Share：
Share this content in WeChat
X
[Chinese][PDF]1112
• Clinical Article •
Prediction of lymphovascular space invasion in endometrial carcinoma based on preoperative multiparameter MRI deep transfer learning features
GUO Ran  PENG Ruchen  LI Yancui  SHEN Xiuzhi  HAO Pan  XIN Ruiqiang 

Cite this article as: GUO R, PENG R C, LI Y C, et al. Prediction of lymphovascular space invasion in endometrial carcinoma based on preoperative multiparameter MRI deep transfer learning features[J]. Chin J Magn Reson Imaging, 2025, 16(3): 70-76, 82. DOI:10.12015/issn.1674-8034.2025.03.011.


[Abstract] Objective This study aimed to develop a model based on deep transfer learning (DTL) features from preoperative multiparametric magnetic resonance imaging (MRI) to predict lymphovascular space invasion (LVSI) status in patients with endometrial carcinoma (EC).Materials and Methods A retrospective analysis was conducted on clinical information and preoperative MRI images of 187 EC patients who were surgically and pathologically confirmed in our hospital from February 2016 to July 2023. The patients were randomly divided into a training set (131 patients) and a test set (56 patients) in a 7∶3 ratio. Regions of interest were delineated on axial T2-weighted imaging, diffusion-weighted imaging, apparent diffusion coefficient (ADC) maps, and contrast-enhanced T1-weighted imaging, manually. Subsequently, 12 DTL models were established using ResNet50, ResNet101, and DenseNet121 networks. Fusion models were then established using three decision-level fusion methods: mean, maximum, and minimum, with the best model selected as the final DTL model. A clinical model was established after screening clinical features through univariate and multivariate logistic regression analysis, and a DTL-clinical combined model was developed using logistic regression incorporating DTL and clinical features. The receiver operating characteristic curve was used to assess the diagnostic performance of the models for LVSI in EC patients, the area under the curve (AUC) was compared using the DeLong test. The calibration curve was used to analyze the goodness of fit of the models, and the decision curve was used to explore the clinical applicability of the models.Results In the test set, the ResNet101 model based on the ADC images showed the highest AUC value of 0.850 [95% confidence interval (CI): 0.736 to 0.963] for diagnosing LVSI in EC patients. The fusion model established using the mean fusion method had the highest AUC value of 0.932 (95% CI: 0.868 to 0.996) in the test set, representing the best DTL model. Logistic regression analysis indicated that age was an independent risk factor for LVSI. The DTL-clinical combined model had an AUC of 0.934 (95% CI: 0.871 to 0.997) in the test set, with significantly better diagnostic performance than the clinical model [AUC: 0.554 (95% CI: 0.436 to 0.671), P < 0.001] and no statistical difference compared to the DTL model (P = 0.909). The combined model demonstrated good fit in both the training and test sets (Hosmer-Lemeshow test: P = 0.814 and 0.402, respectively) and offered greater clinical net benefit.Conclusions The DTL model based on preoperative multiparametric MRI, as well as the combined model integrating DTL features with clinical features, can effectively predict the LVSI status of EC patients, outperforming clinical models. DTL demonstrates excellent performance on our small-sample EC MRI data, providing important clinical assistance for preoperative LVSI prediction.
[Keywords] endometrial carcinoma；lymphvascular space invasion；multiparametric magnetic resonance imaging；deep learning；transfer learning

GUO Ran   PENG Ruchen   LI Yancui   SHEN Xiuzhi   HAO Pan   XIN Ruiqiang*  

Department of Radiology, Beijing Luhe Hospital, Capital Medical University, Beijing 101149, China

Corresponding author: XIN R Q, E-mail: rxin@ccmu.edu.cn

Conflicts of interest   None.

Received  2024-10-11
Accepted  2025-03-04
DOI: 10.12015/issn.1674-8034.2025.03.011
Cite this article as: GUO R, PENG R C, LI Y C, et al. Prediction of lymphovascular space invasion in endometrial carcinoma based on preoperative multiparameter MRI deep transfer learning features[J]. Chin J Magn Reson Imaging, 2025, 16(3): 70-76, 82. DOI:10.12015/issn.1674-8034.2025.03.011.

[1]
SIEGEL R L, MILLER K D, WAGLE N S, et al. Cancer statistics, 2023[J]. CA A Cancer J Clin, 2023, 73(1): 17-48. DOI: 10.3322/caac.21763.
[2]
HAN B F, ZHENG R S, ZENG H M, et al. Cancer incidence and mortality in China, 2022[J]. J Natl Cancer Cent, 2024, 4(1): 47-53. DOI: 10.1016/j.jncc.2024.01.006.
[3]
KOSKAS M, AMANT F, MIRZA M R, et al. Cancer of the corpus uteri: 2021 update[J]. Int J Gynaecol Obstet, 2021, 155(Suppl 1): 45-60. DOI: 10.1002/ijgo.13866.
[4]
OAKNIN A, BOSSE T J, CREUTZBERG C L, et al. Endometrial cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up[J]. Ann Oncol, 2022, 33(9): 860-877. DOI: 10.1016/j.annonc.2022.05.009.
[5]
SUN B W, ZHANG X B, DONG Y Y, et al. Prognostic significance of lymphovascular space invasion in early-stage low-grade endometrioid endometrial cancer: a fifteen-year retrospective Chinese cohort study[J/OL]. World J Surg Oncol, 2024, 22(1): 203 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/39080611/. DOI: 10.1186/s12957-024-03483-6.
[6]
TORTORELLA L, RESTAINO S, ZANNONI G F, et al. Substantial lymph-vascular space invasion (LVSI) as predictor of distant relapse and poor prognosis in low-risk early-stage endometrial cancer[J/OL]. J Gynecol Oncol, 2021, 32(2): e11 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/33470061/. DOI: 10.3802/jgo.2021.32.e11.
[7]
RESTAINO S, TORTORELLA L, DINOI G, et al. Semiquantitative evaluation of lymph-vascular space invasion in patients affected by endometrial cancer: Prognostic and clinical implications[J]. Eur J Cancer, 2021, 142: 29-37. DOI: 10.1016/j.ejca.2020.10.011.
[8]
HUI C, MENDOZA M G, VON EYBEN R, et al. Does lymph node assessment change the prognostic significance of substantial LVSI and p53 status in endometrial endometrioid carcinoma?[J]. Gynecol Oncol, 2023, 177: 150-156. DOI: 10.1016/j.ygyno.2023.09.001.
[9]
BEREK J S, MATIAS-GUIU X, CREUTZBERG C, et al. FIGO staging of endometrial cancer: 2023[J]. Int J Gynaecol Obstet, 2023, 162(2): 383-394. DOI: 10.1002/ijgo.14923.
[10]
HAN Y, SUN M Y. Advances in MRI studies of lymphvascular space invasion in endometrial carcinoma[J]. Chin J Magn Reson Imag, 2024, 15(7): 216-220, 234. DOI: 10.12015/issn.1674-8034.2024.07.036.
[11]
LAMBIN P, RIOS-VELAZQUEZ E, LEIJENAAR R, et al. Radiomics: extracting more information from medical images using advanced feature analysis[J]. Eur J Cancer, 2012, 48(4): 441-446. DOI: 10.1016/j.ejca.2011.11.036.
[12]
LONG L, SUN J Q, JIANG L L, et al. MRI-based traditional radiomics and computer-vision nomogram for predicting lymphovascular space invasion in endometrial carcinoma[J]. Diagn Interv Imaging, 2021, 102(7/8): 455-462. DOI: 10.1016/j.diii.2021.02.008.
[13]
CHEN J Y, WANG X R, LV H Y, et al. Development and external validation of a clinical-radiomics nomogram for preoperative prediction of LVSI status in patients with endometrial carcinoma[J]. J Cancer Res Clin Oncol, 2023, 149(15): 13943-13953. DOI: 10.1007/s00432-023-05044-y.
[14]
LIU D, HUANG J Y, ZHANG Y F, et al. Multimodal MRI-based radiomics models for the preoperative prediction of lymphovascular space invasion of endometrial carcinoma[J/OL]. BMC Med Imaging, 2024, 24(1): 252 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/39304802/. DOI: 10.1186/s12880-024-01430-1.
[15]
SHI B B, GRIMM L J, MAZUROWSKI M A, et al. Prediction of occult invasive disease in ductal carcinoma in situ using deep learning features[J]. J Am Coll Radiol, 2018, 15(3 Pt B): 527-534. DOI: 10.1016/j.jacr.2017.11.036.
[16]
KIM H E, COSA-LINAN A, SANTHANAM N, et al. Transfer learning for medical image classification: a literature review[J/OL]. BMC Med Imaging, 2022, 22(1): 69 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/35418051/. DOI: 10.1186/s12880-022-00793-7.
[17]
SRINIVAS C, NANDINI PRASAD K S, ZAKARIAH M, et al. Deep transfer learning approaches in performance analysis of brain tumor classification using MRI images[J/OL]. J Healthc Eng, 2022, 2022: 3264367 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/35299683/. DOI: 10.1155/2022/3264367.
[18]
AGARWAL D, MARQUES G, DE LA TORRE-DÍEZ I, et al. Transfer learning for Alzheimer's disease through neuroimaging biomarkers: a systematic review[J/OL]. Sensors, 2021, 21(21): 7259 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/34770565/. DOI: 10.3390/s21217259.
[19]
ZHONG X R, CAO R M, SHAKERI S, et al. Deep transfer learning-based prostate cancer classification using 3 Tesla multi-parametric MRI[J]. Abdom Radiol, 2019, 44(6): 2030-2039. DOI: 10.1007/s00261-018-1824-5.
[20]
ATASEVER S, AZGINOGLU N, TERZI D S, et al. A comprehensive survey of deep learning research on medical image analysis with focus on transfer learning[J]. Clin Imaging, 2023, 94: 18-41. DOI: 10.1016/j.clinimag.2022.11.003.
[21]
PETERS E E M, LÉON-CASTILLO A, HOGDALL E, et al. Substantial lymphovascular space invasion is an adverse prognostic factor in high-risk endometrial cancer[J]. Int J Gynecol Pathol, 2022, 41(3): 227-234. DOI: 10.1097/PGP.0000000000000805.
[22]
MENG X X, YANG D W, DENG Y H, et al. Diagnostic accuracy of MRI for assessing lymphovascular space invasion in endometrial carcinoma: a meta-analysis[J]. Acta Radiol, 2024, 65(1): 133-144. DOI: 10.1177/02841851231165671.
[23]
LUO Y, MEI D D, GONG J S, et al. Multiparametric MRI-based radiomics nomogram for predicting lymphovascular space invasion in endometrial carcinoma[J]. J Magn Reson Imaging, 2020, 52(4): 1257-1262. DOI: 10.1002/jmri.27142.
[24]
LEFEBVRE T L, UENO Y, DOHAN A, et al. Development and validation of multiparametric MRI-based radiomics models for preoperative risk stratification of endometrial cancer[J]. Radiology, 2022, 305(2): 375-386. DOI: 10.1148/radiol.212873.
[25]
UENO Y, FORGHANI B, FORGHANI R, et al. Endometrial carcinoma: MR imaging-based texture model for preoperative risk stratification-a preliminary analysis[J]. Radiology, 2017, 284(3): 748-757. DOI: 10.1148/radiol.2017161950.
[26]
PENG Y J, LIU X W, TANG X, et al. Preoperative predicting lymphov-ascular space invasion in endometrial carcinoma by nomogram based on mpMRI radiomics[J]. Chin J Magn Reson Imag, 2022, 13(7): 61-67. DOI: 10.12015/issn.1674-8034.2022.07.011.
[27]
DENG J, DONG W, SOCHER R, et al. ImageNet: a large-scale hierarchical image database[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition. June 20-25, 2009, Miami, FL, USA. IEEE, 2009: 248-255. DOI: 10.1109/CVPR.2009.5206848.
[28]
WANG M, MEI T, GONG Y L. The quality and clinical translation of radiomics studies based on MRI for predicting Ki-67 levels in patients with breast cancer[J/OL]. Br J Radiol, 2023, 96(1150): 20230172 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/37724784/. DOI: 10.1259/bjr.20230172.
[29]
WANG W L, LIANG H, ZHANG Z Y, et al. Comparing three-dimensional and two-dimensional deep-learning, radiomics, and fusion models for predicting occult lymph node metastasis in laryngeal squamous cell carcinoma based on CT imaging: a multicentre, retrospective, diagnostic study[J/OL]. EClinicalMedicine, 2024, 67: 102385 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/38261897/. DOI: 10.1016/j.eclinm.2023.102385.
[30]
LI S Y, WANG Y D, SUN Y Y, et al. Both intra- and peri-tumoral radiomics signatures can be used to predict lymphatic vascular space invasion and lymphatic metastasis positive status from endometrial cancer MR imaging[J]. Abdom Radiol, 2024, 49(11): 4140-4150. DOI: 10.1007/s00261-024-04432-3.
[31]
LIU X F, YAN B C, LI Y, et al. Radiomics feature as a preoperative predictive of lymphovascular invasion in early-stage endometrial cancer: A multicenter study[J/OL]. Front Oncol, 2022, 12: 966529 [2025-03-01]. https://pubmed.ncbi.nlm.nih.gov/36059674/. DOI: 10.3389/fonc.2022.966529.
[32]
YAN B, JIA Y X, LI Z H, et al. Preoperative prediction of lymphovascular space invasion in endometrioid adenocarcinoma: an MRI-based radiomics nomogram with consideration of the peritumoral region[J]. Acta Radiol, 2023, 64(9): 2636-2645. DOI: 10.1177/02841851231181681.

PREV The value of combining whole-tumor ADC histogram parameters with imaging biomarkers in predicting perineural and lymphovascular invasion in rectal adenocarcinoma
NEXT Predictive value of machine learning model based on ADC radiomics in evaluating the invasion depth of endometrial carcinoma
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn