Respiratory-gating Reduces Motion Artifacts Due to Uterine Displacement in T2-weighted MR Imaging of the Female Pelvis.

The analysis included 103 women aged 20–84 years (mean: 45.8 ± 13.1 years). The uterine position was classified as type 1, type 2, and type 3 in 75 (72.8%), 19 (18.4%), and 9 (8.7%) patients, respectively. Image quality of C-T2WI was categorized as poor, moderate, and excellent in 53 (51.5%; type 1 = 45, type 2 = 7, and type 3 = 1 cases), 34 (33.0%; type 1 = 24, type 2 = 9, and type 3 = 1 cases), and 16 (15.5%; type 1 = 6, type 2 = 3, and type 3 = 7 cases) patients, respectively. Conversely, image quality of RESP-T2WI was categorized as poor, moderate, and excellent in 27 (26.2%; type 1 = 22, type 2 = 4, and type 3 = 1 cases), 44 (42.7%; type 1 = 37, type 2 = 5, and type 3 = 2 cases), and 32 (31.1%; type 1 = 16, type 2 = 10, and type 3 = 6 cases) patients, respectively ( Fig. 6 ). The inter-observer agreement κ was 0.63 ( P  < 0.001). Discrepancies between the 2 radiologists were observed in 81 cases (39.3%). The concordant and discordant results between C-T2WI and RESP-T2WI were 52 (50.5%) and 51 (49.5%), respectively. Cases with higher and lower visual assessment of RESP-T2WI compared to C-T2WI were 43 (41.7%; type 1 = 31, type 2 = 10, and type 3 = 2 cases) and 8 (7.8%; type 1 = 4, type 2 = 2, and type 3 = 2 cases) cases, respectively ( Table 3 ). The correlation coefficient between the abdominal wall and uterine displacement was 0.73 ± 0.22 ( P  < 0.05). The abdominal wall displacements for type 1, type 2, and type 3 uteri were 2.75 ± 2.21, 3.51 ± 2.26, and 1.92 ± 0.67 mm, respectively, and uterine displacements of each type were 1.41 ± 0.96, 0.96 ± 0.62, and 0.46 ± 0.24 mm, respectively. Nine cases showed a weak correlation (range: 0.02–0.32) between abdominal wall and uterine displacements, without a significant difference (range: 0.08–0.89) in correlation strength among cases. Four (3 of type 1 and 1 of type 2) of these 9 cases did not show an improvement in the quality of RESP-T2WI, and 1 of the 4 cases displayed degradation of the visual assessment from moderate to poor. The remaining 5 cases (3 cases from type 1 and 2 cases from type 2) showed improvement in the quality of RESP-T2WI (2 cases, poor to excellent; 2 cases, moderate to excellent; 1 case, poor to moderate). In addition, 7 of 8 cases degraded image quality in RESP-T2WI showed strong correlation (range: 0.56–0.93) between abdominal wall and uterine displacements, with a significant difference (all P  < 0.01). Figure 7 shows the correlation between the maximum amplitude of uterine displacement and the quality of FSE-T2WIs. In the C-T2WI, the maximum amplitudes of uterine displacement for the poor, moderate, and excellent image quality groups were 2.35 ± 1.35, 1.61 ± 0.79, and 1.34 ± 0.61 mm, respectively ( P  < 0.001). Significantly different values were observed between the poor and moderate image quality groups ( P  = 0.022), as well as between the poor and excellent image quality groups ( P  < 0.001). No significant differences were observed between the moderate and excellent groups ( P  = 0.050). Similarly, in RESP-T2WI, the maximum amplitudes of uterine displacement for the poor, moderate, and excellent image quality groups were 2.24 ± 1.56, 2.18 ± 1.33, and 1.48 ± 0.74 mm, respectively ( P  = 0.084). No significant differences were observed between the groups ( P  = 0.084). Figure 8 presents the relationship between the maximum amplitude of uterine displacement and the uterine type. The maximum amplitudes of the uterine displacement for type 1, type 2, and type 3 were 2.21 ± 1.38, 1.61 ± 0.77, and 0.86 ± 0.44 mm, respectively ( P  < 0.001). Significantly different values were observed between type 1 and type 3 ( P  < 0.001) and type 2 and type 3 ( P  = 0.017). No significant difference was observed between type 1 and type 2.

This study was approved by the Research Ethics Committee of Fukushima Medical University (approval number: REC2024-131). All procedures involving human participants followed the ethical standards of the institutional and national research committees and the 1964 Declaration of Helsinki and its subsequent amendments or comparable ethical standards. The need for written informed consent was waived due to the retrospective nature of the study and data anonymization. This retrospective study involved patients who had female pelvic MRI examination between June 2023 and November 2024. Patients under 18 years of age, with a previous total hysterectomy, without records of T2WI with respiratory gating and cine images, and with inappropriate cine images were excluded ( Fig. 1 ). The clinical diagnoses are summarized in Table 1 . All MRI examinations were performed using a 3T-MRI scanner (Skyra; Siemens, Erlangen, Germany) equipped with an 18-channel phased-array body coil. This study utilized C-T2WI, RESP-T2WI, and half-Fourier-acquired single-shot turbo spin-echo (HASTE) static images of female pelvis acquired during routine pelvic MRI scans. In addition, HASTE-cine images, comprising 32 frames captured at 1 fps in the sagittal plane, were acquired before FSE-T2WI and focused on the uterus to determine uterine body displacement. These cine images were used to measure the uterine and abdominal wall displacements. No antispasmodic medications, dietary preparations, or urinary preparations were administered prior to imaging. The imaging parameters for the C-T2WI, RESP-T2WI, HASTE-cine, and HASTE-static images are listed in Table 2 . The uterine position was classified into 3 types based on FSE-T2WI, referring to the work of Tusboyama et al., as follows: type 1, uterus on the bladder; type 2, uterus not on the bladder and without retroflexion; and type 3, uterus contacting the vertebrae or a retroflexed uterus ( Fig. 2) . 7 Sagittal C-T2WI and RESP-T2WI were individually assessed by 2 experienced radiologists with 13 and 9 years of experience, respectively. To ensure consistency, the evaluation criteria were standardized before assessment, and HASTE-static images were used as reference images for identifying and grading motion-related artifacts. Visual assessment focuses on artifacts such as ghosting and motion blurring. Ghosting artifacts were rated on a 4-point scale according to the method outlined by Ishikawa et al. 12 : 0 = no apparent ghosts, 1 = visible but non-disturbing ghosts, 2 = visible and slightly disturbing ghosts, and 3 = ghosts interfering with the diagnostic interpretation. Motion blur was assessed on a 4-point scale: 0, sharply defined organ margins with clear internal structures, 1, minimal motion blur of margins and/or internal structures for diagnosis, 2, moderate motion blur that could affect diagnosis; and 3, marked motion blur (non-diagnostic). The overall image quality of each FSE-T2WI was categorized as excellent, moderate, or poor, based on the combined scores of ghosting artifacts and motion blur: excellent (0 points for both parameters), poor (3 points for either parameter), and moderate (1 or 2 points for either parameter) ( Fig. 3 ). Discrepancies between the evaluators were resolved by discussion to reach a consensus. Subsequently, the rates of concordant and discordant visual assessments between C-T2WI and RESP-T2WI were calculated. Cases exhibiting improved or deteriorated image quality on RESP-T2WI relative to C-T2WI were identified and further analyzed. In addition, the association between visual assessment results and uterine position type was investigated. The initial frame of the HASTE-cine image dataset was omitted from the analysis because of its non-steady-state longitudinal magnetization, leading to a distinct signal intensity compared with subsequent frames. The initial frame was referred to as a discarded frame. The first frame in the dataset, excluding the discarded frame, was defined as the reference image. The displacements of the uterus (head–foot direction) and abdominal wall (anteroposterior direction) were calculated pixel-by-pixel from the reference image in the data set onwards ( Fig. 4 ). Uterine displacement was determined using the “imregdemons” function in MATLAB 2022b (MathWorks, Natick, MA, USA). This function uses the Demons algorithm for non-rigid body registration to generate a displacement vector deforming one image based on density differences and gradients from the displacement of 2 images. 14 , 15 Subsequent images in the dataset were registered to the initial image, with ROIs for uterine and abdominal wall displacements defined on the first image ( Fig. 5 ). The ROI for the abdominal wall was defined over the height range from the pubic bone to the uterus. These ROIs were then overlaid onto the displacement maps, and the displacements of the uterus and abdominal wall were quantified over a 30 second period. Positive and negative peaks were identified from the time-series data of the uterine displacement. The amplitude of uterine displacement was determined as the difference between the adjacent positive and negative peaks. 12 Previous studies have established the primary direction of uterine body displacement as head–foot; 12 , 13 therefore, the maximum amplitude of uterine displacement was calculated in this direction. The maximum amplitude signifies the maximum range of uterine displacement. Peaks were detected using the “findpeaks” function in MATLAB 2022b. Subsequently, the maximum amplitude values, classified based on visual assessment of C-T2WI and RESP-T2WI, were plotted for each group. Unless stated otherwise, values are presented as mean ± standard deviation. First, to assess the reliability of visual assessments between the 2 radiologists, the inter-observer agreement (κ) in the visual assessment of the image quality of FSE-T2WI was calculated. The agreement rates were categorized as follows: moderate (0.41–0.60), substantial (0.61–0.80), and near-perfect agreement (≥ 0.81). 16 Before performing group or correlation analyses, data distribution and variance characteristics were examined to select appropriate statistical tests. To investigate whether the motion of the uterus corresponded with that of the abdominal wall, Spearman’s rank correlation coefficient was computed. Absolute values were used to assess the strength of the correlation. The associations between the amplitude of uterine displacement and both image quality and uterine type were evaluated using the Kruskal–Wallis test with Steel–Dwass post-hoc analysis. All statistical analyses were performed using R statistical analysis software for Windows version 4.3.1 (R: Language and environment for statistical computing; R Foundation for Statistical Computing, Vienna, Austria). 17 Significance was defined as P  < 0.05.

This study demonstrated the effectiveness of RESP-T2WI in maintaining image quality of the female pelvis, even with substantial uterine displacement. While C-T2WI showed a significant decrease in image quality with increasing uterine displacement amplitude, RESP-T2WI was unaffected by displacement magnitude. Consequently, in clinical settings, RESP-T2WI can be considered a viable approach for decreasing the number of non-diagnostic cases.

A strong correlation was observed between abdominal wall and uterine displacements, indicating that uterine displacement was affected by respiration. Consequently, respiratory gating was particularly effective in patients with type 1 and type 2 uteri. The results of this study showed that the quality of RESP-T2WI was superior to that of C-T2WI of the female pelvis, because the incidence of poor image quality group for RESP-T2WI was half that of C-T2WI, whereas the incidence of excellent image quality group for RESP-T2WI was double. The correlation coefficient between abdominal wall displacement (anteroposterior direction) and uterine displacement (head–foot direction) was high at 0.73 ± 0.22. This correlation suggested that uterine displacement in the head–foot direction is caused by respiration. During expiration, the abdominal wall is displaced from the anterior to the posterior direction, while the diaphragm is displaced from the foot to the head direction simultaneously. This series of displacements can be transmitted to the bladder and to type 1 and type 2 uteri via bowel and adipose tissue that are located superiorly (Supplementary Movies 1 and 2). Consequently, the uterus would be displaced in accordance with respiration. This finding supports the utility of RESP-T2WI. The maximum amplitude of uterine displacement was the highest in the poor image quality group of C-T2WI. Our results are consistent with our previous study. 12 Moreover, the maximum amplitudes of type 1 and 2 uteri were significantly higher than those of type 3 uteri. Therefore, in cases of type 1 and type 2 uteri, motion-related artifacts would be prone to occur. In type 1 cases, RESP-T2WI is particularly useful for reducing motion-related artifacts. This is because the type 1 uterus may be sensitive to abdominal wall displacement due to respiration because the uterus is close to the abdominal wall. In type 2 uteri, respiratory gating was also effective, despite the uterus being far away from the abdominal wall and not located on the bladder. This is because the displacement of the intestine or adipose tissue surrounding the uterus, in accordance with respiration, may be transmitted to the uterus. In contrast, respiratory gating may be less effective in type 3 uteri because the uterus is located farther from the abdominal wall and its displacement amplitude tends to be smaller, which inherently result in fewer motion-related artifacts. This finding is consistent with a previous report by Tsuboyama, et al., which showed that the uterine type in contact with the vertebrae was not a risk factor for non-diagnostic quality. 7 Some cases showed no noticeable improvement across all uterine types, suggesting that the effect of respiratory gating may vary among individuals. Cases with lower visual assessment in RESP-T2WI compared to C-T2WI may have been affected by factors such as sudden rectal displacement, gating mismatch, or unstable breathing during imaging. In these instances, unstable breathing can lead to inconsistent scan times and degraded image quality. Similarly, a lower correlation between abdominal wall and uterine displacements, potentially resulting from bowel displacement or bladder dilation, may diminish the effectiveness of respiratory gating. In such cases, abdominal wall movement no longer serves as a reliable surrogate for uterine motion, leading to the observed image degradation. Regarding the reliability of image quality evaluation, our inter-observer agreement κ was equal to or higher than those reported in prior studies. Previous studies have compared the quality of uterine FSE-T2WI with those of BLADE in Siemens MRI scanners, which is the same as periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER), 18 and HASTE, with inter-observer agreement κ values for artifacts of 0.65 and 0.51, respectively. 8 , 11 Therefore, the visual assessment in the present study can be considered justified. We anticipate that employing RESP-T2WI could decrease the need for re-imaging and enhance image quality, despite its longer scan time (approximately 5 mins 11s on average). In the future, the integration of deep learning reconstruction into RESP-T2WI will reduce imaging time. Other measures of motion-related artifacts include sequences that are resilient to body displacements, such as PROPELLER 18 and HASTE. Although the time extension is not as long as that of respiratory gating, there are some limitations, such as streak artifacts or a lower SNR. 8 11 This study had several limitations. First, uterine and abdominal wall displacements during HASTE-cine may vary from those during FSE-T2WI due to differences in bowel peristalsis and bladder dilation. Second, this study did not consider age or disease status, which could have potentially affected the results. Third, the imaging time for RESP-T2WI is generally longer than for C-T2WI and can be further prolonged depending on the patient’s respiratory cycle. However, a single respiratory-gated imaging would be preferable to re-imaging of C-T2WI in terms of total examination time and T2WI quality, as the latter does not reliably improve image quality. Despite these limitations, our results show that RESP-T2WI of the female pelvis can effectively reduce the number of non-diagnostic cases.

In the field of gynecology, imaging diagnosis is valuable for localizing diseases and evaluating treatment efficacy. 1 – 3 Fast spin echo-T2-weighted image (FSE-T2WI) is particularly important in imaging diagnosis because of its high contrast resolution. 4 , 5 However, the uterus, surrounded by organs, such as the abdominal wall, bladder, and bowel, undergoes displacement over time, leading to motion-related artifacts, 6 , 7 and clinical challenges may arise when motion-related artifacts complicate imaging diagnosis. Previous studies have reported the usefulness of sequences that are resilient to body displacements and those acquired in shorter imaging durations compared to physiological displacements in mitigating motion artifacts. 8 – 11 However, tradeoffs exist, such as the appearance of streak artifacts and a decrease in the SNR and contrast. 8 – 11 Therefore, motion artifact countermeasures using these sequences have not yet provided a fundamental solution to image quality degradation. Moreover, it is unclear to what extent these sequences can effectively counteract uterine displacement. Breathing-related uterine displacement is presumed to be the underlying mechanism of motion-related artifacts in female pelvic MRI. Previous studies have noted uterine displacement primarily in the head–foot direction rather than in the anteroposterior or left-right directions. 12 , 13 In abdominal and cardiac MRI, motion-compensation techniques such as respiratory or electrocardiogram gating have been proven effective in minimizing motion artifacts. We therefore hypothesized that aligning interventions with the uterine displacement mechanism would reduce the incidence of non-diagnostic cases due to motion-related artifacts on FSE-T2WI of the female pelvis. However, to date, there is a lack of literature regarding the efficacy of FSE-T2WI with respiratory gating (RESP-T2WI) for female pelvic MRI. Accordingly, this study aimed to determine whether RESP-T2WI enhances image quality compared to conventional FSE-T2WI (C-T2WI) by quantitatively evaluating uterine motion and its correlation with abdominal wall displacement across different uterine positions.