurological or psychiatric disorders, and met all MRI scanning safety requirements. The exclusion criteria included significant brain abnormalities, a history of brain surgery, systemic diseases affecting brain structure (e.g., hypertension or diabetes), or pregnancy. Initially, 33 individuals were enrolled; however, one was excluded due to the presence of a cerebral lipoma. Additionally, due to a computer system error, data from one individual was corrupted, leading to exclusion. As a result, the final cohort consisted of 31 healthy volunteers (mean age: 26.5 ± 3.7 years; comprising 16 males and 15 females). Each subject underwent an MRI scan using a 3D T1-weighted MPRAGE sequence, with the following settings: Repetition Time (TR) = 10.0 ms, Echo Time (TE) = 3.4 ms, Field of View (FOV) = 230 mm × 230 mm × 187.2 mm, Sampling Matrix = 192 × 192 × 156, Reconstruction Matrix = 480 × 480 × 156, Spatial Resolution = 0.48 mm × 0.48 mm × 1.20 mm, Slice Thickness = 1.20 mm, and Flip Angle = 12°. The scan time for the 3D MP-RAGE sequence was 4 min and 15 s. Scans were performed in two different postures—supine (0 degrees) and upright (90 degrees). For supine scans, participants’ heads were supported by foam pads to minimize motion. In the upright posture, restraint straps and additional foam pads were used to secure and align the head, effectively reducing motion artifacts.

A comprehensive volumetric analysis of 109 brain regions was performed using the United Imaging analysis system to assess structural changes between the two postures. The automated whole-brain segmentation from 3D T1-weighted MPRAGE images was conducted via the uAI Research Portal (United Imaging Intelligence, China) (Wu et al., 2023), a clinical research platform developed using Python (version 3.7.3). This process utilized the PyRadiomics package¹ (Liu et al., 2022), enabling precise segmentation of the whole brain into 109 distinct regions and providing volumetric measurements for each region to assess structural changes between the two postures (Figure 2).

To assess both the reproducibility and repeatability of this analysis method, an experiment was conducted with three selected subjects. Measurements of brain volume in the supine posture were performed for each subject at three time points: Day 1 (first scan), Day 30 (second scan, initial scan), and Day 30 (third scan, taken 30 min after the second scan). This approach aimed to determine the stability of measurements over time and the reliability of the method. The coefficient of variation (COV) (Bland and Altman, 2010) was calculated using the formula: COV = (Standard deviation/Mean) × 100% (Abdi, 2010). And the average COV was calculated across the three subjects. In this study, the COV was expressed as a percentage, providing a depiction of measurement variability and aiding in the assessment of the precision and consistency of the measurements.

Subsequently, brain volume was measured in both the supine and upright positions for all 31 subjects to evaluate changes in brain region volumes between these two postures. Additionally, the percentage change in brain volume between the upright and supine measurements was calculated using the formula: (V_upright − V_supine)/(V_upright) × 100%, where V_supine and V_upright represent the brain volume in the supine and upright postures, respectively. To ensure the validity of the parametric analysis, the Shapiro–Wilk test was conducted for each brain region to assess whether the volume data conformed to a normal distribution. Only regions that satisfied the normality assumption were included in the subsequent two-tailed paired Student’s t-test. For regions that did not meet the normality assumption, the non-parametric Wilcoxon signed-rank test was applied instead. These statistical tests were used to evaluate volume differences between the two positions. Multiple comparison corrections were also applied to control for Type 1 errors.

Table 1 presents the demographic characteristics of the participants, all of whom are young adults. This selection minimizes the influence of age-related brain volume decline. Additionally, the balanced gender ratio helps reduce the potential impact of sex differences on brain structure.

Imaging was conducted on all subjects using the spin MRI, with no adverse effects. Figure 2 illustrates the segmentation of 109 distinct brain regions from 3D T1-weighted MPRAGE images of a sample subject in both upright and supine positions, performed using the uAI Research Portal. Volumetric measurements for each region were obtained to evaluate structural differences between the two positions.

As shown in Figure 3, the average COV across the three subjects was 4.61% ± 2.60%, indicating good measurement repeatability. The COV values remained consistent across the three time points: Day 1 (first scan), Day 30 (second scan, initial scan), and Day 30 (third scan, conducted 30 min after the second scan). The absence of significant fluctuation further suggests that the scanning method provides reliable precision and temporal stability. Overall, the low mean COV reflects the high repeatability and low variability of the brain volume measurements in the supine posture.

To further assess the measurement precision of the scanning method, we calculated the coefficient of variation (COV) for brain volume measurements in three subjects across the three designated time points: Day 1 (first scan), Day 30 (second scan, initial scan), and Day 30 (third scan, conducted 30 min after the second scan). For each subject, the COV was determined by dividing the standard deviation of the three volume measurements by their mean (Figure 3b). The final mean COV was obtained by averaging the individual COV values across the three subjects (Figure 3a).

Figure 4 shows the overlap of the coefficient of variation (COV) values with the percentage volume changes across 109 brain regions between upright and supine postures. The gray bars represent brain regions with volume changes within the ±5% range, while the blue bars indicate regions with volume changes exceeding 5%. Most brain regions exhibit volume changes within the ±5% range, with only a few regions exceeding this threshold, suggesting that the volume changes between the two postures are very stable for the majority of brain regions.

Figure 4 illustrates that most of the gray bars, representing the average postural volume changes across the 31 subjects, fall within the red box, which denotes the average COV across the three selected subjects for each segmented brain region. This suggests that most brain regions are highly stable, with their postural changes resembling those observed in the same supine posture across different time points.

Figure 5 presents a scatter plot illustrating the relationship between percentage volume changes and p-values. The majority of brain regions display minimal volume changes, with p-values approaching 1, whereas a few regions near the threshold of significance with p-values close to 0.05, further underscoring the overall stability of brain volumes. Multiple comparison corrections were also applied, and the results remained consistent, reinforcing the stability and robustness of the observed findings.

Most brain regions exhibited minimal volume changes between upright and supine postures. Notably, there was no significant difference in intracranial volume (ICV) between the two positions: supine position ICV was 1407981.60 ± 138923.61 mm³, and upright position ICV was 1402848.68 ± 135176.12 mm³ (p = 0.89). Figure 6a demonstrates that the Inferior Lateral Ventricle showed the most substantial negative volume change (−18.56%), which exceeded the ±5% threshold; however, this difference was not statistically significant (p = 0.27). In contrast, the hippocampus depicted in Figure 6b showed a negligible volume change of 0.01% (p = 0.97), exemplifying the typical stability observed across most brain regions. Figure 6c highlights the Choroid Plexus, which showed the largest positive volume change at 3.19% (p = 0.64), yet maintained a relatively stable distribution.

To the best of our knowledge, this study is the first to employ a cryogen-free, rotatable superconductive MRI system in the investigation of brain volume changes. While the system is capable of maintaining a continuous magnetic field during scanner rotation, it is important to note that all scans in this study were conducted at fixed positions (either 0 or 90 degrees), rather than during continuous rotation. We investigated the effects of postural changes on brain volume in healthy individuals by comparing supine and upright positions using this innovative rotatable MRI scanner, quantifying the statistical differences between these two postures. Our results indicate that postural changes have a minimal impact on brain volume in healthy individuals, which supports existing knowledge about the high stability of brain structure.

Specifically, the volume changes in most brain regions between the upright and supine positions were within a ±5% range, with the majority of regions showing near-zero volume changes and p-values close to 1, further validity the stability and reliability of the brain structure. This is consistent with prior upright MRI studies, which have shown that gravity exerts a negligible effect on brain structure, even in conditions such as brain atrophy (Nakada and Tasaka, 2001). However, as illustrated in Figure 6a, the inferior lateral ventricle, where CSF is contained and circulated through, showed a positive shift in volume from upright to supine, yet the overall variability remained low. This region, adjacent to cerebrospinal fluid (CSF)-filled areas, has been documented in prior research to experience a significant increase in CSF flow when supine, suggesting that these regions may be more susceptible to postural changes and may exhibit unique anatomical or hemodynamic characteristics (Muccio et al., 2021; Alperin et al., 2015). Previous studies have reported that CSF flow, velocity, and stroke volume decrease significantly in the upright posture, with these changes closely related to body position (Yukun et al., 2025). Although our study did not include direct CSF flow measurements, the volumetric trend observed may reflect posture-dependent redistribution of CSF. These findings provide preliminary structural evidence that may contribute to understanding how posture influences CSF dynamics.

Further investigations integrating structural, functional, and flow-sensitive imaging modalities are warranted to elucidate the mechanisms by which posture influences CSF distribution and ventricular morphology.

Furthermore, previous studies have demonstrated that, compared to the supine position, the upright posture reduces CSF oscillatory volume by approximately 48%, significantly increases intracranial compliance (ICC), and decreases intracranial pressure (ICP) by approximately 2.4 times (Alperin et al., 2005b). These findings highlights the dynamic role of CSF in regulating intracranial pressure across different postural orientations, effectively maintaining intracranial pressure equilibrium. Notably, prior research has also shown that, even in cases of cerebrospinal fluid leakage, brain imaging differences between the upright and supine positions remain insignificant (Schievink and Tourje, 2007). This further underscores that, despite the significant effects of upright posture on CSF distribution and ICP, the brain’s intrinsic regulatory mechanisms ensure the stability of volume and morphology across brain regions (Lee et al., 2015a).

On the other hand, the hippocampus (Figure 6b) demonstrated the overall stability of most brain regions, with minimal variation between postures. Likewise, despite relatively larger volume changes in the choroid plexus, its low variability further highlights the consistency of measurements across different postural conditions.

The results depicted in Figure 3 demonstrate the good precision and repeatability of the current method. Measurement values for all three subjects exhibited minimal fluctuation across different time points indicating that the method provides consistent volume measurements. By calculating COV, we found that the measurement consistency for the vast majority of brain regions is high, indicating that the current measurement method demonstrates good repeatability and stability. The volume changes in brain regions induced by posture shifts mostly fall within the measurement error range, making them undetectable by the current method.

A limitation of this study is that the COV measurements were derived from data obtained from only three subjects. Such a small sample size may not be sufficiently robust to provide a comprehensive evaluation of the precision and consistency of the method at this scale. Although the results indicate that the method operates effectively within a reasonable margin, the limited sample size could introduce variability in the accuracy of the measurements. To achieve a more reliable estimation of the COV, it would be ideal to conduct multiple scans with a larger cohort, each undergoing at least three separate measurements. However, implementing such a protocol would significantly extend the required scanning time and necessitate multiple visits from the volunteers, which presents considerable logistical challenges. Consequently, the COV values observed in this study may not fully reflect the true precision of the method, highlighting the need for further studies involving larger sample sizes and repeated scans to confirm the reliability and generalizability of the findings.

Another limitation of this study concerns the accuracy of choroid plexus segmentation. Although we utilized the PyRadiomics package for segmentation of 109 brain regions, the tool is based on FreeSurfer, which has shown limited performance in choroid plexus segmentation (Zhao et al., 2020). As the choroid plexus is a critical structure influenced by body position and CSF flow, previous studies have demonstrated that deep learning-based approaches perform significantly better in this context (Yazdan-Panah et al., 2023; Eisma et al., 2024; Zhao et al., 2020). However, due to the scope of this study, these advanced deep learning techniques were not implemented. It is worth noting that a parallel study using this rotatable MRI system has analyzed the choroid plexus and provides supplementary data (Wang et al., 2025). We acknowledge that the segmentation method used in this study may not provide the same level of precision as newer techniques, which may affect the reliability of the volumetric findings related to the choroid plexus. In future work, we plan to explore these advanced deep learning techniques to improve segmentation accuracy.

In conclusion, while most brain regions showed no significant differences between supine and upright positions, these findings deepen our understanding of the potential clinical implications of postural effects on brain structure and provide valuable insights for future studies in clinical brain research.