Study on the value of MRI in locating the internal OS of the cervix and influencing factors

We retrospective collected imaging data of stage I patients with EC (according to the 2009 FIGO staging standards 11 ) who underwent total hysterectomy from January 1 to December 31, 2021. Their clinical and pathological data were collected through the medical record system. Notably, histopathology was the core content of the 2023 FIGO staging standards. Although there are significant changes in stage I in the new staging, it is still mostly confined to the uterus. The staging is based on infiltration limited to the endometrium and involving half the myometrium 24 . According to the histopathological characteristics of the cases in this study, the proportion of cases with invasive histological types and lymphovascular space invasion (LVSI) is small. Therefore, this study still used the 2009 old FIGO staging. Inclusion criteria: (1) Plain and dynamic enhancement scanning of MRI was performed before hysterectomy, and imaging data was complete; (2) No contraindications to MRI examination. Exclusion criteria: (1) Treatments such as radiotherapy, chemotherapy, and uterine artery embolization had been performed after the MRI examination and before the hysterectomy; (2) A history of pelvic surgery, excluding cesarean section. This study was approved by the Ethics Committee of Beijing Obstetrics and Gynecology Hospital, Capital Medical University, who exempted informed consent for this retrospective analysis. Tis study was performed in accordance with the principles of the Declaration of Helsinki. Participants' names and other identifiers in images have been removed. We used the GE Discovery 750 3.0 T superconducting MR equipment, and signals were collected with an 8-channel phased-array coil. Patients were supine, and the scan range was from the iliac crest to the lower edge of the hip joint. Regular sequence scans included transverse T1WI: TR 528 ms, TE 10 ms; transverse and sagittal T2WI: TR 4000 ms, TE 87 ms. The regular scan was set to a thickness of 5 mm, interlayer distance of 1 mm, field of view (FOV) of 300 mm × 300 mm, and a matrix of 320 × 256. The diffusion-weighted imaging (DWI) had b values of 0 and 1000 s/mm 2 . The dynamic enhancement scan employed a 3D volume rapid scan using the Liver Acquisition with Volume Acceleration (LAVA-Flex) sequence, with sagittal plane: FOV 300 mm × 300 mm, matrix 296 × 224, layer thickness 3 mm, layer spacing 1 mm, TR 4.0 ms, TE set at the preset minimum, with a total acquisition of 6-phase enhancement images, the scanning duration was 2 min 15 s. Before the examination, a venous catheter was inserted to establish a venous access route. The contrast agent, gadodiamide (GE Pharmaceutical Co, Ltd, 287 mg/mL), was used, the injection rate was 2.0–2.5 mL/s, and the injection dose was 0.1 mmol/kg body weight. Bservation indicators on MRI images: (1) Display of cervical stroma on T2WI images; (2) Whether there are differences in the enhancement degree of the uterus and cervix on dynamic enhancement scanning images; (3) When the position of the uterus is anteverted or retroverted, whether a physiological curve is formed at the junction of the uterus and cervix on T2WI images. Methods for locating the internal os and measuring cervical length on MRI images (Fig.  3 ). Method 1: Measure the cervix length starting from the cross-section, showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images. Method 2: measure the cervical length beginning from the cross-section where the physiological curvature of the uterus is formed. Method 3: measure the length of cervical stroma on T2WI images. T2WI images were chosen for measurement to avoid the impact of the enhancement differences between the uterus and cervix. Moreover, the cervical length on MRI images is always measured in the median sagittal plane of the cervix, ending at the external os of the cervix. Figure 3 Sagittal T2WI, measure the cervical length from the start of the low signal of the cervical stroma ( a ). Sagittal T2WI, measures the cervix length starting from the cross-section where the physiological curvature of the uterus is formed ( b ). Sagittal dynamic enhancement scanning image, measure the cervix length beginning from the uterine and cervical enhancement junction ( c ). Sagittal T2WI, measure the cervical length from the start of the low signal of the cervical stroma ( a ). Sagittal T2WI, measures the cervix length starting from the cross-section where the physiological curvature of the uterus is formed ( b ). Sagittal dynamic enhancement scanning image, measure the cervix length beginning from the uterine and cervical enhancement junction ( c ). The cervical length measurement on MRI images was conducted jointly by two physicians with more than five years of experience in gynecological imaging diagnosis. In case of any opinion discrepancy, a unified decision is made after consultation between the two physicians. These patients’ surgery information was recorded, and their pathological results were reviewed. According to the pathologists, the length of the cervix was measured after the uterine specimens were cut in the median plane to reveal the long axis of the cervix. The starting point was the endometrium and cervical mucosa junction, and the endpoint was the external os of the cervix. The length of the cervix was recorded in detail in the pathology report. This study analyzed the impact of the position of the uterus, clinical symptoms, menopause, adenomyosis of the cervix, Nabothian cysts, history of cesarean section, and the depth of muscular infiltration on the location of the internal os on MRI. In this study, the position of the uterus was divided into two categories: those in an anteverted, retroverted, or middle position were classified as tilted. Those in anteverted retroflexed, anteverted anteflexed, retroverted anteflexed, or retroverted retroflexed positions were classified as flexed. That is, those forming a physiological curve at the junction of the uterus and cervix 25 , 26 . The main clinical symptom was abnormal uterine bleeding, with a bleeding time of three months as the dividing line. Furthermore, Nabothian cysts of the cervix were divided into two categories based on the location of the cyst: cysts in the internal os area and non-internal os area cysts (including no cysts). Our data were analyzed using SPSS statistical software (version 17.0; IBM SPSS). Based on whether the data conformed to normality, measurement data were presented as (mean ± standard deviation) or median(interquartile range, IQR). The cervical length of the uterine specimen was used as the gold standard. The Paired Sample T test or the Wilcoxon rank test was used to compare whether there were differences in the cervical length measured by the three methods on MRI images. Furthermore, logistic regression was used to analyze the factors affecting the location of the internal os on MRI. P < 0.05 was considered statistically significant. This study was approved by the Ethics Committee of Beijing Obstetrics and Gynecology Hospital, Capital Medical University (reference: 2023-KY-020-01). Informed consent was waived for this retrospective analysis by the Ethics Committee of Beijing Obstetrics and Gynecology Hospital, Capital Medical University.

In 2021, 346 patients with EC underwent hysterectomy. Among these, 231 were stage I EC. Forty-three patients underwent CT scans, and 13 were excluded due to incomplete MRI data. Finally, a total of 175 stage I EC cases were included in the analysis. Notably, the ages of the patients ranged from 26 to 79 years, with an average of 52 ± 3.6 years. The patients' clinical and pathological characteristics are shown in Table 1 . Table 1 Information of 175 stage I patients with endometrial cancer. Clinical and pathological characteristics Age (mean ± standard deviation) 52 ± 3.6 Variable N(%) Stage Ia* 138 (79.0) Stage Ib* 37 (21.0) Reproductive age 63 (36.0) Menopause 112 (64.0) BMI (kg/m 2 )   < 18.5 2 (1.1)  18.5 ≤–< 24.9 48 (27.4)  25 ≤–< 29.9 96 (54.9)   ≥ 30 29 (16.6) Tumor type* (%)  Endometrioid carcinoma 153 (87.4)  Serous carcinoma 15 (8.6)  Clear cell carcinoma 7 (4) Degree of tumor differentiation* (%)  Well-differentiated 114 (65.1)  Moderately differentiated 39 (22.3)  Poorly differentiated 22 (15.6) BMI body mass index. *Based on the 2009 FIGO staging standards 11 . Information of 175 stage I patients with endometrial cancer. BMI body mass index. *Based on the 2009 FIGO staging standards 11 . The cross-section showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images included 106 cases. The physiological curve that formed at the junction of the uterus and cervix on T2WI images involved 152 cases. One hundred forty-eight patients clearly showed cervical stroma on T2WI images. From MRI images, On MRI images, the cervical length was 2.87 cm (95% CI 2.80 cm–2.93 cm), 2.89 cm(95% CI 2.84 cm–2.94 cm), 2.90 cm (95% CI 2.84 cm–2.95 cm). The length of the cervix in the uterine specimen after hysterectomy was 2.88 cm (95% CI 2.81 cm–2.90 cm), 2.90 cm (95% CI 2.84 cm–2.95 cm), and 2.90 cm(95% CI 2.84 cm–2.95 cm), respectively. The box diagram shows the overall distribution of cervical length (Fig.  1 ). Our data indicated for multiple outliers which were mainly seen for the length of cervical stroma on T2WI images and the corresponding cervix length on gross uterine specimens. Figure 1 L1, L3, and L5 shows the length of the cervix starting from the cross-section showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images, and starting from the cross-section where the physiological curvature of the uterus is formed, and the length of cervical stroma on T2WI images. L2, L4, and L6 shows the corresponding cervix length on gross uterine specimens. L1, L3, and L5 shows the length of the cervix starting from the cross-section showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images, and starting from the cross-section where the physiological curvature of the uterus is formed, and the length of cervical stroma on T2WI images. L2, L4, and L6 shows the corresponding cervix length on gross uterine specimens. The Wilcoxon rank test results showed no significant difference in the cervical length measured by the three methods on MRI images and the cervical length measured on uterine specimens (P values all > 0.05), as shown in Table 2 . Table 2 Cervical length measurement results. Groups Cervical length median (IQR) 95% CI p value Uterine specimen 1 2.88 (0.50)cm 2.81 cm–2.90 cm 0.208 Method 1: 2.87 (0.50)cm 2.80 cm–2.93 cm Uterine specimen 2 2.90 (0.50)cm 2.84 cm–2.95 cm 0.571 Method 2: 2.89 (0.50)cm 2.84 cm–2.94 cm Uterine specimen 3 2.90 (0.46)cm 2.84 cm–2.95 cm 0.840 Method 3: 2.90 (0.420)cm 2.84 cm–2.95 cm IQR interquartile range. Method 1: measure the cervix length starting from the cross-section, showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images. Method 2: measure the cervical length beginning from the cross-section where the physiological curvature of the uterus is formed. Method 3: measure the length of cervical stroma on T2WI images. Cervical length measurement results. IQR interquartile range. Method 1: measure the cervix length starting from the cross-section, showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images. Method 2: measure the cervical length beginning from the cross-section where the physiological curvature of the uterus is formed. Method 3: measure the length of cervical stroma on T2WI images. In this study, there were 103 patients with abnormal uterine bleeding for less than 3 months and 72 patients with 3 months or longer. There were 138 patients with EC infiltrating the submucosal layer and 37 patients in the stage of deep infiltration. Additionally, 63 were of reproductive age, and 112 were postmenopausal. There were also 4 patients with no history of childbirth and 171 with a history of childbirth, including 43 patients with a history of cesarean section. MRI images showed 42 cases of uterine tilt and 134 cases of inflexion. Furthermore, 108 cases of Nabothian cysts were located in the internal os area, 39 cases not in the internal os area, and 28 cases with no cyst. There were also 18 patients with cervical adenomyosis. The logistic regression analysis showed that A history of cesarean section (p < 0.001), irregular vaginal bleeding for more than three months (p < 0.001), cervical adenomyosis (p = 0.043) was not conducive to locating the internal os of the cervix by MRI, while the menopausea was a a contributing factor (p = 0.001). A history of cesarean section is unfavorable for the measurement of cervical stroma and the formation of the physiological curvature of the uterus and cervix on T2WI images (Fig.  2 ). Abnormal uterine bleeding over three months and premenopause are unfavorable for forming the cross-section between the uterine body and cervix on dynamic enhanced T1WI images. Adenomyosis of the cervix is not favorable for observing the cross-section of the physiological curvature of the uterus and cervix on T2WI images (Fig.  2 ). Detailed results are shown in Table 3 . Figure 2 The sagittal T2WI image, the uterus is in an anteverted position, with the thickened endometrium protruding into the cervical canal. The patient had a history of cesarean section surgery and a scar formed in the muscle layer of the lower part of the anterior uterine wall. This is accompanied by deep endometriosis and adenomyosis of the posterior cervical wall, affecting the formation of the physiological curvature of the uterus, and the distinction between the lower part of the uterine body and the cervical stroma is unclear ( a ). The sagittal dynamic enhanced scan image, the enhancement degree of the uterine body and the cervix is different, the uterine body is stronger than the cervix, and the cross-section formed by the difference in enhancement degree can be used to locate the internal os of the cervix ( b ). Table 3 Factors influencing the localization of the internal os of the cervix on MRI. Factors Method 1 Method 2 Method 3 β OR 95% CI P value β OR 95% CI P value β OR 95% CI P value Depth of Myometrial Infiltration 0–0.13 10.88 00.25–3.07 00.837 − 1.27 00.28 00.05–1.49 00.136 11.46 44.292 00.31–58.56 10.275 Vaginal bleeding for ≥ 3 months –3.13 00.04 00.02–0.12  < 0.001 − 0.59 00.56 00.14–2.24 00.409 − 1.07 10.34 00.07–1.69 00.189 Menopause 1.38 3.97 1.36–11.55 0.011 00.003 11.00 00.24–4.16 00.996 0.084 1.088 0.190–6.225 0.925 Nabothian Cyst 0.092 1.096 0.387–3.104 0.863 0.34 1.410 0.39–5.07 0.598 -0.51 0.60 0.12–3.13 0.546 Cesarean Section − 0.49 0.61 0.19–2.04 0.613 -3.44 0.03 0.01–0.13  < 0.001 -3.05 0.05 0.01–0.22  < 0.001 Uterine Position 0.14 1.15 0.35–3.71 0.821 -1.16 0.32 0.05–1.88 0.204 -1.61 0.20 0.02–2.43 0.206 Adenomyosis 0.17 1.19 0.21–6.66 0.846 -1.48 0.23 0.05–1.14 0.071 -2.16 0.12 0.01–0.93 0.043 OR odds ratio. Method 1:measure the cervix length starting from the cross-section, showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images. Method 2: measure the cervical length beginning from the cross-section where the physiological curvature of the uterus is formed. Method 3: measure the length of cervical stroma on T2WI images. The sagittal T2WI image, the uterus is in an anteverted position, with the thickened endometrium protruding into the cervical canal. The patient had a history of cesarean section surgery and a scar formed in the muscle layer of the lower part of the anterior uterine wall. This is accompanied by deep endometriosis and adenomyosis of the posterior cervical wall, affecting the formation of the physiological curvature of the uterus, and the distinction between the lower part of the uterine body and the cervical stroma is unclear ( a ). The sagittal dynamic enhanced scan image, the enhancement degree of the uterine body and the cervix is different, the uterine body is stronger than the cervix, and the cross-section formed by the difference in enhancement degree can be used to locate the internal os of the cervix ( b ). Factors influencing the localization of the internal os of the cervix on MRI. OR odds ratio. Method 1:measure the cervix length starting from the cross-section, showing enhancement differences between the uterus and cervix on dynamic contrast-enhanced scanning images. Method 2: measure the cervical length beginning from the cross-section where the physiological curvature of the uterus is formed. Method 3: measure the length of cervical stroma on T2WI images.

We confirmed the location of the internal os of the cervix on MRI images by compared with surgical specimens. The three methods of locating the internal os of the cervix were equally effective and accurate. Our results about the internal cervical os on MRI images are the same as Lakhman et al 6 . We identifed that there are some influencing factors not conducive to MRI for observing of the internal os. The literature has pointed out that the internal cervical os is located at the entry point of the uterine blood vessels 6 , 9 , 12 . At this level, the uterine artery is divided into ascending branches supplying the uterine body and descending branches supplying the cervix. The ascending branch is larger than the descending branch 13 , 14 . Early literature indicates that blood supply to the cervix is as fast as the blood flow entering the uterine myometrium 15 . Therefore, the enhancement degree of the uterine myometrium is higher than that of the cervical myometrium. Then, the dividing line formed by this difference in enhancement degree can be used to locate the internal cervical os. Based on studying the method of locating the internal cervical os using MRI, there are influencing factors can potentially interfere with the results, including the position of the uterus, clinical symptoms, menopause, cervical adenomyosis, Nabothian cysts, history of cesarean section, and stages of EC. A cesarean section disrupts the muscle layer of the lower uterine segment (LUS), forming a fibrous scar which is thin and shows a low signal on T2WI images 16 , 17 , making it unfavorable for displaying the cervical stroma and the physiological curvature of the uterine body and cervix. Cervical adenomyosis shows a low signal intensity on T2WI images with varying degrees of enhancement following contrast injection 18 . Moreover, deep endometrial ectopia can involve the serosal surface of the uterus and extend to the cervix, causing adhesion between the uterine body and the cervix 19 , hindering the display of physiological curvature of the uterine body and cervix on MRI images. This study shows that if the bleeding duration exceeds 3 months or if the woman is not menopausal, the enhancement division between the uterine body and the cervix on MRI dynamic enhanced scan images is more challenging to display. It may have something to do with blood circulation. Clinically, six months is used as the time interval for acute or chronic vaginal bleeding 20 . However, looking at the patient characteristics of this group, three months was used as the time division for uterine bleeding. A study on non-pregnancy-related uterine bleeding pointed out that abnormal bleeding can cause changes in blood circulation and lead to hormonal instability within the body 21 . Menopause is a time of ovarian function decline. During this time, hormone changes are significant, and vasomotor symptoms may appear, leading to stable hormone levels and blood circulation following menopause 22 . Nabothian cysts of the cervix mainly occur within the cervical mucosa, but sometimes they can extend deep into the cervical stroma 23 . Thus, they do not impact the localization of the internal cervical os on MRI. The depth to which EC infiltrates the myometrium of the uterine body does not affect the anatomy of the cervix. Hence, it does not influence the MRI location of the internal cervical os. However, it should be noted that differentiation from invasive cervical stroma is necessary when the cancer lesion is located in the lower uterine segment 5 . When applying the results of this study to the clinic, the study's limitations need to be considered. Firstly, the gross specimens in this study are uterine specimens received by the pathology department after the hysterectomy. There might be certain errors in the length of the uterus and cervix compared to that in the human body, which cannot be avoided entirely. Secondly, the subjects included in this study are women of reproductive age and postmenopausal women with EC. The study did not categorize the women of reproductive age according to their menstrual cycle, so it is unknown whether different subgroups within the reproductive age group affect the MRI location of the internal cervical os. Moreover, some extreme values in our data were not deleted and included in the analysis. The cervix length measured by MRI was inconsistent with the gross specimen, possibly related to uterine extreme anterior flexion. In conclusion, all three measurement methods on T2WI and DCE-MRI can accurately locate the internal os of the cervix. When there are interference factors, the three methods complement each other.

Fertility preservation is an essential component of clinical treatment plans and a critical determinant of patient quality of life after treatment 1 . For patients with cervical cancer (CC) and patients with endometrial cancer (EC) desiring fertility preservation, accurate locating of the cervical internal os through imaging evaluation is required to determine the distance of the tumor from the cervical internal os. For women with CC who wish to have children, MRI needs to indicate the pattern of tumor growth and three-dimensional size (whether it is < 2.0 cm) and the distance from the tumor to the internal cervical os (< 1.0 cm) 2 , 3 . For patients with EC, locating the internal cervical os is critical to distinguishing the involvement of the lower uterine segment or cervix 4 , 5 . According to the literature, two methods exist for locating the internal cervical os on MRI images 6 – 10 . Firstly, the literature pointed out that the internal cervical os is located at the entry point of the uterine blood vessels on MRI images 6 , 7 . However the location was marked inconsistent. Secondly, the internal cervical os is located at the transition from the low signal intensity of the cervical stroma on T2WI images to the position of medium signal intensity of the uterine myometrium 7 , 9 , 10 . However, in actual clinical practice, when some patients have chronic inflammation of the cervix, nabothian cysts, or distorted boundaries between the uterus and cervix, it becomes difficult to determine the boundary between the cervical stroma and the uterine body. In such cases, other methods must be explored to locate the internal os of the cervix. Therefore, we retrospectively analyzed the data of stage I patients with EC who came to the hospital for surgery, recorded the cervical length of the uterus specimens after hysterectomy and used the removed uterus specimens for comparison to explore the new method and accuracy of MRI image localization of the cervical internal os. Furthermore, the factors affecting the localization of the cervical internal os were analyzed.