Share：
Share this content in WeChat
X
[Chinese] [PDF] 11 1
• Clinical Article •
Study on predicting LVSI status in preoperative cervical cancer patients without lymph node metastasis using habitat radiomics based on DCE-MRI quantitative parametric maps
LI Feixiang  SUN Yun  GONG Juntao  HUANG Gang 

Cite this article as: LI F X, SUN Y, GONG J T, et al. Study on predicting LVSI status in preoperative cervical cancer patients without lymph node metastasis using habitat radiomics based on DCE-MRI quantitative parametric maps[J]. Chin J Magn Reson Imaging, 2025, 16(10): 89-97. DOI:10.12015/issn.1674-8034.2025.10.014.


[Abstract] Objective To investigate the diagnostic value of a radiomic habitat model based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) quantitative parametric maps for predicting the status of lymph-vascular space invasion (LVSI) in patients with cervical cancer before surgery.Materials and Methods A total of 102 cervical cancer patients who underwent radical hysterectomy at Gansu Provincial People's Hospital between May 2015 and October 2024 were retrospectively analyzed. Patients were stratified into LVSI-positive [LVSI (+)] and LVSI-negative [LVSI (-)] groups according to postoperative pathological findings. Clinical parameters and DCE-MRI quantitative metrics were compared between the two groups. Univariate and multivariate regression analyses were performed to identify independent risk factors associated with LVSI status in cervical cancer. On DCE-MRI images, Tissue4D is applied to determine the peak blood flow phase via the time-intensity curve (TIC) of the internal iliac artery. During this phase, the entire tumor contour is contoured as the volume of interest (VOI) to obtain the transport constant (Ktrans) parameter map. The K-means method was employed to determine the optimal number of clusters, leveraging the voxels and eigenvalues of the Ktrans parametric maps to categorize the VOI into distinct subregions. Intratumoral radiomics features and habitat radiomics features were extracted. Feature dimensionality reduction was performed on each feature dataset of the training set using t-tests, Pearson analysis, and the least absolute shrinkage and selection operator (LASSO) regression. Machine learning algorithms, including the support vector machine (SVM), adaptive boosting (AdaBoost), and multilayer perceptron (MLP), were utilized to construct intratumoral radiomics models and habitat radiomics models. Feature fusion (pre-fusion) and result fusion (post-fusion) methods were adopted to construct a combined model integrating habitat radiomics and intratumoral radiomics for model development. The receiver operating characteristic (ROC) curve, calibration curve, and decision curve were used to evaluate model performance.Results Among the 102 cervical cancer patients, 38 cases were LVSI (+) and 64 cases were LVSI (-). Univariate logistic regression analysis revealed that age, height, body weight, body mass index, Ktrans, and rate constant (Kep) were factors associated with LVSI status in cervical cancer [odds ratios (ORs) = 0.989, 0.997, 0.991, 0.978, 0.045, 0.372; P = 0.011, 0.010, 0.008, 0.010, 0.038, 0.018, respectively]. Multivariate logistic regression analysis of clinical parameters did not identify any independent risk factors associated with LVSI for the construction of a clinical model (P > 0.05). The optimal number of habitat subregion clusters was determined to be three. Intratumoral and habitat radiomics models were constructed using 18 and 8 optimal radiomics features derived from the habitat regions and the entire tumor, respectively. Among these models, the pre-fusion model integrating intratumoral and habitat radiomics features based on the AdaBoost classifier (Pre_AdaBoost model) demonstrated the highest predictive performance compared to the intratumoral model, habitat model, and post-fusion model. In the training and validation sets, the Pre_AdaBoost model achieved the highest diagnostic capabilities, with area under the ROC curve (AUC), sensitivities, and specificities of 0.916 [95% confidence interval (CI): 0.856 to 0.977], 88.5%, 77.8% and 0.831 (95% CI: 0.691 to 0.972), 91.7%, 57.9%, respectively. The AUC values were 0.916 in the training set and 0.831 in the test set, indicating high clinical net benefit.Conclusions The combined model integrating habitat radiomics and intratumoral radiomics based on DCE-MRI quantitative parametric maps demonstrated significant value in predicting LVSI in cervical cancer, potentially facilitating personalized treatment decisions.
[Keywords] cervical cancer；habitat imaging；lymph-vascular space invasion；radiomics；dynamic contrast-enhanced magnetic resonance imaging

LI Feixiang1   SUN Yun1   GONG Juntao1   HUANG Gang2*  

1 The First Clinical Medical College of Gansu University of Chinese Medicine, Lanzhou 730000, China

2 Department of Radiology, Gansu Provincial Hospital, Lanzhou 730000, China

Corresponding author: HUANG G, E-mail: huang_g2024@163.com

Conflicts of interest   None.

Received  2025-05-28
Accepted  2025-09-10
DOI: 10.12015/issn.1674-8034.2025.10.014
Cite this article as: LI F X, SUN Y, GONG J T, et al. Study on predicting LVSI status in preoperative cervical cancer patients without lymph node metastasis using habitat radiomics based on DCE-MRI quantitative parametric maps[J]. Chin J Magn Reson Imaging, 2025, 16(10): 89-97. DOI:10.12015/issn.1674-8034.2025.10.014.

[1]
JIN X Y, ZHANG Y, ZHONG Y X, et al. Trends and prediction analysis of the disease burden of malignant tumors in China and the world from 1990 to 2021[J]. Med J Chin People's Liberation Army, 2025, 50(5): 513-522. DOI: 10.11855/j.issn.0577-7402.2096.2025.0401.
[2]
NGUYEN T, NOUGARET S, CASTILLO P, et al. Cervical cancer in the pregnant population[J]. Abdom Radiol (NY), 2023, 48(5): 1679-1693. DOI: 10.1007/s00261-023-03836-x.
[3]
RAFIEE A, MOHAMMADIZADEH F. Association of lymphovascular space invasion (LVSI) with histological tumor grade and myometrial invasion in endometrial carcinoma: A review study[J/OL]. Adv Biomed Res, 2023, 12: 159 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/37564444/. DOI: 10.4103/abr.abr_52_23.
[4]
QIU H F, WANG M, WANG S W, et al. Integrating MRI-based radiomics and clinicopathological features for preoperative prognostication of early-stage cervical adenocarcinoma patients: in comparison to deep learning approach[J/OL]. Cancer Imaging, 2024, 24(1): 101 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/39090668/. DOI: 10.1186/s40644-024-00747-y.
[5]
DENG Y R, CHEN X J, XU C Q, et al. A preoperative nomogram predicting risk of lymph node metastasis for early-stage cervical cancer[J/OL]. BMC Womens Health, 2023, 23(1): 568 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/37924031/. DOI: 10.1186/s12905-023-02726-0.
[6]
MARTH C, LANDONI F, MAHNER S, et al. Cervical cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up[J/OL]. Ann Oncol, 2018, 29(Suppl 4): iv262 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/29741577/. DOI: 10.1093/annonc/mdy160.
[7]
ABU-RUSTUM N R, YASHAR C M, AREND R, et al. NCCN guidelines® insights: cervical cancer, version 1.2024[J]. J Natl Compr Canc Netw, 2023, 21(12): 1224-1233. DOI: 10.6004/jnccn.2023.0062.
[8]
BIDUS M A, CAFFREY A S, YOU W B, et al. Cervical biopsy and excision procedure specimens lack sufficient predictive value for lymph-vascular space invasion seen at hysterectomy for cervical cancer[J/OL]. Am J Obstet Gynecol, 2008, 199(2): 151.e1-151.e4 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/18674657/. DOI: 10.1016/j.ajog.2008.02.017.
[9]
WANG Y, CHEN X, PU H, et al. Roles of DWI and T2-weighted MRI volumetry in the evaluation of lymph node metastasis and lymphovascular invasion of stage IB-IIA cervical cancer[J]. Clin Radiol, 2022, 77(3): 224-230. DOI: 10.1016/j.crad.2021.12.011.
[10]
YANG C S, HASSAN H ABU, OMAR N F, et al. The value of amide proton transfer imaging in predicting parametrial invasion and lymph-vascular space invasion of cervical cancer[J/OL]. Magn Reson Imag, 2025, 116: 110282 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/39603395/. DOI: 10.1016/j.mri.2024.110282.
[11]
XU C, YU Y, LI X R, et al. Value of integrated PET-IVIM MRI in predicting lymphovascular space invasion in cervical cancer without lymphatic metastasis[J]. Eur J Nucl Med Mol Imaging, 2021, 48(9): 2990-3000. DOI: 10.1007/s00259-021-05208-3.
[12]
CHENG J M, LUO W X, TAN B G, et al. Whole-tumor histogram analysis of apparent diffusion coefficients for predicting lymphovascular space invasion in stage IB-IIA cervical cancer[J/OL]. Front Oncol, 2023, 13: 1206659 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/37404753/. DOI: 10.3389/fonc.2023.1206659.
[13]
SONG Q L, TIAN S F, MA C J, et al. Amide proton transfer weighted imaging combined with dynamic contrast-enhanced MRI in predicting lymphovascular space invasion and deep stromal invasion of IB1-IIA1 cervical cancer[J/OL]. Front Oncol, 2022, 12: 916846 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/36172148/. DOI: 10.3389/fonc.2022.916846.
[14]
LI S J, LIU J, ZHANG W H, et al. T1 mapping and multimodel diffusion-weighted imaging in the assessment of cervical cancer: a preliminary study[J/OL]. Br J Radiol, 2023, 96(1148): 20220952 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/37183908/. DOI: 10.1259/bjr.20220952.
[15]
LIU L Y, WANG S, YU T, et al. Value of diffusion-weighted imaging in preoperative evaluation and prediction of postoperative supplementary therapy for patients with cervical cancer[J/OL]. Ann Transl Med, 2022, 10(2): 120 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/35282103/. DOI: 10.21037/atm-21-5319.
[16]
GATENBY R A, GROVE O, GILLIES R J. Quantitative imaging in cancer evolution and ecology[J]. Radiology, 2013, 269(1): 8-15. DOI: 10.1148/radiol.13122697.
[17]
WU Y, WANG S X, CHEN Y Q, et al. A multicenter study on preoperative assessment of lymphovascular space invasion in early-stage cervical cancer based on multimodal MR radiomics[J]. J Magn Reson Imaging, 2023, 58(5): 1638-1648. DOI: 10.1002/jmri.28676.
[18]
HUANG G, CUI Y Q, WANG P, et al. Multi-parametric magnetic resonance imaging-based radiomics analysis of cervical cancer for preoperative prediction of lymphovascular space invasion[J/OL]. Front Oncol, 2022, 11: 663370 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/35096556/. DOI: 10.3389/fonc.2021.663370.
[19]
CUI L P, YU T, KAN Y Y, et al. Multi-parametric MRI-based peritumoral radiomics on prediction of lymph-vascular space invasion in early-stage cervical cancer[J]. Diagn Interv Radiol, 2022, 28(4): 312-321. DOI: 10.5152/dir.2022.20657.
[20]
DU W, WANG Y, LI D D, et al. Preoperative prediction of lymphovascular space invasion in cervical cancer with radiomics-based nomogram[J/OL]. Front Oncol, 2021, 11: 637794 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/34322375/. DOI: 10.3389/fonc.2021.637794.
[21]
NAPEL S, MU W, JARDIM-PERASSI B V, et al. Quantitative imaging of cancer in the postgenomic era: Radio(geno)mics, deep learning, and habitats[J]. Cancer, 2018, 124(24): 4633-4649. DOI: 10.1002/cncr.31630.
[22]
CAI Z P, XU Z Y, CHEN Y F, et al. Multiparametric MRI subregion radiomics for preoperative assessment of high-risk subregions in microsatellite instability of rectal cancer patients: a multicenter study[J]. Int J Surg, 2024, 110(7): 4310-4319. DOI: 10.1097/JS9.0000000000001335.
[23]
YANG Y, HAN Y, ZHAO S J, et al. Spatial heterogeneity of edema region uncovers survival-relevant habitat of Glioblastoma[J/OL]. Eur J Radiol, 2022, 154: 110423 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/35777079/. DOI: 10.1016/j.ejrad.2022.110423.
[24]
LI S L, DAI Y M, CHEN J Y, et al. MRI-based habitat imaging in cancer treatment: current technology, applications, and challenges[J/OL]. Cancer Imaging, 2024, 24(1): 107 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/39148139/. DOI: 10.1186/s40644-024-00758-9.
[25]
WANG W, FAN X F, YANG J, et al. Preliminary MRI study of extracellular volume fraction for identification of lymphovascular space invasion of cervical cancer[J]. J Magn Reson Imaging, 2023, 57(2): 587-597. DOI: 10.1002/jmri.28423.
[26]
GAO J, LI P, CHEN Z K, et al. A survey on deep learning for multimodal data fusion[J]. Neural Comput, 2020, 32(5): 829-864. DOI: 10.1162/neco_a_01273.
[27]
KOCH J, WOLF S, BEYERER J. A transformer-based late-fusion mechanism for fine-grained object recognition in videos[C]//2023 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW). January 3-7, 2023, Waikoloa, HI, USA. IEEE, 2023: 1-10. DOI: 10.1109/WACVW58289.2023.00015.
[28]
GADZICKI K, KHAMSEHASHARI R, ZETZSCHE C. Early vs late fusion in multimodal convolutional neural networks[C]//2020 IEEE 23rd International Conference on Information Fusion (FUSION). July 6-9, 2020, Rustenburg, South Africa. IEEE, 2020: 1-6.
[29]
KLIONSKY D J, ABDEL-AZIZ A K, ABDELFATAH S, et al. Guidelines for the use and interpretation of assays for monitoring autophagy (4th edition)1[J]. Autophagy, 2021, 17(1): 1-382. DOI: 10.1080/15548627.2020.1797280.
[30]
CHEVRIER S, LEVINE J H, ZANOTELLI V R T, et al. An immune atlas of clear cell renal cell carcinoma[J/OL]. Cell, 2017, 169(4): 736-749.e18 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/28475899/. DOI: 10.1016/j.cell.2017.04.016.
[31]
HORNING S J. A new cancer ecosystem[J/OL]. Science, 2017, 355(6330): 1103 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/28302798/. DOI: 10.1126/science.aan1295.
[32]
SINGH Y, FARRELLY C M, HATHAWAY Q A, et al. Topological data analysis in medical imaging: current state of the art[J/OL]. Insights Imaging, 2023, 14(1): 58 [2025-05-27]. https://pubmed.ncbi.nlm.nih.gov/37005938/. DOI: 10.1186/s13244-023-01413-w.

PREV Super-resolution reconstruction technique enhances the diagnostic efficacy of deep learning-based prediction of lymphvascular space invasion in endometrial cancer
NEXT Analysis of patellofemoral joint cartilage injuries and influencing factors in amateur marathon runners based on MRI and X-ray
  


LINK


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