Application of 3D reconstruction software and MRI measurements in predicting delivery Mode: A comparative study

Read full text Summarise with VIRAT
Authors: Cuicui Jiang, Fei Wang, Xiaoling Chen, Zhimei Du, Yuanyuan Mao, Meizheng Xu, Qiaoying Wang, Yurong Wu, Baihui Zhao, Qiong Luo, Qiumin Zhu
Cited by: 0
References: 32

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Background Predicting delivery mode remains challenging in obstetrics, as traditional methods cannot fully capture the spatial fetal-pelvic relationship crucial for diagnosing cephalopelvic disproportion. MRI-based 3D reconstruction offers a novel way to quantify this spatial alignment. This study evaluated whether software-derived 3D parameters offer superior predictive performance compared to conventional MRI and aimed to develop an individualized prediction model. Methods This study enrolled 216 term pregnant women who underwent pelvic MRI. Clinical characteristics, MRI pelvimetric measurements, and four software-derived parameters (dd1-dd4) were collected. Participants were randomly assigned to training and test sets (7:3). Four feature combinations and four machine-learning algorithms (LR, SVM, RF, XGB) were used to construct predictive models. Model performance was assessed using accuracy, AUC, precision, recall, calibration curves, decision curve analysis, and SHAP interpretability analysis. A nomogram was developed based on the best-performing model. Results The inclusion of software-calculated spatial parameters improved model performance. The SVM model trained with clinical features and software-calculated spatial parameters achieved the best overall balance, with an accuracy of 0.800 (95%CI: 0.702-0.879), an F1-score of 0.711, and the highest AUC (0.893, 95%CI: 0.806-0.952) among all configurations. DCA demonstrated that this model provided the highest clinical net benefit within the clinically relevant threshold probability range (10-50%). SHAP analysis confirmed that the software-derived spatial parameters were among the most influential predictors, capturing spatial fetal-pelvic relationships not reflected in conventional static MRI measurements. A nomogram was developed based on the best-performing model to facilitate individualized prediction of vaginal delivery probability. Conclusion The integration of 3D reconstruction software-derived spatial parameters enhances the predictive performance of MRI pelvimetry for delivery mode. This approach, which quantifies fetal-pelvic spatial geometry, represents a promising step toward individualized labor management and decision-making in Chinese women.

Fields of Study & Subjects

Maternal and Perinatal Health Interventions Maternal and fetal healthcare Pregnancy-related medical research

Publication details

Journal / source
Journal of Radiation Research and Applied Sciences
Volume
19
Issue
2
Pages
102340–102340
Publication date
27 March 2026
Language
EN