Share：
Share this content in WeChat
X
[Chinese][PDF]763153
• MRI of Uterine Diseases •
Predicting the histological classification of cervical squamous cell carcinoma based on multi-sequence MRI radiomics model
LI Xiaoran  GUO Yan  XU Chen  KANG Ye  SUN Hongzan 

Cite this article as: Li XR, Guo Y, Xu C, et al. Predicting the histological classification of cervical squamous cell carcinoma based on multi-sequence MRI radiomics model. Chin J Magn Reson Imaging, 2020, 11(7): 481-486. DOI:10.12015/issn.1674-8034.2020.07.001.


[Abstract] Objective: To investigate the value of radiomics model based on no-contrast-enhanced multi-sequence MRI in predicting the histological classification of cervical squamous cell carcinoma.Materials and Methods: One hundred and three patients with cervical squamous cell carcinoma histological confirmed underwent conventional non-contrast-enhanced magnetic resonance imaging examination and were randomly divided into training dataset and test dataset according to 3︰1. The tumor region of interest was delineated in each sequence image, and the radiomics features of tumours based on T1WI, T2WI and FS-T2WI were extracted and filtered. The naive Bayesian algorithm was used to separately create the model by T1WI, T2WI, FS-T2WI and combined all features and to perform cross-validation in training dataset. Each model was tested and evaluated in testing dataset.Results: The AUC value of each model predicting non-keratinized results in the training dataset was 0.718, 0.705, 0.756, and 0.863, and which were statistically significant (P<0.001). In the testing dataset, combined model predicting non-keratinized results are the best. The AUC value was 0.860, P=0.003, the accuracy rate was 0.720, and the recall rate is 0.900. The pairwise comparison of the combined model with T1 model prediction in the delong test had a statistically significant difference in the predicting non-keratinized ROC curve (P<0.05).Conclusions: The radiomics model based on MRI of multi-sequence can predict the non-keratinized type of cervical squamous cell carcinoma, and the prediction model combined with multi-sequences predicted greater.
[Keywords] magnetic resonance imaging；uterine cervical neoplasms；pathology；radiomics；prediction；histological classification；machine Learning

LI Xiaoran Department of Radiology, Shengjing Hospital of China Medical University, Shenyang 110004, China

GUO Yan GE Healthcare, Shenyang 110000, China

XU Chen Department of Radiology, Shengjing Hospital of China Medical University, Shenyang 110004, China

KANG Ye Department of Pathology, Shengjing Hospital of China Medical University, Shenyang 110004, China

SUN Hongzan* Department of Radiology, Shengjing Hospital of China Medical University, Shenyang 110004, China

*Correspondence to: Sun HZ, E-mail: sunhongzan@126.com

Conflicts of interest   None.

ACKNOWLEDGMENTS  Liaoning Province Natural Science Foundation Project No. 2019-MS-373
Received  2019-12-15
Accepted  2020-05-21
DOI: 10.12015/issn.1674-8034.2020.07.001
Cite this article as: Li XR, Guo Y, Xu C, et al. Predicting the histological classification of cervical squamous cell carcinoma based on multi-sequence MRI radiomics model. Chin J Magn Reson Imaging, 2020, 11(7): 481-486. DOI:10.12015/issn.1674-8034.2020.07.001.

[1]
Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, 2012. CA Cancer J Clin, 2015, 65(2): 87-108.
[2]
Kumar S, Shah JP, Bryant CS, et al. Prognostic significance of keratinization in squamous cell cancer of uterine cervix: a population based study. Arch Gynecol Obstet, 2009, 280(1): 25-32.
[3]
Liu Y, Zhang Y, Cheng R, et al. Radiomics analysis of apparent diffusion coefficient in cervical cancer: a preliminary study on histological grade evaluation. J Magn Reson Imaging, 2019, 49(1): 280-290.
[4]
Tsujikawa T, Rahman T, Yamamoto M, et al. (18)F-FDG PET radiomics approaches: comparing and clustering features in cervical cancer. Ann Nucl Med, 2017, 31(9): 678-685.
[5]
Wu Q, Shi D, Dou S, et al. Radiomics analysis of multiparametric MRI evaluates the pathological features of cervical squamous cell carcinoma. J Magn Reson Imaging, 2019, 49(4): 1141-1148.
[6]
Demsar JC, Erjavec A, Gorup C, et al. Orange: data mining toolbox in python. J Mach Learn Res. 2013, 14: 2349-2353.
[7]
张庆，徐香玖，周星，等. 3.0 T MR动态对比增强对宫颈癌组织学特性和临床分期的分析．临床放射学杂志, 2015, 34(10): 1607-1610.
[8]
叶晓华，周诚，王宏，等. MR体素内不相干运动成像评价宫颈癌组织学特征的初步研究．临床放射学杂志, 2016, 35(7): 1048-1052.
[9]
杨蔚强，田海萍，陈兵，等．标准化ADC评估宫颈癌的组织学类型及分化程度．中国医学计算机成像杂志, 2017, 23(6): 521-525.
[10]
Meng N, Wang J, Sun J, et al. Using amide proton transfer to identify cervical squamous carcinoma/adenocarcinoma and evaluate its differentiation grade. Magn Reson Imaging, 2019, 61: 9-15.
[11]
Shen WC, Chen SW, Liang JA, et al. [18]Fluorodeoxyglucose Positron Emission Tomography for the Textural Features of Cervical Cancer Associated with Lymph Node Metas0tasis and Histological Type. Eur J Nucl Med Mol Imaging, 2017, 44(10): 1721-1731.

PREV Application progress of radiomics in cervical cancer
NEXT Clinical value of reduced field-of-view diffusion-weighted imaging in cervical cancer
  


LINK


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