A method for predicting immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images, and a method for measuring urethral curvature in the prediction model.

By combining a logistic regression model based on preoperative MRI images with urethral curvature and other indicators, the problem of accurate prediction of urinary continence after robot-assisted prostatectomy was solved, achieving precise preoperative prediction of urinary continence and optimization of surgical procedures.

CN123118449APending Publication Date: 2026-11-17THE FIRST AFFILIATED HOSPITAL OF FUJIAN MEDICAL UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611047995.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-15
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Current technologies cannot accurately predict the immediate recovery of urinary continence after robot-assisted radical prostatectomy. Existing indicators cannot fully reflect the functional anatomical structures that support urethral closure and maintain urinary continence, and there is a lack of reliable preoperative predictive indicators.

Method used

Based on preoperative MRI images, a predictive model is constructed using a Logistic regression model. Combining indicators such as membranous urethral length, prostate volume, and urethral curvature, a simple and reliable geometric calculation method for urethral curvature is adopted to objectively quantify urethral curvature and achieve accurate prediction of immediate postoperative urinary incontinence.

Benefits of technology

It provides a reliable preoperative prediction tool that can accurately predict postoperative urinary continence, providing a basis for the formulation and adjustment of surgical plans, simplifying the measurement process and not increasing examination time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123118449A_ABST
    Figure CN123118449A_ABST
Patent Text Reader

Abstract

This invention provides a method, model, and urethral curvature measurement method for predicting immediate postoperative urinary continence in prostate cancer patients based on preoperative MRI images. The prediction method involves collecting and filtering MRI image samples from patients, dividing the multi-center MRI image sample set into a training set, an internal validation set, and an external validation set. Predictive indicators are selected from the training set of this center to build a prediction model. Using the predictive indicators as feature variables and whether immediate urinary continence is achieved after RS-RARP as the target variable, a logistic regression algorithm is used to train the model, resulting in a trained prediction model. The model is then validated using both the internal and external validation sets. Predictive indicators are extracted from the MRI images to be predicted, input into the prediction model, and individual scores for each indicator are obtained. A total score is accumulated, and the corresponding probability of immediate urinary continence recovery after RS-RARP is output based on the total score. This invention achieves accurate prediction of immediate postoperative urinary continence without requiring additional scanning sequences.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of medical auxiliary prediction technology, and in particular to a method, prediction model and urethral curvature measurement method for predicting immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images. Background Technology

[0002] Prostate cancer (PCa) is the second most common malignant tumor among men worldwide. Radical prostatectomy (RP) is one of the standard radical treatment options for localized and locally advanced prostate cancer. Postoperative urinary incontinence, as the most common complication, can severely impair patients' quality of life.

[0003] With the continuous development of minimally invasive surgical techniques, robot-assisted radical prostatectomy (RARP) has become a mainstream surgical procedure. In 2010, Galfano et al. first proposed Retzius-sparing RARP (RS-RARP), which uses a posterior approach to protect the neurovascular bundle (NVB). Compared with the traditional anterior approach, it can improve the early postoperative urinary continence (UC) recovery. However, in clinical practice, the immediate urinary continence recovery rate after RS-RARP is only 38% to 80%, and some patients still cannot immediately regain urinary continence after surgery. Therefore, finding reliable preoperative predictive indicators for risk stratification of immediate urinary continence recovery after RS-RARP and for surgical planning has important clinical value.

[0004] In recent years, preoperative multiparametric magnetic resonance imaging (mpMRI) has received widespread attention in predicting postoperative urinary continence recovery after prostate cancer surgery. Existing studies have explored the potential of membranous urethral length (MUL), prostate volume (PV), intravesical prostatic protrusion length (IPPL), and morphological and anatomical parameters of the prostate apex as predictive factors. However, these indicators cannot fully reflect the functional anatomical structures supporting urethral closure and maintaining urinary continence. Other indicators have been proposed for predicting postoperative urinary continence recovery, but previous studies have not yet explored how to quantify them, nor have they verified their predictive efficacy for immediate urinary continence after RS-RARP.

[0005] In view of this, there is an urgent need to study a prostate cancer immediate postoperative urinary incontinence prediction scheme based on preoperative imaging characteristics, so as to achieve a more accurate prediction of immediate postoperative urinary incontinence and make it a clinically practical and scalable tool, so that more patients can benefit from the surgery. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a method, prediction model and urethral curvature measurement method for predicting immediate postoperative urinary incontinence in prostate cancer based on preoperative MRI images. By designing a simple and reliable geometric calculation method for urethral curvature, the urethral curvature can be objectively and repeatedly quantified, achieving a relatively accurate prediction of immediate postoperative urinary incontinence, and providing a reliable basis for the formulation and adjustment of surgical plans.

[0007] In a first aspect, the present invention provides a method for predicting immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images. The method involves collecting and screening MRI image sample sets of patients, and after preprocessing, dividing the MRI image sample sets into training sets, internal validation sets, and external validation sets. Predictive indicators are obtained by screening the training set, and a predictive model is built using the predictive indicators. The predictive indicators include membranous urethra length, prostate volume and urethral curvature. The predictive model adopts a logistic regression nomogram model. Using the training set, with the prediction index as the feature variable and whether urinary continence is restored immediately after RS-RARP as the target variable, the prediction model is trained by the logistic regression algorithm to obtain the trained prediction model. The model is then validated by the internal validation set and the external validation set respectively. After the validation is passed, a prediction model with qualified accuracy and generalization ability is obtained. Predictive indicators are extracted from the MRI images to be predicted, input into a qualified prediction model, and individual scores for each predictive indicator are obtained. The total score is obtained by accumulating the individual scores of each predictive indicator. The corresponding probability is obtained by drawing a vertical line based on the total score, which is used as the output of the probability of immediate recovery of urinary continence after RS-RARP. The immediate urinary control is defined as the patient not needing to use an incontinence pad or only using one protective pad daily within one week after catheter removal. The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the beginning of the membranous urethra.

[0008] Furthermore, the images in the MRI image sample set were all acquired using a 3.0T magnetic resonance scanner, coupled with an external phased array coil via the abdomen.

[0009] Furthermore, the method for obtaining the urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .

[0010] Furthermore, the length of the membranous urethra is obtained as follows: a sagittal T2-weighted image slice is selected, and the lower edge of the prostate tip and the upper edge of the penile root are obtained by manual segmentation. The urethral length between the two is calculated as the length of the membranous urethra. The prostate volume was obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices were selected, and the maximum long diameter, wide diameter and high diameter of the prostate were measured respectively. The prostate volume was calculated using the formula: Volume = Length × Width × Height × 0.523.

[0011] Furthermore, the phrase "collecting and screening the MRI image sample set of the patient, and after preprocessing, dividing the MRI image sample set into a training set, an internal validation set, and an external validation set" specifically includes: collecting information from different patients and screening the MRI image sample set. The screening of the MRI image sample set must simultaneously meet the following conditions: (1) prostate cancer confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter MRI scan completed within 30 days before surgery; (3) TNM stage of cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) no adjuvant therapy such as endocrine therapy or radiotherapy before surgery; (5) no history of urethral trauma or urethral surgery; (6) no severe lower urinary tract symptoms before surgery; (7) non-recurrent / metastatic prostate cancer. Preprocessing is performed, including file format standardization, image normalization, noise reduction, artifact removal, and sequence registration to the same anatomical coordinate system to ensure uniformity of the three-dimensional measurement benchmark; The MRI image sample set of patients from the same hospital area was used as the training set and internal validation set, while the MRI image sample set of patients from different hospital areas was used as the external validation set.

[0012] Furthermore, the phrase "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Acquire all images in the training set and perform preprocessing. All image metrics were measured on a Philips IntelliSpace Portal workstation using a T2WI sequence, including: Membranous urethral length: The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis; Membranous urethral width: The maximum transverse diameter of the membranous urethra is measured using axial images; Prostate volume: Prostate width × length × height × 0.523; Intravesical prostate protrusion length: The vertical distance from the apex of the prostate protrusion to the base of the bladder on a midsagittal image; Prostatic urethral length: The length of the urethra from the apex of the prostate to the opening of the bulbourethral opening in a sagittal image; Pubic symphysis-prostate apex distance: the distance between the extended line of the superior pubic crest and the extended line of the prostate apex in sagittal MRI images; Urethral curvature: The reciprocal of the radius of curvature of the urethra from the bladder neck to the membranous urethral proximal segment; Step S12: Using all measurement indicators as candidate variables, perform dimensionality reduction, eliminate variable collinearity, and select the optimal variable combination through LASSO regression to determine the predictive indicators to be included in the predictive model. Then, verify their independent predictive value through multifactor regression and integrate them to construct the predictive model.

[0013] Furthermore, the model validation uses the area under the receiver operating characteristic curve to assess the model's discriminative ability, while the model's predictive accuracy is tested using calibration curves and mean squared error.

[0014] Secondly, the present invention provides a predictive model for immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images. The model is used to obtain predictive indicators for patients to be predicted. The predictive indicators include membranous urethral length, prostate volume and urethral curvature. The model calculates the individual score for each predictive indicator, accumulates the individual scores of each predictive indicator to obtain the total score, and draws a vertical line based on the total score to obtain the corresponding probability, which is used as the corresponding RS-RARP postoperative immediate urinary incontinence recovery probability output. The immediate urinary control is defined as the patient not needing to use an incontinence pad or only needing to use one protective pad daily within one week after catheter removal. The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation. The prediction model construction process includes: The MRI image sample set of patients was collected and screened. After preprocessing, the MRI image sample set was divided into training set, internal validation set and external validation set. Predictive indicators are obtained by screening the training set, and a predictive model is built using the predictive indicators. The predictive indicators include membranous urethra length, prostate volume and urethral curvature. The predictive model adopts a logistic regression nomogram model. Using the training set, with the prediction index as the feature variable and whether urinary continence is restored immediately after RS-RARP as the target variable, the prediction model is trained by the logistic regression algorithm to obtain the trained prediction model. The model is then validated by the internal validation set and the external validation set respectively. After the validation is passed, a prediction model with qualified accuracy and generalization ability is obtained.

[0015] Furthermore, the method for obtaining the urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .

[0016] Furthermore, the length of the membranous urethra is obtained as follows: a sagittal T2-weighted image slice is selected, and the lower edge of the prostate tip and the upper edge of the penile root are obtained by manual segmentation. The urethral length between the two is calculated as the length of the membranous urethra. Furthermore, the prostate volume is obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices are selected, and the maximum long diameter, wide diameter and high diameter of the prostate are measured respectively. The prostate volume is calculated as follows: Volume = Length × Width × Height × 0.523.

[0017] Furthermore, the phrase "collecting and screening the MRI image sample set of the patient, and after preprocessing, dividing the MRI image sample set into a training set, an internal validation set, and an external validation set" specifically includes: collecting information from different patients and screening the MRI image sample set. The screening of the MRI image sample set must simultaneously meet the following conditions: (1) prostate cancer confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter MRI scan completed within 30 days before surgery; (3) TNM stage of cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) no adjuvant therapy such as endocrine therapy or radiotherapy before surgery; (5) no history of urethral trauma or urethral surgery; (6) no severe lower urinary tract symptoms before surgery; (7) non-recurrent / metastatic prostate cancer. Preprocessing is performed, including file format standardization, image normalization, noise reduction, artifact removal, and sequence registration to the same anatomical coordinate system to ensure uniformity of the three-dimensional measurement benchmark; The MRI image sample set of patients from the same hospital area was used as the training set and internal validation set, while the MRI image sample set of patients from different hospital areas was used as the external validation set.

[0018] Furthermore, the phrase "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Obtain all images from the training set. All image metrics were measured on a Philips IntelliSpacePortal workstation using a T2WI sequence, including: Membranous urethral length: The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis; Membranous urethral width: The maximum transverse diameter of the membranous urethra is measured using axial images; Prostate volume: Prostate width × length × height × 0.523; Intravesical prostate protrusion length: The vertical distance from the apex of the prostate protrusion to the base of the bladder on a midsagittal image; Prostatic urethral length: The length of the urethra from the apex of the prostate to the opening of the bulbourethral opening in a sagittal image; Pubic symphysis-prostate apex distance: the distance between the extended line of the superior pubic crest and the extended line of the prostate apex in sagittal MRI images; Urethral curvature: The reciprocal of the radius of curvature of the urethra from the bladder neck to the membranous urethral proximal segment; Step S12: Using all measurement indicators as candidate variables, perform dimensionality reduction, eliminate variable collinearity, and select the optimal variable combination through LASSO regression to determine the predictive indicators to be included in the predictive model. Then, verify their independent predictive value through multifactor regression and integrate them to construct the predictive model.

[0019] Furthermore, the model validation uses the area under the receiver operating characteristic curve to assess the model's discriminative ability, while the model's predictive accuracy is tested using calibration curves and mean squared error.

[0020] Thirdly, the present invention provides a method for measuring urethral curvature, the method comprising: defining urethral curvature as the reciprocal of the radius of the urethral curve in the initial segment from the bladder neck to the membranous urethra; Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .

[0021] One or more technical solutions provided by this invention have at least the following technical effects or advantages: (1) A new method for quantitative measurement of urethral curvature based on preoperative mpMRI has been established. The geometric calculation process of urethral curvature is simple and can objectively and repeatedly quantify the degree of urethral curvature, providing a new observation dimension for urethral morphology research.

[0022] (2) The multi-index prediction model based on urethral curvature of the present invention can achieve relatively accurate prediction of postoperative urinary incontinence before surgery, providing a highly reliable basis for the formulation and adjustment of surgical plans.

[0023] (3) All the measurement indicators required by the prediction model are derived from the routine preoperative 3.0T mpMRI T2-weighted sequence in clinical practice. No additional scanning sequence is required, and the examination time will not be prolonged. This fully demonstrates the good clinical applicability and feasibility of the model and has outstanding practical application value. Attached Figure Description

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0025] Figure 1 This is a flowchart illustrating the execution of a method for predicting immediate postoperative urinary incontinence in prostate cancer based on preoperative MRI images, according to the present invention. Figure 2This is a schematic diagram of the various measurement parameters of the prostate magnetic resonance imaging and the four-part classification of the prostate apex in this invention; in, Figure 2 A: Length of the membranous urethra; Figure 2 B: Width of the membranous urethra; Figure 2 C: Prostate width; Figure 2 D: Prostate length; Figure 2 E: Prostate height; Figure 2 F: Length of protrusion of the prostate gland within the bladder; Figure 2 G: Length of the prostatic urethra; Figure 2 H: Distance between the pubic symphysis and the apex of the prostate; Figure 2 I: Measurement of urethral curvature (the solid red line represents the urethral arc). The chord length c is the line segment connecting the bladder neck and the starting point of the membranous urethra; h is the vertical distance from the chord to the apex of the urethral arc; the radius R of the fitted circle is marked in yellow, and the curvature = 1 / R; Figure 2 J: Lee A type: The apex of the prostate covers the membranous portion of the urethra on both sides, anterior and posterior; Figure 2 K: Lee B type: The apex of the prostate only covers the anterior part of the membranous urethra; Figure 2 L: Lee C type: The apex of the prostate only covers the membranous urethra behind it; Figure 2 M: LeeD type: The apex of the prostate is not covered by the membranous urethra; Figure 3 for Figure 2 The enlarged view of I is a schematic diagram of the urethral curvature measurement process of the present invention (the red solid line represents the urethral curvature).

[0026] Figure 4 This is a schematic diagram illustrating the process of selecting predictive factors using the LASSO logistic regression algorithm of this invention. in, Figure 4 A: LASSO coefficient variation curves for 9 candidate variables; Figure 4 B: The two dashed lines represent the criteria for optimal λ values: the left dashed line represents the minimum mean square error criterion, and the right dashed line represents the standard error criterion; this study selected 6 predictors based on the standard error criterion (λ=0.015). LASSO: Minimum absolute contraction and selection operator; Figure 5 This is a nomogram predicting immediate urinary incontinence recovery after RS-RARP surgery, as presented in this invention. The unit for membranous urethral length is mm, prostate volume is mL, and urethral curvature is [missing information]. ; Figure 6 The receiver operating characteristic (ROC) curves for model validation are shown in Figure B, where Figure C is the ROC curve for the internal validation set, and Figure D is the ROC curve for the external validation set. Figure 7Calibration curves for model validation: Figure E shows the calibration curve for the training set; Figure F shows the calibration curve for the internal validation set; Figure G shows the calibration curve for the external validation set. Figure 8 The decision curves for model validation are shown in Figure H, where Figure I is the decision curve for the internal validation set, and Figure J is the decision curve for the external validation set.

[0027] UC: Urinary incontinence; RS-RARP: Retzius gap-preserving robot-assisted radical prostatectomy; ROC: Receiver operating characteristic curve; MUL: Membranous urethra length; PV: Prostate volume; DCA: Decision curve analysis.

[0028] Figure 9 Table 1 in this invention is a table of baseline clinical data for patients; Figure 10 Table 2 in this invention is a comparison table of baseline and perioperative data between the immediate continence group and the non-immediate continence group in the training set. Figure 11 Table 3 in this invention is a table of independent predictors of immediate postoperative urinary incontinence after RS-RARP screening using multivariate logistic regression. Figure 12 Table 4 in this invention is a table showing the efficacy of various magnetic resonance indicators in predicting immediate urinary continence after RS-RARP. Figure 13 Table S1 in this invention is a multi-parameter magnetic resonance scanning sequence parameter table; Figure 14 Table S2 in this invention is an inter-observer agreement table for magnetic resonance measurement indicators; Figure 15 Table S3 in this invention is a comparison table of AUC DeLong test results for different prediction models on the training set. Detailed Implementation

[0029] The overall concept of the technical solution in this application is as follows: Accurate preoperative imaging prediction is crucial for helping surgeons formulate scientific surgical plans. This application studies the predictive factors for immediate postoperative urinary incontinence in robot-assisted radical prostatectomy with preservation of the Retzius space (Retzius-sparing RARP, RS-RARP), screening for important influencing factors on the prediction results. For urethral curvature, which may be of significant importance, this application proposes for the first time an objective and repeatable quantitative measurement method, and incorporates urethral curvature along with two other significantly influential factors (membranous urethral length and prostate volume) to jointly construct a prediction model (Nomo model) with high-precision predictive capabilities. The trained prediction model is obtained through training on a training set. Validated by internal and external multi-center cohorts, it exhibits excellent discriminative power, good calibration, and significant clinical practical value, providing a precise quantitative tool for preoperative urinary incontinence risk stratification and individualized surgical planning for prostate cancer patients.

[0030] Example 1 like Figure 1 As shown in this embodiment, a method for predicting immediate urinary incontinence after prostate cancer surgery based on preoperative MRI images is provided. In this invention, immediate urinary incontinence is defined as: within one week after catheter removal, the patient does not need to use a urinary pad or only needs to use one protective pad daily.

[0031] The prediction method of this invention is to collect and screen the MRI image sample set of patients, and after preprocessing, divide the MRI image sample set into training set, internal validation set and external validation set; Predictive indicators were obtained by screening the training set, and a predictive model was built using the predictive indicators. The predictive indicators included the length of the membranous urethra, the volume of the prostate, and the curvature of the urethra. The predictive model adopted the Logistic regression nomogram model (Nomo model). Using the training set, the prediction index is used as the feature variable, and whether urinary continence is restored immediately after RS-RARP is used as the target variable. The prediction model is trained by logistic regression algorithm to obtain the trained prediction model. The model is then validated by internal and external validation sets. After the validation is passed, a prediction model with qualified accuracy and generalization ability is obtained to ensure the discrimination accuracy and generalization ability of the prediction model. Predictive indicators are extracted from the MRI images to be predicted, input into a qualified prediction model, and individual scores for each predictive indicator are obtained. The total score is obtained by accumulating the individual scores of each predictive indicator. The corresponding probability is obtained by drawing a vertical line based on the total score, which is used as the output of the probability of immediate recovery of urinary continence after RS-RARP. The membranous urethra length is defined as the length of the urethra between the lower edge of the prostate tip and the upper edge of the penile root. This urethra segment includes both smooth muscle and striated muscle fibers. The prostate volume is defined as the value of prostate width × length × height × 0.523 (i.e., π / 6). The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the beginning of the membranous urethra.

[0032] To improve the consistency of measurement results, all images in the MRI image sample set were acquired using a 3.0T MRI scanner with an external phased array coil via the abdominal cavity. Specifically, a Siemens MAGNETOM Verio 3.0T MRI scanner with an external phased array coil via the abdominal cavity was used. The patient was in a supine position, and the bladder was moderately filled before scanning. The scanning sequence included: (1) axial, coronal, and sagittal T2-weighted imaging (T2WI); (2) axial T1-weighted imaging (T1WI); (3) diffusion-weighted imaging (DWI); (4) apparent diffusion coefficient (ADC); and (5) dynamic contrast enhancement (DCE). The contrast agent gadopentetate dimeglumine (Gd-DTPA) was injected via the antecubital vein at a flow rate of 3.0 mL / s, and 15 phases were scanned simultaneously. For detailed MRI scanning parameters, see [link to relevant documentation]. Figure 13 Table S1.

[0033] Preferably, the method for obtaining the urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R (in cm) of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: ,unit .

[0034] Preferably, the length of the membranous urethra is obtained as follows: a sagittal T2-weighted image slice is selected, and the lower edge of the prostate tip and the upper edge of the penile root are obtained by manual segmentation. The urethral length between the two is calculated as the length of the membranous urethra, in mm. The prostate volume was obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices were selected, and the maximum long axis (vertical direction), wide axis (left-right direction), and high axis (anteroposterior direction) of the prostate were measured respectively. The prostate volume was calculated using the formula: Volume (ml) = Length (cm) × Width (cm) × Height (cm) × 0.523 (i.e., π / 6), in mL.

[0035] Preferably, the phrase "collecting and screening the patient's MRI image sample set, and after preprocessing, dividing the MRI image sample set into a training set, an internal validation set, and an external validation set" specifically includes: collecting information from different patients and screening the MRI image sample set. The screening of the MRI image sample set must simultaneously meet the following conditions: (1) prostate cancer confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter MRI scan completed within 30 days before surgery; (3) TNM stage of cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) no adjuvant therapy such as endocrine therapy or radiotherapy before surgery; (5) no history of urethral trauma or urethral surgery; (6) no severe lower urinary tract symptoms before surgery; (7) non-recurrent / metastatic prostate cancer. Preprocessing is performed, including file format standardization, image normalization, noise reduction, artifact removal, and sequence registration to the same anatomical coordinate system to ensure uniformity of the three-dimensional measurement benchmark; The MRI image sample set of patients from the same hospital area was used as the training set and internal validation set, while the MRI image sample set of patients from different hospital areas was used as the external validation set.

[0036] This retrospective study was approved by the Ethics Committee of the First Affiliated Hospital of Fujian Medical University (ethics approval number: MRCTA, ECFAH of FMU

[2025] 002). This study included prostate cancer patients who underwent RS-RARP at the hospital between January 2022 and January 2026, divided into a training set and an internal validation set according to admission time; patients who underwent surgery at other hospitals (Nanjing University School of Medicine Affiliated Gulou Hospital, Shandong University Qilu Hospital, and Baoji Central Hospital) between February 2025 and January 2026 served as the external validation set. Training set: 348 patients from our hospital between January 2022 and January 2025; Internal validation set: 103 patients from our hospital between February 2025 and January 2026; External validation set: 129 patients from three external hospitals between February 2025 and January 2026; Inclusion criteria for this study: (1) Prostate cancer was confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter magnetic resonance imaging was completed within 30 days before the operation; (3) TNM staging was cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) Complete clinical and pathological data; (5) Complete follow-up data.

[0037] Exclusion criteria for the subjects in this study: (1) Preoperative neoadjuvant therapy (endocrine therapy, radiotherapy); (2) History of urethral trauma or urethral surgery; (3) Preoperative severe lower urinary tract symptoms; (4) Recurrent / metastatic prostate cancer; (5) Missing or lost to follow-up clinical, imaging, or pathological data.

[0038] The data collected from the above-mentioned research subjects included: Demographic and clinical baseline data: age, body mass index (BMI), hypertension, diabetes, history of prostate surgery, D'Amico risk stratification, PI-RADS score, International Prostate Symptom Score (IPSS), and clinical TNM stage (refer to the 2022 NCCN guidelines). Perioperative and pathological data: preoperative prostate-specific antigen (PSA), operation time (from skin incision to suturing completion), estimated blood loss, blood transfusion, preservation of neurovascular bundle, postoperative drainage volume, drainage tube, catheter removal time, hospital stay, surgical margin status, nerve invasion, and Gleason score for puncture.

[0039] All subjects in the above study underwent RS-RARP posterior approach surgery using the da Vinci Xi surgical system.

[0040] In this embodiment, the step of "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Acquire all images in the training set. All preoperative anatomical parameter measurements were performed according to the Prostate Imaging Reporting and Data System, version 2 (PI-RADS v2), independently measured by two radiologists with 10 and 8 years of imaging diagnostic experience, respectively, under the guidance of a senior urologist. The two measurers were mutually blinded and unaware of the patient's clinical data and follow-up outcomes; any discrepancies in the measurement results were resolved through joint image review. All image parameter measurements were performed on a Philips IntelliSpace Portal workstation with T2WI sequences, such as... Figure 2As shown, it includes: Membranous Urethral Length (MUL): The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis (Figure 2A). Membranous urethral width (MUW): The maximum transverse diameter of the membranous urethra is measured using axial images (Figure 2B). Prostate volume (PV): Prostate width × length × height × 0.523 (Figure 2C-E); Intravesical Prostatic Protrusion Length (IPPL): The vertical distance from the apex of the prostatic protrusion to the base of the bladder on a midsagittal image (Figure 2F). Prostatic urethral length (PUL): The length of the urethra from the apex of the prostate to the opening of the bulbous urethra on a sagittal image (Fig. 2G). Pubic symphysis-prostate apex length (PAL): The distance between the extended line of the superior pubic crest and the extended line of the prostate apex in a sagittal MRI image (Fig. 2H). Urethral curvature (UC): The reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation (Figure 2I). Step S12: Using all measurement indicators as candidate variables, LASSO regression is used to reduce the dimensionality of all candidate variables, eliminate variable collinearity, and screen the optimal variable combination to determine the predictive indicators to be included in the prediction model. The predictive indicators include membranous urethral length, prostate volume, and urethral curvature. Then, multifactor regression is used to verify their independent predictive value, and then the prediction model is integrated to obtain the individualized Nomogram risk prediction model.

[0041] The magnetic resonance imaging measurements used in this study included: MUL, MUW, PV, urethral curvature, IPPL, PUL, PAL, and Lee classification of the prostate apex. The Lee classification of the prostate apex is as follows: Class A: The prostate apex covers the membranous portion of the urethra, both anterior and posterior (…). Figure 2 J); Category B: The apex of the prostate only covers the anterior part of the membranous urethra ( Figure 2 K); C: The apex of the prostate only covers the membranous urethra behind it ( Figure 2 L); D: Uncovered membranous urethra at the apex of the prostate ( Figure 2 M); Statistical analyses were performed using SPSS version 26.0 software (IBM, Armonk, NY, USA). Inter-observer consistency was assessed using the weighted Cohen's kappa (κw) coefficient for categorical variables and the within-group correlation coefficient (ICC) for continuous variables. Kappa values ​​were interpreted as: poor (<0.20), fair (0.21–0.40), moderate (0.41–0.60), good (0.61–0.80), or almost perfectly consistent (0.81–1.00); ICC values ​​were categorized as: poor (<0.40), fair (0.40–0.59), good (0.60–0.74), and excellent (0.75–1.00). Normally distributed data are expressed as mean ± standard deviation. The Student's t-test is used for comparisons between two groups, and one-way ANOVA is used for comparisons of three or more groups. Non-normally distributed data are expressed as median and interquartile range (IQR). The Mann-Whitney U test is used for comparisons between two groups, and the Kruskal-Wallis test is used for comparisons of three or more groups. Categorical variables are expressed as frequency and percentage, and are analyzed using either the chi-square test or Fisher's exact test, respectively.

[0042] This study used univariate analysis and a Lasso regression model to screen for the best predictors, and then incorporated the selected predictors into a multivariate logistic regression analysis. The predictors of immediate urinary tract infection (UC) after RS-RARP were determined by calculating the odds ratio (OR) and 95% confidence interval (CI). The results of the multivariate logistic regression were presented as a nomogram, constructed using R statistical software (R Foundation for Statistical Computation in Vienna, Austria, version 4.1.0). The discriminative power of the nomogram was assessed using the area under the receiver operating characteristic (AUC), and the predictive accuracy was tested using calibration curves and mean squared error (MSE). Decision curve analysis was used to compare the clinical applicability of the nomogram and each independent parameter across the full risk threshold range (0–1). The correlation and collinearity of the final predictors were assessed using Pearson correlation coefficient and variance inflation factor (VIF). ROC curve analysis was performed using MedCalc version 15.2 software (Ostend, Belgium). The optimal cutoff value, sensitivity, and specificity were determined based on the maximum Youden index to evaluate the predictive efficacy of the nomogram. Positive predictive value (PPV) and negative predictive value (NPV) were used to assess the predictive value of each parameter for immediate recovery from urinary tract infection after RS-RARP, and the area under the ROC curve (AUC) and 95% confidence interval were calculated. The DeLong test was used to compare the AUC values ​​of different variables. A two-sided p-value <0.05 was considered statistically significant.

[0043] result Baseline characteristics of all patients' demographics, perioperative pathology, and magnetic resonance imaging. The study included 580 prostate cancer patients who underwent RS-RARP, of whom 451 were from our hospital (348 in the training set and 103 in the internal validation set) and 129 were from other hospitals (78 from Gulou Hospital, 47 from Qilu Hospital, and 4 from Baoji Central Hospital). Baseline data for the overall population can be found here. Figure 9 Table 1: The median patient age was 68 years (range 62–74 years), and the median BMI was 24.09 kg / m² (range 22.43–25.50). Clinical T stage: T1 30.7%, T2 62.2%, T3 7.1%; Gleason score for biopsy: 6 points 30.2%, 7 points 48.6%, ≥8 points 21.2%; median preoperative PSA was 11.12 ng / mL (range 7.49–18.68). The immediate postoperative urinary continence recovery rate was 83.1% (482 / 580); 84.5% (294 / 348) in the training set, 80.6% (83 / 103) in the internal validation set, and 81.4% (105 / 129) in the external validation set. There were no statistically significant differences in demographic, clinicopathological, or MRI indicators among the training set, internal validation set, and external validation set (all P > 0.05).

[0044] One-way comparison of immediate and non-immediate urinary control groups in the training set The imaging measurements showed good observer consistency, with all magnetic resonance imaging parameters having ICC values ​​ranging from 0.86 to 0.98 (see...). Figure 14 (See Table S2). Univariate analysis was used to compare various variables between the immediate and non-immediate urinary continence (UC) groups in the training cohort. The non-immediate urinary continence group had a higher body mass index (BMI) (25.08 vs. 23.84 kg / m², P = 0.005), a larger prostate volume (PV) (58.64 vs. 49.34 mL, P<0.001), and a higher proportion of T3 stage (13.0% vs. 4.4%, P = 0.046). The percentage of bilateral neurovascular bundles preserved was a key factor in protecting urinary continence, and it was significantly higher in the immediate continence group (41.8% vs 20.4%, P < 0.001). Significant differences were also observed in anatomical measurements between the groups: the membranous urethra (MUL) was longer in the immediate continence group (16.00 vs 12.83 mm, P < 0.001), the intravesical prostate protrusion length (IPPL) was shorter (4.27 vs 4.96 mm, P = 0.028), the pubic symphysis-prostate apex distance was shorter (26.12 vs 27.53, P = 0.030), the urethral curvature was greater (0.321 vs 0.285, P < 0.001), and the percentage of D-type prostate apex was higher (59.2% vs 9.3%, P < 0.001) (see [link to relevant documentation]). Figure 10 Table 2); Logistic regression was used to screen independent predictors of immediate postoperative urinary incontinence after RS-RARP. Univariate analysis is susceptible to confounding factors. Figure 10 The statistically significant indicators in Table 2 were included in the LASSO regression, and the predictive variables with non-zero regression coefficients were selected: neurovascular bundle preservation, prostate apex type D, clinical T stage, membranous urethral length, prostate volume, and urethral curvature (Figures 4A and 4B).

[0045] Furthermore, the above six potential predictive indicators and clinically relevant confounding variables (operation duration) were incorporated into a multivariate logistic regression to screen for independent preoperative predictive factors (see...). Figure 11 (Table 3) Membranous urethral length (MUL): OR=1.23, P=0.029, an independent protective factor against immediate urinary continence; Urethral curvature (UC): OR=2.84, P<0.001, an independent protective factor against immediate urinary continence; Prostate volume (PV): OR=0.84, P<0.001, an independent risk factor for immediate urinary continence (the larger the prostate volume, the lower the probability of immediate urinary continence). Construction and validation of predictive RS-RARP postoperative immediate urinary continence nomogram Based on multivariate regression incorporating the three indicators MUL, PV, and UC, a visual prediction nodal plot was constructed on the training set (see...). Figure 5 The correlations among the three indicators were as follows: the correlation coefficient between UC and PV was -0.13, and the correlation coefficient between UC and MUL was 0.17; the VIF for all three was 1.014, indicating no multicollinearity. ROC analysis showed that the AUC values ​​for MUL, PV, and urethral curvature were 0.74, 0.75, and 0.76, respectively (see [link to ROC analysis]). Figure 12 Table 4 and Figure 6(B in the original text). To enhance its clinical application value, a multivariate prediction model was constructed by combining these three parameters using a permutation test. The AUC values ​​of the two-parameter joint test (MUL+PV), (MUL+UC), and (PV+UC) were 0.83, 0.82, and 0.83, respectively.

[0046] Preferably, the model validation uses the area under the receiver operating characteristic curve (AUC) to assess the model's discriminative ability, while the model's predictive accuracy is tested using calibration curves and mean squared error (MSE).

[0047] The nomogram model exhibits excellent discriminative performance: its AUC in the training cohort reaches 0.87 (95% confidence interval: 0.83–0.91; see...). Figure 6 Model B in the model significantly outperformed the performance of individual parameter and two-parameter combined models (including MUL (DeLong test z=4.05, P<0.001), PV (z=4.50, P<0.001), UC (z=3.96, P<0.001), (MUL+PV) combined model (z=2.96, P=0.003), (MUL+UC) combined model (z=2.01, P=0.044), and (PV+UC) combined model (z=2.78, P=0.005)) (see [reference]). Figure 15 (See Table S3). The discriminative performance of this plural plot was further validated: the AUC value in the internal validation queue was 0.93 (95% confidence interval: 0.86–0.97; see Table S3). Figure 6 The C value in the external validation queue also reached 0.93 (95% confidence interval: 0.88–0.97; see C in the external validation queue). Figure 6 (D in the text). The calibration curves showed a high degree of agreement between the observed incidence of immediate urinary continence and the predicted values, with mean squared errors of 0.00078, 0.00433, and 0.0024 for the training cohort, internal validation cohort, and external validation cohort, respectively (see D). Figure 7 ).like Figure 8 As shown, the decision curve analysis indicates that, within a reasonable threshold probability range, the nomogram provides a higher net benefit compared to individual parameters, fully demonstrating its superiority in clinical decision guidance. This study is the first multicenter clinical trial to investigate the impact of male physiological urethral curvature on postoperative urinary continence after robot-assisted radical prostatectomy (RARP), including 580 patients (348 in the training set, 103 in the internal validation set, and 129 in the external validation set). This study innovatively established a quantitative measurement method for urethral curvature and validated its predictive efficacy for immediate postoperative urinary continence recovery after Retzius-preserving robot-assisted radical prostatectomy (RS-RARP) using multicenter data: the areas under the ROC curve (AUC) for the training set, internal validation set, and external validation set were 0.76, 0.84, and 0.83, respectively (see [link to study]. Figure 6 Furthermore, this study combined membranous urethral length (MUL), prostate volume (PV), and urethral curvature to construct a prediction model based on multiparameter magnetic resonance imaging (mpMRI) (i.e., the Nomo model). The model (Nomo model) exhibited excellent predictive performance, with AUCs of 0.87, 0.93, and 0.93 on the training set, internal validation set, and external validation set, respectively (see [link to study]). Figure 6 The model demonstrated superior predictive performance compared to single imaging indicators and combined models of two indicators (all DeLong tests showed P < 0.05). Calibration curves and mean squared error (MSE) confirmed the reliability of the nomogram's predictive accuracy, and decision curve analysis showed that the model significantly improved clinical net benefit across the entire range of risk thresholds. This model can accurately assess the risk of postoperative urinary incontinence in patients, providing a theoretical basis for developing individualized treatment plans.

[0048] Preoperative multiparameter magnetic resonance imaging (mpMRI) is not only a core tool for the diagnosis of prostate cancer (PCa), but also provides rich and valuable anatomical information for intraoperative procedures and postoperative functional assessment. Current research focuses on analyzing anatomical parameters related to the urethral sphincter, prostate gland, pelvic floor support structures, and pelvic-related parameters. Existing studies have confirmed that sphincter-related anatomical parameters are closely related to surgical difficulty, complication risk, neurovascular bundle preservation, and postoperative functional recovery. Membranous urethral length (MUL) refers to the length of the urethra from the apex of the prostate to the root of the penis; this segment of the urethra contains both smooth and striated muscle fibers. Current research indicates that patients with longer preoperative MUL tend to have longer residual membranous urethra postoperatively, which is more conducive to early recovery of urinary continence. Furthermore, studies have found that prostate volume (PV) is only associated with urinary incontinence at 3 and 6 months after RS-RARP, and has no significant impact on long-term urinary continence at 12 months postoperatively, suggesting that prostate volume has a weak effect on long-term urinary continence recovery. However, this study showed that patients with smaller prostate volumes (<53.94 mL) experienced faster immediate postoperative urinary continence recovery. The mechanism may be that a larger prostate volume leads to a wider bladder neck during surgery, significantly increasing the difficulty of bladder neck anastomosis; simultaneously, benign prostatic hyperplasia can cause short-term bladder compensatory dysfunction, thereby inducing postoperative urinary incontinence. Urethral curvature, by optimizing urethral pressure distribution and improving urodynamics, mechanistically prevents urinary incontinence. Based on this, posterior approach surgeries like RS-RARP preserve the natural physiological curvature of the urethra by protecting the pubic ligament, bladder neck, and other urethral supporting tissues, maintaining optimal urethral curvature after anastomosis between the bladder neck and the urethral stump. This study proposes urethral curvature as a quantitative indicator, defined as the reciprocal of the radius of the urethral arc from the bladder neck to the beginning of the membranous urethra; this indicator can objectively and repeatedly quantify the degree of urethral curvature, providing a new observational dimension for urethral morphology research. The results of this study confirm that urethral curvature is significantly associated with immediate recovery of urinary continence after RS-RARP, with the optimal cutoff value for this indicator being 0.323 in the training set.

[0049] From a clinical perspective, the urethral curvature measurement and predictive nomogram model has strong feasibility for clinical application. All required measurement indicators are derived from routine preoperative 3.0T mpMRI T2-weighted sequences, requiring no additional scanning sequences and not prolonging examination time. The geometric calculation process for urethral curvature is simple, and the inter-observer intraclass correlation coefficient (ICC) is 0.86-0.98, confirming excellent reproducibility of this indicator in routine clinical work. The nomogram is visually intuitive and easy to operate, enabling rapid patient risk stratification without the need for complex calculation software. It is suitable for preoperative patient communication and surgical planning and can be directly incorporated into the standardized clinical diagnosis and treatment process for RS-RARP patients.

[0050] This invention establishes a novel, simple, repeatable, and clinically appropriate method for quantitative measurement of urethral curvature using mpMRI, verifying that this indicator can independently predict immediate urinary continence recovery after RS-RARP. The predictive nomogram constructed based on three MRI indicators—membranous urethral length, prostate volume, and urethral curvature—demonstrates excellent discriminative power, good calibration, and significant clinical value through internal and external multicenter cohort validation. It can provide a precise quantitative tool for preoperative urinary incontinence risk stratification and individualized surgical planning in prostate cancer patients.

[0051] Based on the same inventive concept, this application also provides an apparatus corresponding to the method in Embodiment 1, as detailed in Embodiment 2.

[0052] Example 2 This embodiment provides a predictive model for immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images. The model is used to obtain predictive indicators for patients to be predicted. The predictive indicators include membranous urethral length, prostate volume and urethral curvature. The model calculates the individual score for each predictive indicator, accumulates the individual scores of each predictive indicator to obtain the total score, and draws a vertical line based on the total score to obtain the corresponding probability, which is used as the output of the corresponding RS-RARP postoperative immediate urinary incontinence recovery probability. The immediate urinary control is defined as the patient not needing to use an incontinence pad or only needing to use one protective pad daily within one week after catheter removal. The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation. The prediction model construction process includes: The MRI image sample set of patients was collected and screened. After preprocessing, the MRI image sample set was divided into training set, internal validation set and external validation set. Predictive indicators are obtained by screening the training set, and a predictive model is built using the predictive indicators. The predictive indicators include membranous urethra length, prostate volume and urethral curvature. The predictive model adopts a logistic regression nomogram model. Using the training set, with the prediction index as the feature variable and whether urinary continence is restored immediately after RS-RARP as the target variable, a prediction model is trained using a logistic regression algorithm. The trained prediction model is then validated using both an internal validation set and an external validation set. Once the validation is successful, a prediction model with satisfactory accuracy and generalization ability is obtained, ensuring the accuracy and generalization ability of the prediction model.

[0053] Furthermore, the method for obtaining the urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R (in cm) of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .

[0054] Preferably, the length of the membranous urethra is obtained as follows: a sagittal T2-weighted image slice is selected, and the lower edge of the prostate tip and the upper edge of the penile root are obtained by manual segmentation. The urethral length between the two is calculated as the length of the membranous urethra (unit: mm). Preferably, the prostate volume is obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices are selected, and the maximum long diameter, wide diameter and high diameter of the prostate are measured respectively. The prostate volume is calculated using the formula: volume = length × width × height × 0.523 (unit, mL).

[0055] Preferably, the phrase "collecting and screening the patient's MRI image sample set, and after preprocessing, dividing the MRI image sample set into a training set, an internal validation set, and an external validation set" specifically includes: collecting information from different patients and screening the MRI image sample set. The screening of the MRI image sample set must simultaneously meet the following conditions: (1) prostate cancer confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter MRI scan completed within 30 days before surgery; (3) TNM stage of cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) no adjuvant therapy such as endocrine therapy or radiotherapy before surgery; (5) no history of urethral trauma or urethral surgery; (6) no severe lower urinary tract symptoms before surgery; (7) non-recurrent / metastatic prostate cancer. Preprocessing is performed, including file format standardization, image normalization, noise reduction, artifact removal, and sequence registration to the same anatomical coordinate system to ensure uniformity of the three-dimensional measurement benchmark; The MRI image sample set of patients from the same hospital area was used as the training set and internal validation set, while the MRI image sample set of patients from different hospital areas was used as the external validation set.

[0056] Preferably, the "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Obtain all images from the training set. All image metrics were measured on a Philips IntelliSpacePortal workstation using a T2WI sequence, including: Membranous Urethral Length (MUL): The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis (Figure 2A). Membranous urethral width (MUW): The maximum transverse diameter of the membranous urethra is measured using axial images (Figure 2B). Prostate volume (PV): Prostate width × length × height × 0.523 (Figure 2C-E); Intravesical Prostatic Protrusion Length (IPPL): The vertical distance from the apex of the prostatic protrusion to the base of the bladder on a midsagittal image (Figure 2F). Prostatic urethral length (PUL): The length of the urethra from the apex of the prostate to the opening of the bulbous urethra on a sagittal image (Fig. 2G). Pubic symphysis-prostate apex length (PAL): The distance between the extended line of the superior pubic crest and the extended line of the prostate apex in a sagittal MRI image (Fig. 2H). Urethral curvature (UC): The reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation (Figure 2I). Step S12: Using all measurement indicators as candidate variables, perform dimensionality reduction, eliminate variable collinearity, and select the optimal variable combination through LASSO regression to determine the predictive indicators to be included in the predictive model. Then, verify their independent predictive value through multifactor regression and integrate them to construct the predictive model.

[0057] Preferably, the model validation uses the area under the receiver operating characteristic curve (AUC) to assess the model's discriminative ability, while the model's predictive accuracy is tested using calibration curves and mean squared error (MSE).

[0058] Since the model described in Embodiment 2 of this invention is an apparatus used to implement the method of Embodiment 1 of this invention, those skilled in the art can understand the specific structure and variations of the apparatus based on the method described in Embodiment 1 of this invention, and therefore will not be repeated here. All apparatuses used in the method of Embodiment 1 of this invention fall within the scope of protection of this invention.

[0059] Example 3 like Figure 3 As shown, this embodiment provides a method for measuring urethral curvature, the method comprising: defining urethral curvature as the reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation segment; Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .

[0060] To improve the accuracy of urethral curvature measurement, sagittal T2-weighted image slices that most clearly display the urethral arc between the bladder neck and the beginning of the membranous urethra were selected. The distance between the two inflection points (A, B) on either side of the dome was calculated. Figure 3 (marked as line c), and measure the distance from the dome apex to line c along a direction perpendicular to that line ( Figure 3 (The symbol is denoted as h). The urethral curvature is then calculated based on the radius R (in cm) of the imaginary circle defined by the chord length c and height h. In this embodiment, the average curvature of the urethral curve segment is... Defined as: the total tangent deflection angle α (unit: radians) and the arc If the ratio of the arc length *s* to the arc length *s*, then for any arc segment on the circle... The tangent deflection angle α is equal in size to the central angle subtended by the arc. Substituting this into the definition of the mean curvature using the arc length formula s=Rα, we can derive: ; The curvature K at any point on the circle is the limit of K as the arc length approaches zero: .

[0061] This embodiment establishes a novel method for quantitatively measuring urethral curvature based on preoperative mpMRI. Through a simplified geometric calculation process, the degree of urethral curvature can be objectively and repeatedly quantified, providing a new observational dimension for urethral morphology research. The urethral curvature measurement index obtained by this method can be used to further construct a multi-index prediction model, achieving relatively accurate prediction of immediate postoperative urinary incontinence before surgery, and providing a reliable basis for the formulation and adjustment of surgical plans.

[0062] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the present invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images, characterized in that: The MRI image sample set of patients was collected and screened. After preprocessing, the MRI image sample set was divided into training set, internal validation set and external validation set. Predictive indicators are obtained by screening the training set, and a predictive model is built using the predictive indicators. The predictive indicators include membranous urethra length, prostate volume and urethral curvature. The predictive model adopts a logistic regression nomogram model. Using the training set, with the prediction index as the feature variable and whether urinary continence is restored immediately after RS-RARP as the target variable, the prediction model is trained by the logistic regression algorithm to obtain the trained prediction model. The model is then validated by the internal validation set and the external validation set respectively. After the validation is passed, a prediction model with qualified accuracy and generalization ability is obtained. Predictive indicators are extracted from the MRI images to be predicted, input into a qualified prediction model, and individual scores for each predictive indicator are obtained. The total score is obtained by accumulating the individual scores of each predictive indicator. The corresponding probability is obtained by drawing a vertical line based on the total score, which is used as the output of the probability of immediate recovery of urinary continence after RS-RARP. The immediate urinary control is defined as the patient not needing to use an incontinence pad or only using one protective pad daily within one week after catheter removal. The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the beginning of the membranous urethra.

2. The prediction method according to claim 1, characterized in that: The images in the MRI image sample set were all acquired using a 3.0T magnetic resonance scanner, coupled with an external phased array coil via the abdomen.

3. The prediction method according to claim 1, characterized in that: The specific method for obtaining urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: ; The method for obtaining the length of the membranous urethra is as follows: Select a slice of the sagittal T2-weighted image, obtain the lower edge of the prostate tip and the upper edge of the penile root by manual segmentation, and calculate the urethral length between the two as the length of the membranous urethra; The prostate volume was obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices were selected, and the maximum long diameter, wide diameter and high diameter of the prostate were measured respectively. The prostate volume was calculated using the formula: Volume = Length × Width × Height × 0.

523.

4. The prediction method according to claim 1, characterized in that: The phrase "collecting and screening MRI image sample sets from patients, and after preprocessing, dividing the MRI image sample sets into training sets, internal validation sets, and external validation sets" specifically includes: collecting information from different patients and screening MRI image sample sets. The screening of the MRI image sample sets must simultaneously meet the following conditions: (1) prostate cancer confirmed by prostate biopsy and postoperative pathology; (2) 3.0T prostate / pelvic multiparameter MRI scan completed within 30 days before surgery; (3) TNM stage of cT1, cT2, cT3aN0M0 localized tumor without regional lymph node or distant metastasis; (4) no adjuvant therapy such as endocrine therapy or radiotherapy before surgery; (5) no history of urethral trauma or urethral surgery; (6) no severe lower urinary tract symptoms before surgery; (7) non-recurrent / metastatic prostate cancer. Preprocessing is performed, including file format standardization, image normalization, noise reduction, artifact removal, and sequence registration to the same anatomical coordinate system to ensure uniformity of the three-dimensional measurement benchmark; The MRI image sample set of patients from the same hospital area was used as the training set and internal validation set, while the MRI image sample set of patients from different hospital areas was used as the external validation set.

5. The prediction method according to claim 1, characterized in that: The phrase "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Obtain all images from the training set. All image metrics were measured on a Philips IntelliSpacePortal workstation using a T2WI sequence, including: Membranous urethral length: The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis; Membranous urethral width: The maximum transverse diameter of the membranous urethra is measured using axial images; Prostate volume: Prostate width × length × height × 0.523; Intravesical prostate protrusion length: The vertical distance from the apex of the prostate protrusion to the base of the bladder on a midsagittal image; Prostatic urethral length: The length of the urethra from the apex of the prostate to the opening of the bulbourethral opening in a sagittal image; Pubic symphysis-prostate apex distance: the distance between the extended line of the superior pubic crest and the extended line of the prostate apex in sagittal MRI images; Urethral curvature: The reciprocal of the radius of curvature of the urethra from the bladder neck to the membranous urethral proximal segment; Step S12: Using all measurement indicators as candidate variables, perform dimensionality reduction, eliminate variable collinearity, and select the optimal variable combination through LASSO regression to determine the predictive indicators to be included in the predictive model. Then, verify their independent predictive value through multifactor regression and integrate them to construct the predictive model.

6. The prediction method according to claim 1, characterized in that: The model validation uses the area under the receiver operating characteristic curve to assess the model's discriminative ability, while the model's predictive accuracy is tested using calibration curves and mean squared error.

7. A predictive model for immediate postoperative urinary incontinence in prostate cancer patients based on preoperative MRI images, characterized in that: The predictive indicators for obtaining the patient to be predicted include membranous urethra length, prostate volume and urethral curvature. The individual scores of each predictive indicator are calculated, and the total score is obtained by accumulating the individual scores of each predictive indicator. The corresponding probability is obtained by drawing a vertical line based on the total score, which is used as the output of the probability of immediate recovery of urinary incontinence after RS-RARP. The immediate urinary control is defined as the patient not needing to use an incontinence pad or only needing to use one protective pad daily within one week after catheter removal. The urethral curvature is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation. The prediction model construction process includes: The MRI image sample set of patients was collected and screened. After preprocessing, the MRI image sample set was divided into training set, internal validation set and external validation set. Predictive indicators are obtained by screening the training set, and a predictive model is built using the predictive indicators. The predictive indicators include membranous urethra length, prostate volume and urethral curvature. The predictive model adopts a logistic regression nomogram model. Using the training set, with the prediction index as the feature variable and whether urinary continence is restored immediately after RS-RARP as the target variable, the prediction model is trained by the logistic regression algorithm to obtain the trained prediction model. The model is then validated by the internal validation set and the external validation set respectively. After the validation is passed, a prediction model with qualified accuracy and generalization ability is obtained.

8. The prediction model according to claim 7, characterized in that: The specific method for obtaining urethral curvature is as follows: Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: ; The length of the membranous urethra is obtained as follows: a slice of the sagittal T2-weighted image is selected, and the lower edge of the prostate tip and the upper edge of the penile root are obtained by manual segmentation. The urethral length between the two is calculated as the length of the membranous urethra. The prostate volume was obtained as follows: sagittal T2-weighted image slices and horizontal T2-weighted image slices were selected, and the maximum long diameter, wide diameter and high diameter of the prostate were measured respectively. The prostate volume was calculated using the formula: Volume = Length × Width × Height × 0.

523.

9. The prediction model according to claim 7, characterized in that: The phrase "using the training set to filter and obtain prediction indicators" specifically includes: Step S11: Obtain all images from the training set. All image metrics were measured on a Philips IntelliSpacePortal workstation using a T2WI sequence, including: Membranous urethral length: The length of the urethra from the lower edge of the apex of the prostate to the upper edge of the root of the penis; Membranous urethral width: The maximum transverse diameter of the membranous urethra is measured using axial images; Prostate volume: Prostate width × length × height × 0.523; Intravesical prostate protrusion length: The vertical distance from the apex of the prostate protrusion to the base of the bladder on a midsagittal image; Prostatic urethral length: The length of the urethra from the apex of the prostate to the opening of the bulbourethral opening in a sagittal image; Pubic symphysis-prostate apex distance: the distance between the extended line of the superior pubic crest and the extended line of the prostate apex in sagittal MRI images; Urethral curvature: The reciprocal of the radius of curvature of the urethra from the bladder neck to the membranous urethral proximal segment; Step S12: Using all measurement indicators as candidate variables, perform dimensionality reduction, eliminate variable collinearity, and select the optimal variable combination through LASSO regression to determine the predictive indicators to be included in the predictive model. Then, verify their independent predictive value through multifactor regression and integrate them to construct the predictive model.

10. A method for measuring urethral curvature, characterized in that: The measurement method includes: The curvature of the urethra is defined as the reciprocal of the radius of the urethral curve from the bladder neck to the membranous urethral initiation. Select sagittal T2-weighted image slices of the urethral arc between the bladder neck and the beginning of the membranous urethra; In a workstation or measuring tool, after manually dividing and marking the position of the urethral dome and the inflection points on both sides of the dome, the distance between the two inflection points is automatically output and marked as chord length c; the distance from the dome apex to chord length c is measured along a direction perpendicular to chord length c and marked as height h; Construct a urethral curvature calculation module to output the urethral curvature based on the input chord length c and height h; The urethral curvature calculation module includes: The radius R of the defined imaginary circle is calculated using the following geometric formula: R = ; Calculate and output the urethral curvature as: .