KEY TAKEAWAYS
- The study aimed to investigate CECT radiomics for preoperative FIGO staging prediction in EOC and validate model stability with external data.
- The integrated model provided a potential method as a non-invasive diagnostic tool for preoperative FIGO staging in EOC.
Accurate preoperative prediction of the International Federation of Gynecology and Obstetrics (FIGO) stage in epithelial ovarian cancer (EOC) guides treatment planning.
Yinping Leng and the team conducted a study that aimed to investigate contrast-enhanced CT (CECT) radiomics for preoperative FIGO staging prediction in EOC and validate model stability with external data.
About 201 EOC patients were enrolled from three centers, with 106 in the training cohort and 46 and 49 in the internal and external validation cohorts, respectively. The least absolute shrinkage and selection operator (LASSO) regression algorithm was applied for radiomics feature selection.
Five machine learning algorithms logistic regression, support vector machine, light gradient boosting machine (LightGBM), random forest and decision tree, were utilized for radiomics model development. The top-performing algorithm was selected to build the radiomics, clinical, and combined models.
Receiver operating characteristic analysis assessed diagnostic performance, with AUC comparisons conducted using the Delong or F-test. The LASSO algorithm retained seven optimal radiomics features. Among the 5 radiomics models, the LightGBM model showed notable prediction efficiency and robustness with AUCs training, internal valudation and external validation cohort were 0.83, 0.80 , and 0.68 respectively.
Multivariate logistic regression analysis revealed carcinoma antigen 125 and tumor location as independent predictors for EOC FIGO staging. The combined model exhibited superior diagnostic efficiency, with AUCs of 0.95, 0.83, and 0.79 in the training, internal validation, and external validation cohorts, respectively. The F-test demonstrated significantly higher AUC values for the combined model than the radiomics model in the training cohort (P < 0.001).
The study concluded that integrating clinical characteristics and radiomics features into a combined model offers promise as a non-invasive adjunctive diagnostic tool for preoperative evaluation of EOC FIGO staging status. This approach has the potential to improve clinical decision-making and enhance patient outcomes.
Authors affirm that they did not receive any funds, grants, or other support while preparing this manuscript.
Souce: https://pubmed.ncbi.nlm.nih.gov/38448945/
Leng Y, Kan A, Wang X, et al. (2024). “Contrast-enhanced CT radiomics for preoperative prediction of stage in epithelial ovarian cancer: a multicenter study.’’ BMC Cancer. 2024 Mar 6;24(1):307. doi: 10.1186/s12885-024-12037-8. PMID: 38448945; PMCID: PMC10916071.
