A pelvic floor muscle volume assessment and prediction model and its construction method

Abstract

The present invention discloses a pelvic floor muscle volume assessment and prediction model and a construction method thereof, comprising the following steps: S1. Collect MRI examination images of the pelvic floor muscles, and extract relevant parameter features in the images; S2. Perform artificial reconstruction and segmentation on the images according to the relevant parameter features, distinguish and mark the positions of the coccygeus muscle, iliococcygeus muscle, remaining levator ani muscle and levator ani muscle in the images, and calculate the muscle volume at different positions; S3. Determine the cut-off value for two groups of data with high and low muscle volumes at different positions, and describe and test the POP-Q indexes of the two groups to obtain differential indexes; S4. Perform Logit regression analysis according to the differential indexes to obtain the prediction model of the muscle volume at different positions. It can be used to predict the high and low of the pelvic floor muscle volume, so as to assist doctors in making quick decisions in clinical applications, and at the same time improve the accuracy of the diagnosis and treatment of pelvic floor diseases.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical diagnosis, and specifically, it is a pelvic floor muscle volume assessment and prediction model and a method for constructing the same. Background Art

[0002] Pelvic organ prolapse (POP) is the descent of pelvic organs caused by abnormal pelvic floor tissues, resulting in abnormal organ position and dysfunction. The main symptoms are the protrusion of a mass from the vaginal orifice, which may be accompanied by urinary, defecation, and sexual dysfunction. It is a common disease in adult women. Treatment methods include lifestyle intervention, physical therapy, pessary treatment, and surgical treatment, etc. There are many surgical treatment methods. Traditional surgeries mainly use autologous tissues for repair. In some patients, the supporting force of autologous tissues is weak, and the postoperative recurrence rate is relatively high. Although the success rate of pelvic floor reconstruction using mesh is relatively high, the incidence of mesh-related complications is high. Therefore, for patients who need surgical treatment, accurate preoperative assessment of the severity of POP and pelvic floor structural defects is a prerequisite for formulating an appropriate surgical plan.

[0003] Currently, clinicians mainly evaluate the patient's condition through POP-Q scoring, pelvic floor ultrasound, etc., and to a certain extent rely on experience to formulate surgical plans, resulting in relatively high postoperative recurrence rates and reoperation rates. Existing research shows that levator ani muscle injury is closely related to pelvic organ prolapse. Patients with low muscle volume may be more inclined to use mesh for surgery. Therefore, evaluating and predicting the pelvic floor muscle volume is of great significance for surgical decisions and can reflect the situation of the pelvic floor support structure to a certain extent.

[0004] Existing muscle volume assessment relies on magnetic resonance imaging (MRI). Due to economic and equipment limitations, it is difficult to promote widely in clinics. MRI has high soft tissue resolution. Based on MRI-based three-dimensional reconstruction of the pelvic floor, it can directly display the fine anatomical structure of the pelvis and the relationship between tissue organs, and can also perform qualitative and quantitative analysis of lesions. On this basis, preoperative plan design and surgical simulation can be carried out, which is more conducive to precise treatment and improving the surgical success rate. Currently, it has been applied in the diagnosis of skeletal muscle diseases. However, the pelvic floor structure is more complex, and the scanning gray range of different tissue organs cannot be automatically distinguished, and image marking is mainly carried out manually. Manual recognition is time-consuming, and coupled with the high cost of MRI examination, it is very difficult to promote and apply MRI-based three-dimensional reconstruction of the pelvic floor muscle in clinics.

[0005] Chinese Patent CN115530881A discloses a holistic pelvic floor function assessment method and device for multimodal data fusion. The assessment method is an integrated pelvic floor function detection and assessment process that integrates optical, ultrasonic, and pressure detections on a multi-module fusion vaginal and rectal probe. In this process, although the video images detected optically, the two-dimensional images detected ultrasonically, and the curve graphs detected by pressure can be fused to obtain a three-dimensional stereoscopic image to display the overall state of pelvic floor tissues, muscles, bones, nerves, etc. comprehensively, it is still unable to be used to evaluate the relevant numerical values of the pelvic floor muscle volume. Therefore, the present invention specifically provides an assessment and prediction model for the pelvic floor muscle volume for clinical MRI detection, which can be used to predict the level of the pelvic floor muscle volume, so as to assist doctors in making quick decisions in clinical applications and improve the accuracy of the diagnosis and treatment of pelvic floor diseases at the same time. Summary of the Invention

[0006] The object of the present invention is to provide a method for constructing an assessment and prediction model for the pelvic floor muscle volume. This construction method is based on the existing magnetic resonance imaging for evaluating muscle volume, specifically divides the pelvic floor muscles into two categories of high and low muscle volume according to the median, uses the Logit model to predict the volume levels of several muscles, and then through variable screening, predicts according to the POP-Q score, thereby obtaining the high and low levels of the muscle volume at different positions of the pelvic floor muscles, which can be used to guide the design of clinical individualized surgical plans. For this reason, the present invention also provides an assessment and prediction model for the pelvic floor muscle volume constructed by using the above construction method.

[0007] The present invention is realized through the following technical solutions: A method for constructing an assessment and prediction model for the pelvic floor muscle volume, comprising the following steps:

[0008] S1. Collect MRI examination images of the pelvic floor muscles and extract relevant parameter features in the images;

[0009] S2. Manually reconstruct and segment the images according to the relevant parameter features, distinguish and mark the positions of the coccygeus muscle, iliococcygeus muscle, remaining levator ani muscle, and levator ani muscle in the images, and calculate the muscle volumes at different positions;

[0010] S3. Determine the cut-off value for two groups of data with high and low muscle volumes at different positions, describe and test the POP-Q indicators of the two groups, and obtain the difference indicators;

[0011] S4. Perform Logit regression analysis according to the difference indicators, and obtain the prediction models for the muscle volumes at different positions.

[0012] In the step of S1, when collecting the MRI examination images of the pelvic floor muscles, the static sequence sagittal plane, coronal plane and axial sequence cover the entire pelvic range, and the image range of the dynamic scan and defecography is extended to the proximal thigh.

[0013] In the step of S1, relevant parameter features of the organ region and ROI region in the image are extracted, and the relevant parameter features include color, texture and shape.

[0014] In the step of S2, manual reconstruction and segmentation is a process of manually using a tool integrated with an image segmentation algorithm to segment and label different positions of the pelvic diaphragm, colorectum, bladder, uterus and vagina in the extracted image.

[0015] In the step of S2, the remaining levator ani muscle includes the pubovaginalis muscle, pubococcygeus muscle and puborectalis muscle; the levator ani muscle includes the iliococcygeus muscle and the remaining levator ani muscle.

[0016] In the step of S2, the integrated Export / import models and labelmaps algorithm is adopted, and the models information integration module is used to export the Mesh model and label map of each tissue structure segmented by manual reconstruction respectively. Through the information volume measurement algorithm tool, the volume of the target structure is extracted. Multiplying the number of voxels contained in the generated label map by the voxel size of the image can obtain the volume of each tissue structure.

[0017] In the step of S3, the description of the POP-Q indexes of the two groups is to predict the high and low of the muscle volume based on the POP-Q score using Logit regression analysis; the Wilcoxon rank sum test method is used to verify the POP-Q indexes during the test.

[0018] In the step of S3, the difference indexes at least include age, C, Ba, gh, pb, tvl, Bp and D.

[0019] In the step of S4, the prediction models of the muscle volume at different positions at least include the following formulas:

[0020] (1) Iliococcygeus muscle high / low = logit(-5.04379 - 0.13209*Ba - 0.03222C + 0.1995*gh + 0.1231*pb + 0.5807*tvl - 0.20253*Bp + 0.05719*D);

[0021] (2) Coccygeus muscle height = logit(-4.197395 - 0.024958 * Elderly - 0.345093 * Aa + 0.086413 * Ba++ + 0.006805C + 0.465753 * gh - 0.066995 * pb + 0.287507 * tvl + 0.014876 * Ap - 0.121509 * Bp + 0.005364 * D);

[0022] (3) Remaining levator ani muscle height = logit(-3.37790 - 0.76268 * Age - 0.29136 * Aa - 0.01824 * Ba - 0.02412 * C + 0.42796 * gh + 0.11895 * pb + 0.27907 * tvl + 0.31562 * Ap - 0.19732 * Bp + 0.05606 * D);

[0023] (4) Levator ani muscle height = logit(-1.24849 - 0.05855 * Aa - 0.01493 * Ba - 0.00468 * C + 0.09061 * gh + 0.04563 * pb + 0.17617 * tvl - 0.04719 * Bp + 0.01337 * D).

[0024] A pelvic floor muscle volume assessment and prediction model constructed by the above construction method.

[0025] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0026] (1) The present invention obtains a prediction model based on the three-dimensional reconstruction data of the pelvic floor by MRI, utilizes the advantage of high soft tissue resolution of the former and makes up for the disadvantages of time-consuming and expensive; saves the cost of measuring the pelvic floor muscle volume of patients by technologies such as MRI, and avoids the complexity of muscle volume measurement, and can utilize existing indicators to guide clinical decisions as much as possible.

[0027] (2) The prediction model involved in the present invention only needs to input the POP-Q score to obtain the height of the muscle volume, and the POP-Q score is widely used and easy to operate, making the promotion of this model feasible.

[0028] (3) The present invention uses the Logit model to consider the clinical interpretability of muscle volume prediction, and makes the decision-making based on the POP-Q score transparent. This prediction model can timely guide clinical decisions and quickly make clinical decisions based on muscle volume. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic flow diagram of MRI examination.

[0030] Figure 2 Sagittal and coronal plane images of the T2 fast spin echo sequence for MRI examination of Patient 1.

[0031] Figure 3 Axial plane images of the T1 (left) and T2 (right) fast spin echo sequences for MRI examination of Patient 1.

[0032] Figure 4 Axial plane image of the three-dimensional high-resolution T2-weighted imaging for MRI examination of Patient 1.

[0033] Figure 5 MRI diagnosis report of Patient 1.

[0034] Figure 6 Sagittal and coronal plane images of the T2 fast spin echo sequence for MRI examination of Patient 2.

[0035] Figure 7 Axial plane images of the T1 (left) and T2 (right) fast spin echo sequences for MRI examination of Patient 2.

[0036] Figure 8 Axial plane image of the three-dimensional high-resolution T2-weighted imaging for MRI examination of Patient 2.

[0037] Figure 9 MRI diagnosis report of Patient 2. Detailed implementation manners

[0038] The present invention will be further described in detail below in conjunction with embodiments, but the implementation manners of the present invention are not limited thereto.

[0039] Embodiment 1: Establishing an evaluation and prediction model for the muscle volume of the pelvic floor muscles

[0040] This embodiment relates to the construction process of an evaluation and prediction model for the muscle volume of the pelvic floor muscles. The specific steps can be summarized as follows:

[0041] Step 1, Acquisition of three-dimensional pelvic floor images

[0042] Pelvic floor muscle MRI examination images were collected for female patients with pelvic organ prolapse. 300 female patients diagnosed with pelvic organ prolapse were included. Inclusion criteria: POP-Q classification degree II or above, full preoperative informed consent of surgical risks, and patients who understood and chose surgical treatment.

[0043] MRI examinations were performed on the included patients using a 1.5T magnetic resonance device to obtain images. An abdominal phased array coil was used. During the examination, the static sequence sagittal, coronal, and axial sequences covered the entire pelvic range, and the image range of dynamic scanning and defecography was extended to the proximal thigh. After acquisition, all images were stored in the PACS system in DICOM format and saved in the corresponding image database.

[0044] Step 2, Image extraction

[0045] Specifically extract the features such as color, texture, and shape of the organ region and the region of interest (ROI region) in the image.

[0046] Step 3, Manual reconstruction and segmentation

[0047] Perform manual reconstruction and segmentation on the image after feature extraction. Use the integrated ND morphological contour interpolation and grow - cut algorithms and apply segmentation algorithm tools such as Threshold, Region grow, Flood filling, and Margin extension. By segmenting and annotating 2 - 3 groups of pelvic diaphragms (including the coccygeus muscle, levator ani muscle, etc.), the colorectum, bladder, uterus, and vagina and other pelvic structures in the MRI image, thereby distinguishing and marking the positions of the coccygeus muscle, iliococcygeus muscle, remaining levator ani muscle, and levator ani muscle in the image.

[0048] The levator ani muscle described in this embodiment includes the iliococcygeus muscle, pubovaginalis muscle, pubococcygeus muscle, and puborectalis muscle. Since the pubovaginalis muscle, pubococcygeus muscle, and puborectalis muscle are relatively small under MRI and their boundaries are not easy to distinguish, during the construction process, the pubovaginalis muscle, pubococcygeus muscle, and puborectalis muscle are combined into the remaining levator ani muscle. The levator ani muscle is divided into two parts: the iliococcygeus muscle and the remaining levator ani muscle, which is convenient for the identification of the corresponding muscles and the calculation of the volume under MRI.

[0049] Step 4, Predict the muscle volume

[0050] Adopt the integrated Export / import models and labelmaps algorithm, and use the models information integration module to separately export the Mesh model (for visualization) and LabelMap (for volume calculation) of each organizational structure segmented in the Segment module. Through the information volume measurement algorithm tool, extract the volume of the target structure, and then multiply the number of voxels contained in the previously generated Label Map by the voxel size of the image (Voxel Size) to obtain the volume of each organizational structure. Therefore, the volumes of the bilateral muscles of the coccygeus muscle, iliococcygeus muscle, remaining levator ani muscle, and levator ani muscle can be calculated respectively.

[0051] Step 5, POP - Q score prediction and verification

[0052] Based on the predicted muscle volume of the coccygeus muscle, iliococcygeus muscle, and levator ani muscle, a binary classification based on the muscle volume is used to determine whether customized surgery is required. Considering that there is currently no gold standard for muscle volume, a decision model based on the median is adopted for processing, that is: the median (median: 5197, 4843.1, 16422, 21389) is used as the cut-off to divide the muscle volume into two groups: high and low. The POP-Q indicators of the two groups are described and a rank-sum test is performed. The statistical software used is R 3.4.3, and the results are shown in Table 1 below.

[0053] Specifically, the muscle is divided into two levels of high and low volume as the prediction object, that is, the dependent variable, to explore the possibility of establishing a prediction model for muscle volume level. First, based on the POP-Q score, Logit regression analysis is used to predict the high and low muscle volume. Then, a statistical test is performed on the parameters of each variable in the model. The obtained statistical test p-value represents the probability that the parameter is 0. Finally, the rank-sum test method is used to explore whether there are differences in the POP-Q indicators of high and low muscle volume.

[0054] Table 1 POP-Q characteristics of each muscle group

[0055]

[0056]

[0057] Step 6, Establishment of the prediction model

[0058] Logit regression analysis is adopted, using whether the patient is elderly (age > 60) as the correction variable. Variable screening is performed using stepwise regression screening, and it is required that the model retains Ba and C. The statistical software used is R 3.4.3, and the results are shown in Table 2.

[0059] It can be found that in the prediction of high and low muscle volume, the finally selected variables, in addition to Ba and C, all retain gh, pb, tvl, Bp, and D. Therefore, among them, gh, tvl, and Bp have statistical significance except for Bp in the prediction of the coccygeus muscle.

[0060] Table 2 Results of the Logit stepwise regression model for muscle volume

[0061]

[0062]

[0063] The finally constructed prediction model formula (where 1 is high and 0 is low) is as follows:

[0064] (1) iliococcygeus muscle height = logit(-5.04379 -0.13209*Ba-0.03222C+0.1995*gh+0.1231*pb+0.5807*tvl-0.20253*Bp+0.05719*D);

[0065] (2) coccygeal muscle height = logit(-4.197395-0.024958*old-0.345093*Aa+0.086413*Ba++0.006805C+0.465753*gh-0.066995*pb+0.287507*tvl+0.014876*Ap-0.121509*Bp+0.005364*D);

[0066] (3) levator ani height = logit(-3.37790 -0.76268*age-0.29136*Aa-0.01824*Ba-0.02412*C+0.42796*gh+0.11895*pb+0.27907*tvl+0.31562*Ap-0.19732*Bp+0.05606*D);

[0067] (4) Anal levator muscle height = logit(-1.24849 -0.05855*Aa-0.01493*Ba-0.00468*C+0.09061*gh+0.04563*pb+0.17617*tvl-0.04719*Bp+0.01337*D).

[0068] It should be noted that, in the present invention, Aa refers to the midline of the anterior vaginal wall 3 cm away from the hymen edge; Ba refers to the lowest point of the anterior vaginal wall between the apex of the vagina and point Aa; C refers to the farthest end of the apex of the vagina after cervical or hysterectomy; gh refers to the distance between the midpoint of the external urethral orifice and the hymen edge of the midline of the posterior wall; pb refers to the distance from the hymen edge of the midline of the posterior wall to the midpoint of the anus; tvl refers to the maximum depth of the vagina measured when the prolapse is fully reduced to avoid increased pressure or stretching; Ap refers to the midline of the posterior vaginal wall 3 cm away from the hymen edge; Bp refers to the lowest point of the posterior vaginal wall between the apex of the vagina and point Ap; D refers to the posterior fornix (in patients without hysterectomy); CD refers to the difference between point C and point D.

[0069] In current clinical diagnosis and treatment for POP patients, clinicians mainly evaluate the patients' conditions through POP-Q scores, pelvic floor ultrasound, etc. The choice of surgical method for patients and whether to use a mesh are all subjectively determined by doctors. Due to different clinical experiences of each doctor, there are significant differences in the selection of surgical plans, with a high postoperative recurrence rate and a high rate of secondary surgeries. Therefore, by using the pelvic floor muscle volume assessment and prediction model constructed in Example 1, the estimation of the pelvic floor muscle volume of patients can be achieved. Based on this data, clinicians can quickly make clinical decisions, improving the accuracy of individualized surgical plan design.

[0070] The following are two groups of clinical application cases:

[0071] (1) Patient 1: 62 years old.

[0072] Preoperative diagnosis: Anterior vaginal wall prolapse degree III, uterine prolapse degree III, posterior vaginal wall prolapse degree II, urge incontinence (mild). Preoperative POP-Q score: Aa 2 cm, Ba 5 cm, C 2.5 cm, gh 6 cm, pb 2.5 cm, tvl 7.5 cm, Ap 0 cm, Bp 0 cm, D -2 cm.

[0073] (1) Directly use a 1.5T MRI magnetic resonance device to examine the patient. Refer to Figure 1 the process shown, and the specific examination steps are as follows:

[0074] 1. Preparation before examination:

[0075] Patient training: 1) Before the examination, fully inform the patient of the purpose of the examination and the examination process of pelvic floor MRI, inform the patient of the action commands, force application time, force application frequency, etc. during the examination, and conduct cooperation training; 2) Conduct sufficient pelvic floor muscle movement training for the patient, including Valsalva maneuver, anal contraction maneuver, and defecation maneuver.

[0076] Patient preparation: 1) For those with fecal and urinary incontinence, prepare adult diapers or disposable diapers in advance and pad them under the buttocks; 2) Instruct the patient to empty the bladder 2 hours before the examination, arrive at the examination room 30 minutes in advance for waiting, and drink 300 - 500 ml of water during this period to ensure moderate bladder distension during the examination.

[0077] 2. Examination process:

[0078] Using a 1.5T magnetic resonance device and a phased array body coil, the patient lies supine, with a wedge-shaped sponge pad or pillow placed under the knees to keep the knee joint in a flexed state; the coil is placed centered on the pelvis and covers all pelvic floor structures and organ prolapse. Correctly input the patient's name, age, registration number and other information, retrieve the pelvic floor function scan sequence package, and start the scanning program. The examination sequences include: 1. Static sequences: high-resolution fast spin echo sequence T2-weighted imaging, T1-weighted imaging, and three-dimensional high-resolution T2-weighted imaging, without fat suppression; T2-weighted imaging is routinely scanned in the standard axial, coronal, and sagittal planes of the pelvis; T1-weighted imaging is an optional sequence and is generally only scanned in the transverse plane; the patient should maintain calm breathing during scanning; 2. Dynamic sequences: single-shot fast spin echo sequence (SSFSE) and balanced steady-state free precession fast imaging sequences (such as True-FISP, FIESTA, B-FFE); the mid-sagittal plane is routinely scanned, and each dynamic sequence should not exceed 20 s; the patient is imaged during the Valsalva maneuver and the anal contraction maneuver.

[0079] After the scan is completed and the patient's safety is evaluated, the patient leaves the examination room. The MRI images of Patient 1 (see Figures 2 to 4 ) are archived in a timely manner and uploaded to the PACS in a timely manner. The diagnostic doctor issues a standardized report on female pelvic floor MRI (see Figure 5 ). Finally, based on the MRI images, the standardized MRI report, and the preoperative POP-Q score, the surgical plan required by the patient is obtained.

[0080] (2) Using the pelvic floor muscle volume assessment and prediction model constructed by the present invention, the prediction model formulas for the four muscle heights (see Example 1), combined with the preoperative POP-Q score, can measure that the volumes of the coccygeus muscle, iliococcygeus muscle, and levator ani muscle are all low. The MRI measurement of muscle volume is consistent with the prediction of muscle volume by the prediction model. Therefore, based on this measurement result, a reconstruction surgical method mainly using a mesh is selected.

[0081] This patient chose to undergo total pelvic floor reconstruction (using a six-wing mesh). The operation lasted for 125 minutes, with 50 ml of intraoperative bleeding, and no complications during the perioperative period. Follow-up visits were conducted at 1, 3, 6, and 12 months after the operation, and there were no complications and no recurrence.

[0082] (2) Patient 2, 59 years old.

[0083] Preoperative diagnosis: uterine prolapse degree III, anterior vaginal wall prolapse degree II, posterior vaginal wall prolapse degree II, cervical elongation, and old perineal laceration degree II. Preoperative POP-Q score: Aa - 1 cm, Ba 0 cm, C 1.5 cm, gh 5.5 cm, pb 2 cm, tvl 8 cm, Ap 0 cm, Bp 0 cm, D - 4 cm.

[0084] (1) Directly use an MRI 1.5T magnetic resonance device to examine the patient and operate according to the aforementioned process.

[0085] After the examination, archive the MRI images of patient 2 (see Figures 6 to 8 ), upload them to the PACS in a timely manner, and have a diagnostic doctor issue a standardized report on female pelvic floor MRI (see Figure 9 ). Finally, based on the MRI images, the standardized MRI report, and the preoperative POP-Q score, obtain the surgical plan required for the patient.

[0086] (2) Adopt the pelvic floor muscle volume assessment and prediction model constructed by the present invention, the prediction model formulas for the height and low of the four muscles (see Example 1), and combine with the preoperative POP-Q score to obtain that the volumes of the coccygeus muscle, iliococcygeus muscle, and levator ani muscle are all high, which is consistent with the MRI measurement results. Therefore, select the reconstruction surgical method using autologous tissue.

[0087] This patient chose to undergo single-port laparoscopic total hysterectomy and high uterosacral ligament suspension. The operation duration was 215 minutes, the intraoperative blood loss was 20 ml, and there were no complications during the perioperative period. Follow-up visits were conducted at 1, 3, 6, and 12 months after the operation, and there were no complications and no recurrence.

[0088] After the prediction model is constructed, clinical patients can no longer undergo MRI examinations and pelvic floor three-dimensional reconstructions, and can predict the height and low of the pelvic floor muscle volume through the POP-Q score, thereby assisting in the design and implementation of individualized surgical plans.

[0089] In summary, compared with the existing MRI examination for evaluating the condition of POP patients, the pelvic floor muscle volume assessment and prediction model constructed by the present invention can obtain the height and low of the muscle volume based on the POP-Q score, which can not only improve the accuracy of individualized surgical plan design, but also save the examination process and improve the efficiency of patient diagnosis and treatment evaluation.

[0090] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. Any simple modification or equivalent change made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for constructing a pelvic floor muscle volume assessment prediction model, characterized in that: The following steps are involved: S1. Collect MRI examination images of pelvic floor muscles and extract relevant parameter features in the images; S2. manually reconstruct and segment the image according to relevant parameter features, distinguish and mark the positions of the coccygeal muscle, iliococcygeus muscle, levator ani muscle and levator ani muscle in the image, and calculate the muscle volume at different positions; S3. Determine the cut-off value for the two groups of data on muscle volume at different locations, and describe and test the POP-Q index of the two groups to obtain the difference index; S4. Logit regression analysis is performed based on the difference index to obtain a prediction model for muscle volume at different locations.

2. The construction method according to claim 1, characterized in that: In step S1, when acquiring MRI examination images of the pelvic floor muscles, the static sequence sagittal, coronal and axial sequences include the entire pelvic range, and the image range of dynamic scanning and defecography is expanded to the proximal thigh.

3. The construction method according to claim 1, characterized in that: In the step S1, relevant parameter features of the organ region and the ROI region in the image are extracted, and the relevant parameter features include color, texture, and shape.

4. The construction method according to claim 1, characterized in that: In the step S2, manual reconstruction and segmentation is a process of manually using a tool integrated with an image segmentation algorithm to segment and mark the pelvic diaphragm, colorectum, bladder, uterus and vagina at different locations on the pelvic structure in the extracted image.

5. The construction method according to claim 1, characterized in that: In step S2, the levator ani muscles include the pubovaginal muscle, the pubococcygeus muscle and the puborectalis muscle; the levator ani muscles include the iliococcygeus muscle and the levator ani muscles.

6. The construction method according to claim 1, characterized in that: In the step S2, an integrated Export / import models and labelmaps algorithm is adopted, and the models information integration module is used to export the Mesh model and label map of each manually reconstructed and segmented tissue structure respectively. The volume of the target structure is extracted through the information volume measurement algorithm tool, and the volume of each tissue structure can be obtained by multiplying the number of voxels contained in the generated label map by the voxel size of the image.

7. The construction method according to claim 1, characterized in that: In step S3, the description of the POP-Q index of the two groups is based on the POP-Q score using Logit regression analysis to predict the level of muscle volume; the Wilcoxon rank sum test method is used to verify the POP-Q index during the test.

8. The construction method according to claim 1, characterized in that: In step S3, the difference index at least includes age, C, Ba, gh, pb, tvl, Bp and D. The C refers to the farthest end of the vaginal apex after cervix or hysterectomy; Ba refers to the farthest point in the upper part of the anterior vaginal wall between the front end of the vagina or the anterior fornix and the Aa point; Aa refers to the midline of the anterior vaginal wall 3 cm away from the edge of the hymen; gh refers to the midline distance from the midline of the external urethral orifice to the midline of the posterior edge of the hymen; pb refers to the distance from the posterior edge of the vulvar fissure to the midpoint of the anus; tvl refers to the total vaginal length; Bp refers to the farthest point in the upper part of the posterior vaginal wall between the vaginal apex or the posterior fornix and the Ap point, and Bp corresponds to the Ap point; Ap refers to the midline of the posterior vaginal wall 3 cm away from the hymen, and Ap corresponds to the Aa point; D refers to the position of the posterior fornix when there is a cervix.

9. The construction method according to claim 8, characterized in that: In step S4, the prediction model of muscle volume at different positions includes at least the following formula: (1) Iliocaudal muscle height = logit (-5.04379-0.13209*Ba-0.03222C+0.1995*gh+0.1231*pb+ 0.58071*tvl-0.20253* Bp+0.05719*D); (2) Coccygeal muscle height = logit (-4.197395-0.024958*old age-0.345093*Aa+0.086413*Ba+0.006805C+0.465753*gh-0.066995*pb+0.287507*tvl+0.014876*Ap-0.121509*Bp+0.005364*D); (3) Height of levator ani muscle = logit (-3.37790-0.76268*age-0.29136*Aa-0.01824*Ba-0.02412*C+0.42796*gh+0.11895*pb+0.27907*tvl+0.31562*Ap-0.19732*Bp+0.05606*D); (4) Anal levator muscle height = logit (-1.24849-0.05855*Aa-0.01493*Ba-0.00468*C+0.09061*gh+ 0.04563*pb+ 0.17617*tvl-0.04719*Bp+0.01337*D), The elderly refers to age > 60.

Images