Clinical research of primiparous women with postpartum pelvic floor dysfunction using AI-assisted pelvic floor ultrasound combined with tomographic ultrasound imaging

doi:10.3877/cma.j.issn.1673-5250.2025.04.014

Objective

To investigates the application of AI-assisted pelvic floor ultrasound combined with tomographic ultrasound imaging (TUI) in assessing the structure and function of pelvic floor muscles in patients with postpartum pelvic floor dysfunction (PFD).

Methods

A total of 115 postpartum PFD patients admitted to Cangzhou People′s Hospital from April 2021 to October 2023 were selected as the research subjects and included in the PFD group. Additionally, 96 primiparous women without PFD during the same period were selected as the control group. All subjects′ ultrasound images were independently diagnosed by two examiners using a double-blind method. The detrusor wall thickness (DWT), bladder neck descent (BND), retrovesical angle (RA), and urethral rotation angle (URA) measured by intelligent pelvic floor ultrasound under resting state and Valsalva maneuver were compared between the two groups. In addition, the levator hiatus (HA) area, anteroposterior diameter, transverse diameter, and thickness under resting state, anal contraction, and Valsalva maneuver were also compared. The intraclass correlation coefficient (ICC) was used to assess the consistency of parameter measurements from intelligent pelvic floor ultrasound between the two physicians. The diagnostic performance of intelligent pelvic floor ultrasound combined with TUI technology for PFD was evaluated by drawing receiver operating characteristic (ROC) curves. The procedures followed in this study were in accordance with the requirements of the Medical Ethics Committee of Cangzhou People′s Hospital and were approved (Approval No.K2020141). Informed consents were obtained from all subjects.

Results

① There were no statistically significant differences between the two groups in age, postpartum body mass index (BMI), distribution of delivery modes, or episiotomy rates (P>0.05). ② Under both resting conditions and during the Valsalva maneuver, the DWT, BND, RA, and URA were significantly higher in the PFD group compared to the control group (P<0.05). Additionally, the HA area, anteroposterior diameter, transverse diameter, and levator ani muscle thickness were all significantly greater in the PFD group under resting, anal contraction, and Valsalva conditions (P<0.05). ③ The inter-observer agreement for AI-assisted pelvic floor ultrasound measurements was high, with ICC all exceeding 0.75. ROC curve analysis demonstrated that the area under the curve (AUC) for diagnosing PFD was 0.727 (95%CI: 0.657-0.797) using AI-assisted pelvic floor ultrasound alone, 0.777 (95%CI: 0.711-0.842) using TUI alone, and 0.884 (95%CI: 0.878-0.935) when the two modalities were combined.

Conclusions

AI-assisted pelvic floor ultrasound combined with TUI enables dynamic evaluation of structural and functional changes in the pelvic floor muscles of patients with PFD. The measurements demonstrate high reproducibility, providing a reliable imaging basis for the clinical diagnosis of PFD.

Key words: Pelvic floor disorders, Ultrasonography, Artificial intelligence, Ultrasonography, tomographic, Pelvic floor disorders, ROC curve, Diagnostic accuracy, Female

表1 2组初产妇一般临床资料比较

组别	例数	年龄(岁，±s)	分娩方式[例数(%)]		产后接受超声检查时间(d，±s)	会阴侧切[例数(%)]	产后BMI(kg/m²，±s)
组别	例数	年龄(岁，±s)	阴道分娩	剖宫产	产后接受超声检查时间(d，±s)	会阴侧切[例数(%)]	产后BMI(kg/m²，±s)
PFD组	115	26.5±2.2	59(51.3)	56(48.7)	54.2±6.7	34(29.6)	23.2±0.6
对照组	96	26.6±2.2	50(52.1)	46(47.9)	53.8±7.0	30(31.2)	23.2±0.5
统计量		t＝0.40	χ²＝0.01		t＝0.37	χ²＝0.07	t＝0.78
P值		0.689	0.910		0.712	0.790	0.435

表1 2组初产妇一般临床资料比较

组别	例数	年龄(岁，±s)	分娩方式[例数(%)]		产后接受超声检查时间(d，±s)	会阴侧切[例数(%)]	产后BMI(kg/m²，±s)
组别	例数	年龄(岁，±s)	阴道分娩	剖宫产	产后接受超声检查时间(d，±s)	会阴侧切[例数(%)]	产后BMI(kg/m²，±s)
PFD组	115	26.5±2.2	59(51.3)	56(48.7)	54.2±6.7	34(29.6)	23.2±0.6
对照组	96	26.6±2.2	50(52.1)	46(47.9)	53.8±7.0	30(31.2)	23.2±0.5
统计量		t＝0.40	χ²＝0.01		t＝0.37	χ²＝0.07	t＝0.78
P值		0.689	0.910		0.712	0.790	0.435

表2 2组初产妇不同状态下超声检查的盆底解剖结构参数比较(

±s)

组别	例数	DWT(mm)		BND(mm)		RA(°)		URA(°)
组别	例数	静息状态	Valsalva动作	静息状态	Valsalva动作	静息状态	Valsalva动作	静息状态	Valsalva动作
PFD组	115	4.7±0.5	5.8±0.4	28.6±3.6	29.6±3.6	157.2±10.3	159.6±11.7	77.2±6.7	78.6±7.2
对照组	96	2.7±0.3	2.1±0.3	13.3±1.2	11.3±1.6	91.2±8.4	92.1±8.5	31.4±4.8	32.2±5.1
t值		32.88	80.30	39.52	46.08	50.37	47.10	55.94	53.40
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

表2 2组初产妇不同状态下超声检查的盆底解剖结构参数比较(

±s)

组别	例数	DWT(mm)		BND(mm)		RA(°)		URA(°)
组别	例数	静息状态	Valsalva动作	静息状态	Valsalva动作	静息状态	Valsalva动作	静息状态	Valsalva动作
PFD组	115	4.7±0.5	5.8±0.4	28.6±3.6	29.6±3.6	157.2±10.3	159.6±11.7	77.2±6.7	78.6±7.2
对照组	96	2.7±0.3	2.1±0.3	13.3±1.2	11.3±1.6	91.2±8.4	92.1±8.5	31.4±4.8	32.2±5.1
t值		32.88	80.30	39.52	46.08	50.37	47.10	55.94	53.40
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

图1 研究组1例自述产后有不自主漏尿史的PFD初产妇(女性，27岁)经阴道及盆底超声声像图(图1A：经阴道超声可见子宫大小、形态基本正常，轮廓尚规则，宮壁回声均匀，内膜线居中，宫腔内液性暗区深度约为0.49 cm；图1B：静息状态下盆底超声可见子宫明显下降；图1C：缩肛动作下，宫体形态、位置均未见显著异常；图1D：最大Valsalva动作后，可见宫底下缘最低点位于参考线下方5.4 mm；图1E：最大Valsalva动作后，可见肛提肌裂孔无明显扩张；图1F：肛提肌及尿道间隙结构未见明显异常)注：PFD为盆底功能障碍

表3 2组初产妇不同状态下提肛肌HA、前后径、左右径及厚度比较(

±s)

组别	例数	HA(cm²)			前后径(cm)
组别	例数	静息状态	缩肛动作	Valsalva动作	静息状态	缩肛动作	Valsalva动作
PFD组	115	17.0±3.1	14.0±2.4	18.9±3.1	5.3±0.4	4.4±0.3	6.0±0.7
对照组	96	12.1±1.2	11.3±1.4	13.0±1.2	3.3±0.3	3.7±0.2	4.0±0.3
t值		14.30	10.07	17.34	40.00	17.64	25.60
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
组别	例数	左右径(cm)			厚度(cm)
组别	例数	静息状态	缩肛动作	Valsalva动作	静息状态	缩肛动作	Valsalva动作
PFD组	115	5.6±0.8	5.2±0.7	5.7±0.6	0.6±0.1	0.57±0.11	0.75±0.17
对照组	96	4.2±0.5	4.8±0.3	4.3±0.4	0.4±0.1	0.27±0.06	0.31±0.05
t值		14.89	5.09	20.70	16.57	23.91	24.48
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

表3 2组初产妇不同状态下提肛肌HA、前后径、左右径及厚度比较(

±s)

组别	例数	HA(cm²)			前后径(cm)
组别	例数	静息状态	缩肛动作	Valsalva动作	静息状态	缩肛动作	Valsalva动作
PFD组	115	17.0±3.1	14.0±2.4	18.9±3.1	5.3±0.4	4.4±0.3	6.0±0.7
对照组	96	12.1±1.2	11.3±1.4	13.0±1.2	3.3±0.3	3.7±0.2	4.0±0.3
t值		14.30	10.07	17.34	40.00	17.64	25.60
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
组别	例数	左右径(cm)			厚度(cm)
组别	例数	静息状态	缩肛动作	Valsalva动作	静息状态	缩肛动作	Valsalva动作
PFD组	115	5.6±0.8	5.2±0.7	5.7±0.6	0.6±0.1	0.57±0.11	0.75±0.17
对照组	96	4.2±0.5	4.8±0.3	4.3±0.4	0.4±0.1	0.27±0.06	0.31±0.05
t值		14.89	5.09	20.70	16.57	23.91	24.48
P值		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

表4 2名超声科医师对2组初产妇智能盆底超声测量的一致性评价结果

指标	ICC	ICC 95%CI
DWT	0.914	0.878~0.940
BND	0.955	0.935~0.969
RA	0.842	0.772~0.891
URA	0.927	0.887~0.953
HA	0.841	0.770~0.890
前后径	0.887	0.840~0.920
左右径	0.940	0.913~0.958
提肛肌厚度	0.938	0.910~0.957

表4 2名超声科医师对2组初产妇智能盆底超声测量的一致性评价结果

指标	ICC	ICC 95%CI
DWT	0.914	0.878~0.940
BND	0.955	0.935~0.969
RA	0.842	0.772~0.891
URA	0.927	0.887~0.953
HA	0.841	0.770~0.890
前后径	0.887	0.840~0.920
左右径	0.940	0.913~0.958
提肛肌厚度	0.938	0.910~0.957

图2 智能盆底超声和TUI技术诊断PFD的ROC曲线注：TUI为断层超声成像，PFD为盆底功能障碍，ROC曲线为受试者工作特征曲线

表5 智能盆底超声、TUI技术及二者联合诊断PFD的ROC曲线分析结果

预测因子	ROC曲线			敏感度(%)	特异度(%)	约登指数(%)
预测因子	AUC	AUC 95%CI	P值	敏感度(%)	特异度(%)	约登指数(%)
智能盆底超声	0.727	0.657~0.797	<0.001	72.1	61.3	33.4
TUI技术	0.777	0.711~0.842	<0.001	76.3	64.1	40.4
二者联合	0.884	0.878~0.935	<0.001	89.6	79.8	69.4

表5 智能盆底超声、TUI技术及二者联合诊断PFD的ROC曲线分析结果

预测因子	ROC曲线			敏感度(%)	特异度(%)	约登指数(%)
预测因子	AUC	AUC 95%CI	P值	敏感度(%)	特异度(%)	约登指数(%)
智能盆底超声	0.727	0.657~0.797	<0.001	72.1	61.3	33.4
TUI技术	0.777	0.711~0.842	<0.001	76.3	64.1	40.4
二者联合	0.884	0.878~0.935	<0.001	89.6	79.8	69.4

[1]	张珂，王澜静，焦玲洁，等. 产后盆底功能障碍性疾病与盆底肌收缩力及其相关因素分析[J]. 实用妇产科杂志，2014，30(10): 757-759.
[2]	李晓阳，刘柏隆，周祥福. 大数据及人工智能对女性盆底功能障碍性疾病的诊断及风险预测[J/OL]. 中华腔镜泌尿外科杂志(电子版)，2023，17(6)：549-552. DOI：10.3877/cma.j.issn.1674-3253.2023.06.001.
[3]	孙玉坤，吴丽娜，杨秋晔，等. 经会阴盆底超声智能识别联合多平面成像技术评估盆膈裂孔形态及功能[J]．临床超声医学杂志，2023，25(3): 217-220. DOI: 10.16245/j.cnki.issn1008-6978.2023.03.017.
[4]	陶均佳，应涛，陈爱萍，等．断层超声成像技术在分娩时耻骨直肠肌损伤诊断中的应用[J]．中国医科大学学报，2023, 52(2): 179-182.
[5]	江苏省中西医结合学会盆底疾病专业委员会，上海市医师协会肛肠专业委员会，北京整合医学学会盆底疾病专业委员会，等. 盆底功能障碍性疾病诊断及康复治疗专家共识 [J/OL]. 中华临床医师杂志(电子版)，2024，18(2): 113-121. DOI: 10.3877/cma.j.issn.1674-0785.2024.02.001.
[6]	中华医学会泌尿外科分会女性泌尿学组，牛晓宇，潘铁军，等．女性盆腔器官脱垂的风险预警及早期干预专家共识[J]．实用妇产科杂志，2024, 40(7): 532-537.
[7]	Brennen R, Lin KY, Denehy L, et al. The effect of pelvic floor muscle interventions on pelvic floor dysfunction after gynecological cancer treatment: a systematic review [J]. Phys Ther, 2020, 100(8): 1357-1371. DOI: 10.1093/ptj/pzaa081.
[8]	闵洁，彭国平，陈艳，等. 经会阴三维盆底超声在女性压力性尿失禁诊断及病情程度评估中的应用[J]. 临床与病理杂志，2022, 42(9): 2159-2165.
[9]	花卉，许晨莹，刘金凤. 盆底肌训练时机对高龄初产妇产后盆底肌张力及性功能的影响[J/OL]. 中华妇幼临床医学杂志(电子版)，2024，20(3)：284-291. DOI：10.3877/cma.j.issn.1673-5250.2024.03.007.
[10]	李多，鲁蓉．盆底超声评估女性肛提肌形态及功能的研究进展 [J]. 中南大学学报(医学版), 2023, 48(8): 1267-1273.
[11]	叶琴，薛恩生，梁荣喜，等. 经阴道三维超声容积成像与断层超声显像技术对宫腔粘连的诊断及漏诊分析[J]. 中华超声影像学杂志，2020, 29(3)：255-259. DOI: 10.3760/cma.j.cn131148-20190812-00478.
[12]	Tennfjord MK, Engh ME, Bφ K. The influence of early exercise postpartum on pelvic floor muscle function and prevalence of pelvic floor dysfunction 12 months postpartum [J]. Phys Ther, 2020, 100(9): 1681-1689. DOI: 10.1093/ptj/pzaa084.
[13]	赵春桃，梁峰雪，杨瑞敏，等. 三维盆底超声预测产妇发生盆腔脏器脱垂的价值及影响因素[J/OL]. 中华妇幼临床医学杂志(电子版)，2022，18(5)：606-614. DOI：10.3877/cma.j.issn.1673-5250.2022.05.016.
[14]	胡海燕，郭宏斌，张静芳，等. 盆底二维超声结合三维超声断层成像在未育女性压力性尿失禁中的初步临床应用[J]. 生物医学工程与临床，2022，26(1)：22-28. DOI: 10.13339/j.cnki.sglc.20211217.014.
[15]	Papanikolaou DT, Lampropoulou S, Giannitsas K, et al. Pelvic floor muscle training: novel versus traditional remote rehabilitation methods. A systematic review and Meta-analysis on their effectiveness for women with urinary incontinence [J]. Neurourol Urodyn, 2023, 42(4): 856-874. DOI: 10.1002/nau.25150.

[1]	Yao Jiang, Cheng Jiang, Xiang Yu, Ying Tan, Xin Wen, Huiying Wen, Guiyan Peng, Shengli Li. Performance of an attention-enhanced multi-task model for uterine anatomical structure detection and segmentation[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(08): 703-710.
[2]	Minglang Chen, Kai Xu, Zhixi Huang, Bocheng Liang, Jie He, Haishan Huang, Weibo Ma, Ying Tan, Zhiying Zou, Xiaotang Liu, Guiyan Peng, Jiaxi Chen, Xiaohong Zhong. MobileNetV4: an intelligent tool for diagnosis of aortic arch branching anomalies based on prenatal ultrasound images[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(08): 711-720.
[3]	xian Li, Zhixi Huang, Bocheng Liang, Shuyuan Ouyang, Minglang Chen, Yingli Zhao, Weibo Ma, Jing Miao, Lei Wang, Ying Yuan. Intelligent prediction of neonatal birth weight based on prenatal sequential ultrasound dataYang[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(08): 721-732.
[4]	Liling Lu, Xiuzhen Yang, Bin Xu, Lei Zhao, Jingjing Qian, Xiaoying Li, Biao Wang, Jingjing Ye. Ultrasonic manifestations and diagnostic value of torsion of the pedicle of old ovarian cysts in infants and young children[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(08): 748-753.
[5]	Qingqing Liu, Jin Yu, Weize Xu, Zhiwei Zhang, Xiaohua Pan, Qiang Shu, Jingjing Ye. Value of OBICnet image classification model in ultrasound screening for congenital heart disease in children[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(08): 754-760.
[6]	Xiaofang Fu, Qinghan Yang, Changqin Sun, Mengjie Dou, Junpu Hu, Hao Sun, Faqin Lyu. Clinical value of a YOLO 11-based ultrasound detection model for limb long bone fracture ends[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(06): 541-546.
[7]	Lang Wang, Yihua He, Zhengyi Li, Cuiyun Liu, Jianfeng Xie, Jian Chen. Predictive value of maternal and fetal ultrasound parameters for adverse pregnancy outcomes in pregnant women with recurrent miscarriage[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(06): 556-563.
[8]	Guannan He, Ying Tan, Yuhuan Lu, Bin Pu, Shuihua Yang, Rentie Zhang, Ming Chen, Zhihong Shi, Xiaohong Zhong, Xi Chen, Liuyi Yan, Shengli Li. Application of artificial intelligence in quality control of standard views for fetal echocardiography: a multi-center study[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(05): 388-396.
[9]	Yin Chen, Ying Tan, Bohan Tan, Guannan He, Lei Wang, Xin Wen, Qiaozhen Zhu, Bocheng Liang, Shengli1 Li. Intelligent quality assessment and diagnosis of fetal omphalocele using YOLO V8 in ultrasound imaging[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2025, 22(04): 305-310.
[10]	Jiale Zhang, Lu Tian, Guosheng Wu, Yingying Liu, Zhi Li, Qiong Wu, Shizhao Ji. Application of artificial intelligence in the diagnosis and treatment of burns in naval warfare[J]. Chinese Journal of Injury Repair and Wound Healing(Electronic Edition), 2025, 20(05): 426-430.
[11]	Shuang Shi, Yi Wang, Na Shi, Wei Xu. Precision applications and future prospects of artificial intelligence in chronic wound management via multi-source information fusion[J]. Chinese Journal of Injury Repair and Wound Healing(Electronic Edition), 2025, 20(05): 431-435.
[12]	Renjie Zheng, Zunshu Zhang, Chen Huang. Application and challenges of artificial intelligence-based pathomics in the diagnosis and management of gastric cancer[J]. Chinese Archives of General Surgery(Electronic Edition), 2025, 19(04): 274-280.
[13]	Yi Wang, Jianqiao Kong, Peng Zhang, Yang Dai, Hengping Li. Laparoscopic ultrasound-guided duodenoscopy for the treatment of gallbladder combined with common bile duct stones[J]. Chinese Journal of Operative Procedures of General Surgery(Electronic Edition), 2025, 19(06): 658-661.
[14]	Lingjie Wang, Aiping Wang, Chaojun Li, Yueyou Ding, Deye Yang, Qing Zhao, Zhaoqiang Cui, Jingkun Wang, Hongyu Wang. Expert Consensus on Research and Development Strategies for Innovative Heart and Vascular Health Technologies (2024, Shanghai)[J]. Chinese Journal of Clinicians(Electronic Edition), 2025, 19(05): 323-336.
[15]	lin Liu, Huahua Zhang, Min Yang, Mingru Zhang, Yinsha Miao, Zijun He, Zhe Zhao, Jinshan Ji, Jin Wang, Jiming Han. Evaluation of the effect of DeepSeek's practical application in emergency and disaster medicine teaching[J]. Chinese Journal of Hygiene Rescue(Electronic Edition), 2025, 11(04): 238-243.

Viewed

Full text

Abstract

Please choose a citation manager

Content to export