AUTHOR=Xue Beihui , Jiang Jia , Chen Lei , Wu Sunjie , Zheng Xuan , Zheng Xiangwu , Tang Kun 

TITLE=Development and Validation of a Radiomics Model Based on 18F-FDG PET of Primary Gastric Cancer for Predicting Peritoneal Metastasis

JOURNAL=Frontiers in Oncology

VOLUME=Volume 11 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/oncology/articles/10.3389/fonc.2021.740111

DOI=10.3389/fonc.2021.740111

ISSN=2234-943X

ABSTRACT=Abstract
Objectives: This study aimed to develop a preoperative positron emission tomography (PET)-based radiomics model for predicting peritoneal metastasis (PM) of gastric cancer（GC）.
Methods: The preoperative fluorine-18-fludeoxyglucose (18F-FDG) PET images of 355 patients (109PM+, 246PM-) confirmed by histopathological examination were retrospectively reviewed. Patients were randomly divided into training set and validation set according to 7:3 ratio. Radiomics features and relevant data were extracted from PET images. The radiomics scores based on radiomics features, were built by logistic regression after using maximum relevance and minimum redundancy (mRMR) and least shrinkage and selection operator (LASSO) method. The performance of the models was determined by their discrimination, calibration, and clinical usefulness in the training and validation sets.
Results: After dimensionality reduction,12 radiomics feature parameters were obtained to construct radiomics signatures. The multivariate logistic regression analysis showed that only carbohydrate antigen 125(CA125), maximum standardized uptake value (SUVmax) and the radiomics signature had statistically significant differences between patients with and without PM (P<0.05). A radiomics model was developed based on the logistic analyses with AUC of 0.86 in the training set and 0.87 in the validation set. The clinical prediction model based on CA125 was 0.76 in the training set and 0.69 in the validation set. The comprehensive model contained rad-score and clinical factor (CA125) as well as metabolic parameter (SUVmax) showed the best performance with AUC of 0.90 in the training set and 0.88 in the validation set. The calibration curve demonstrated good consistency between the nomogram-predicted probability of PM and the actual rate. The Decision curve analysis (DCA) also confirmed the clinical utility of the nomogram.
Conclusions: The comprehensive model based on rad-score and other factors (SUVmax, CA125) can make a meaningful contribution for predicting PM status in GC patients preoperatively.