(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: develop and validate a nomogram for the prediction of local recurrence after radical surgery for esophageal squamous cell carcinoma (ESCC) and to provide selective postoperative adjuvant radiotherapy.
Methods: total of 197 patients were enrolled in this study (training group: n=137; test group: n=60). The two-person blindly outlines the region of interest on the preoperative enhanced CT and extracts radiomics features. The features with good reproducibility were screened with an intra-class correlation coefficient greater than 0.9. The LASSO method was used for radiomics signature establishment. Multivariate analysis was used to establish a predictive nomogram in conjunction with radiomics signatures and clinicopathological factors. The calibration curve and the receiver operating characteristic (ROC) curve were used to evaluate the performance of the nomogram.
Results: radiomics signature and pathological N stage demonstrated a significant association with the occurrence of postoperative local events and finally entered the predicted nomogram. The calibration curve showed that the nomogram prediction results were consistent with the real situation (C-index: training group 0.789, test group 0.768). The ROC curve showed that the use of nomogram for local event prediction was significantly better than the pathological N stage alone (AUC: training group 0.789: 0.641, test group: 0.768: 0.664).
Conclusion: predictive nomogram constructed by combined CT radiomics signature and pathological N staging can improve the predictive accuracy of local recurrence, and provide a basis for selective postoperative adjuvant radiotherapy.
Keywords
CT, Quantitative Imaging, Chest Radiography
Taxonomy
Contact Email
