CT-based radiomics analysis for evaluating the differentiation degree of esophageal squamous carcinoma.
10.11817/j.issn.1672-7347.2019.03.004
- Author:
Leishu CHENG
1
;
Lei WU
1
;
Shuting CHEN
2
;
Weitao YE
2
;
Zaiyi LIU
3
,
4
;
Changhong LIANG
3
,
4
Author Information
1. School of Medicine, South China University of Technology, Guangzhou 510006, China.
2. Department of Radiology, Guangdong Provincial People's Hospital, Guangzhou 510080, China.
3. School of Medicine, South China University of Technology, Guangzhou 510006
4. Department of Radiology, Guangdong Provincial People's Hospital, Guangzhou 510080, China.
- Publication Type:Journal Article
- MeSH:
Carcinoma, Squamous Cell;
Esophageal Neoplasms;
Humans;
ROC Curve;
Retrospective Studies;
Tomography, X-Ray Computed
- From:
Journal of Central South University(Medical Sciences)
2019;44(3):251-256
- CountryChina
- Language:Chinese
-
Abstract:
To build a CT-based radiomics predictive mode to evaluate the differentiation degree of the esophageal squamous carcinoma.
Methods: A total of 160 patients with surgical pathology, complete clinical data and chest CT scanning before operation were retrospectively collected from January 2008 to August 2016. All patients were assigned randomly to a primary data set and an independent validation. Texture analysis was performed on CT images, while the carcinomas were performed by manual segmentation to extract the radiomics features. Radiomics features were extracted and 9 radiomics signatures were finally selected after dimension reduction. Radiomics features were extracted and established via Matlab. Multivariable logistic regression analysis was performed to build the predictive model. A 10-fold cross-validation was used for selecting parameters in the least absolute shrinkage and selection operator (LASSO) model by minimum criteria. The receiver operating characteristic (ROC) curves and areas under ROC curve (AUC) were used to compare the model performance in the primary validation and the independent validation for evaluating the differentiation degree of esophageal squamous carcinoma.
Results: Radiomics signature showed great effect in discriminating primary data set and independent validation. The predictive model had a good performance in primary data set. The AUC was 0.791, the sensitivity was 81.6%, and specificity was 72.3%. In the independent validation, the AUC was 0.757, the sensitivity was 70.0%, and the specificity was 73.0%.
Conclusion: The predictive model can be used for evaluating the differentiation degree of esophageal squamous carcinoma efficiently, which can be helpful to clinicians in diagnosis and choice of treatment for esophageal squamous carcinoma.
