Deep Segmentation Feature-Based Radiomics Improves Recurrence Prediction of Hepatocellular Carcinoma

作     者：Jifei Wang Dasheng Wu Meili Sun Zhenpeng Peng Yingyu Lin Hongxin Lin Jiazhao Chen Tingyu Long Zi-Ping Li Chuanmiao Xie Bingsheng Huang Shi-Ting Feng 

作者机构：Department of RadiologyThe First Affiliated HospitalSun Yat-sen UniversityGuangzhouChina Medical AI LabSchool of Biomedical EngineeringHealth Science CentreShenzhen UniversityShenzhenChina Shenzhen-Hong Kong Institute of Brain Science-Shenzhen Fundamental Research InstitutionsShenzhenChina Department of Medical Imaging and Interventional RadiologySun Yat-sen University Cancer CenterState Key Laboratory of Oncology in South ChinaCollaborative Innovation Center for Cancer MedicineGuangzhouChina 

出 版 物：《Biomedical Engineering Frontiers》 (生物医学工程前沿（英文）)

年 卷 期：2022年第3卷第1期

页      面：126-137页

核心收录：

学科分类：1002[医学-临床医学] 100214[医学-肿瘤学] 10[医学] 

基　　金：funded by the National Natural Science Foundation of China (81771908,81971684) Natural Science Foundation of Guangdong Province,PR China (2020A1515010571) Medical Research Foundation of Guangdong Province,PR China (A2019092) Shenzhen-Hong Kong Institute of Brain Science-Shenzhen Fundamental Research Institutions (2019SHIBS0003) Shenzhen University Top Ranking Project (860/000002100108) Nature Science Foundation of Shenzhen (JCYJ20200109114014533) 

主　　题：Carcinoma signature Deep 

摘      要：Objective and Impact *** study developed and validated a deep semantic segmentation feature-based radiomics(DSFR)model based on preoperative contrast-enhanced computed tomography(CECT)combined with clinical information to predict early recurrence(ER)of single hepatocellular carcinoma(HCC)after curative *** prediction is of great significance to the therapeutic decision-making and surveillance strategy of *** prediction is important for ***,it cannot currently be adequately ***,208 patients with single HCC after curative resection were retrospectively recruited into a model-development cohort(n=180)and an independent validation cohort(n=28).DSFR models based on different CT phases were *** optimal DSFR model was incorporated with clinical information to establish a DSFR-C *** integrated nomogram based on the Cox regression was *** DSFR signature was used to stratify high-and low-risk ER ***.A portal phase-based DSFR model was selected as the optimal model(area under receiver operating characteristic curve(AUC):development cohort,0.740;validation cohort,0.717).The DSFR-C model achieved AUCs of 0.782 and 0.744 in the development and validation cohorts,*** the development and validation cohorts,the integrated nomogram achieved C-index of 0.748 and 0.741 and time-dependent AUCs of 0.823 and 0.822,respectively,for recurrence-free survival(RFS)*** RFS difference between the risk groups was statistically significant(P0.0001 and P=0.045 in the development and validation cohorts,respectively).***-based DSFR can predict ER in single HCC after curative resection,and its combination with clinical information further improved the performance for ER prediction.