European journal of surgical oncology : the journal of the European Society of Surgical Oncology and the British Association of Surgical Oncology · 2026
CT Habitat-peritumoral fusion radiomics predicts aggressiveness and prognosis in pheochromocytomas and paragangliomas to inform intensified surgical strategies: A multicenter study.
СтатьяМногоцентровое исследованиеНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Zhou Yongjie, Zhang Zhilong, Zhou Keyang, Tan Yongming, Zhao Jinhong, Zhong Linhua, Wei Xiang, Zhu Jianhua, Li Zihan, Du Jingjing, Liu Lan, Fang Lei
Аннотация
BACKGROUND: This study aimed to develop a dual-domain radiomics framework integrating probability-driven high-risk habitats and peritumoral microenvironmental features to accurately predict the aggressiveness of pheochromocytomas and paragangliomas (PPGLs).
METHODS: This retrospective study included 356 patients with abdominal PPGLs from four institutions, who were divided into a training set (n = 182) and two external test sets (n = 70, n = 104). Radiomic features were extracted from the whole-tumor and 1-mm peritumoral regions on contrast-enhanced CT images. Probability-driven habitat analysis utilizing a Support Vector Machine and K-means clustering was implemented to segment high-risk subregions. Following rigorous cascaded feature selection, a Random Forest-based fusion model was constructed to integrate key habitat and peritumoral features for predicting aggressiveness. Additionally, SHapley Additive exPlanations (SHAP) analysis was employed to evaluate feature importance and model interpretability.
RESULTS: The habitat-peritumoral fusion model demonstrated superior predictive performance and robustness compared to traditional single-modality models, achieving area under the receiver operating characteristic curve values of 0.884 and 0.854 in external test sets 1 and 2, respectively. The model maintained excellent diagnostic efficacy across both >6 cm and ≤6 cm tumor subgroups. Furthermore, the model-derived risk score successfully stratified metastasis-free survival in the overall cohort (P < 0.001) and the clinically ambiguous ≤6 cm subgroup (P = 0.003), serving as an independent prognostic factor (HR = 10.73, P = 0.028).
CONCLUSION: The fusion model is a robust, non-invasive tool for preoperatively identifying high-risk PPGLs. Particularly for tumors ≤6 cm where decision-making is challenging, it provides objective evidence to individualize surgical strategies and stratify postoperative surveillance.
METHODS: This retrospective study included 356 patients with abdominal PPGLs from four institutions, who were divided into a training set (n = 182) and two external test sets (n = 70, n = 104). Radiomic features were extracted from the whole-tumor and 1-mm peritumoral regions on contrast-enhanced CT images. Probability-driven habitat analysis utilizing a Support Vector Machine and K-means clustering was implemented to segment high-risk subregions. Following rigorous cascaded feature selection, a Random Forest-based fusion model was constructed to integrate key habitat and peritumoral features for predicting aggressiveness. Additionally, SHapley Additive exPlanations (SHAP) analysis was employed to evaluate feature importance and model interpretability.
RESULTS: The habitat-peritumoral fusion model demonstrated superior predictive performance and robustness compared to traditional single-modality models, achieving area under the receiver operating characteristic curve values of 0.884 and 0.854 in external test sets 1 and 2, respectively. The model maintained excellent diagnostic efficacy across both >6 cm and ≤6 cm tumor subgroups. Furthermore, the model-derived risk score successfully stratified metastasis-free survival in the overall cohort (P < 0.001) and the clinically ambiguous ≤6 cm subgroup (P = 0.003), serving as an independent prognostic factor (HR = 10.73, P = 0.028).
CONCLUSION: The fusion model is a robust, non-invasive tool for preoperatively identifying high-risk PPGLs. Particularly for tumors ≤6 cm where decision-making is challenging, it provides objective evidence to individualize surgical strategies and stratify postoperative surveillance.