Ultrasound in medicine & biology · 2026
Integrative Multi-scale Radiomics and Interpretable AI for Prediction and Burden Stratification of Axillary Lymph Node Metastasis in Breast Cancer.
СтатьяНазвание 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.
Yang Xinying, Cai Mengjun, Pan Jiazhen, Du Liwen, Li Cuiying, Zha Hailing, Du Yu
Аннотация
OBJECTIVE: To develop and validate multi-scale radiomics for predicting axillary lymph node metastasis (ALNM) and burden stratification in breast cancer (BC) patients.
METHODS: A total of 475 consecutive women with pathologically diagnosed BC from three centers were retrospectively included. Ultrasound (US) radiomics signature capturing tumor heterogeneity were extracted from the intratumoral region (Intra_RS), peritumoral region (Peri_RS), and subregional intratumoral habitats (ITH). Dimensionality reduction and feature selection of radiomics features were conducted using the least absolute shrinkage and selection operator. Model performance was evaluated using receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA). Model interpretability was achieved using SHAP analysis.
RESULTS: In the training cohort, multivariable analysis identified clinical T stage, lymphovascular invasion, margin status, Intra_RS, and ITH as independent predictors of ALNM. ITH presented an area under curve (AUC) of 0.87 (95% CI, 0.82-0.91) outperformed the clinical (AUC 0.71), Intra_RS (AUC 0.80), and Peri_RS (AUC 0.79) models (both p < 0.05, DeLong test). The combined model outperformed other models in both the internal (AUC 0.91, 95% confidence intervals [CI] 0.86-0.96) and external validation (AUC 0.87, 95% CI, 0.81-0.94) cohorts (both p < 0.05). In the external validation cohort, DCA indicated favorable clinical utility. SHAP analysis revealed that ITH contributed most substantially to model predictions (value = 0.420). For burden stratification, the ITH achieved the highest discriminatory performance in the external validation cohort, yielding a macro-average accuracy of 0.72 and a micro-average accuracy of 0.82.
CONCLUSION: The multi-scale radiomics model demonstrates strong potential for noninvasively predicting ALNM and burden stratification in BC patients.
METHODS: A total of 475 consecutive women with pathologically diagnosed BC from three centers were retrospectively included. Ultrasound (US) radiomics signature capturing tumor heterogeneity were extracted from the intratumoral region (Intra_RS), peritumoral region (Peri_RS), and subregional intratumoral habitats (ITH). Dimensionality reduction and feature selection of radiomics features were conducted using the least absolute shrinkage and selection operator. Model performance was evaluated using receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA). Model interpretability was achieved using SHAP analysis.
RESULTS: In the training cohort, multivariable analysis identified clinical T stage, lymphovascular invasion, margin status, Intra_RS, and ITH as independent predictors of ALNM. ITH presented an area under curve (AUC) of 0.87 (95% CI, 0.82-0.91) outperformed the clinical (AUC 0.71), Intra_RS (AUC 0.80), and Peri_RS (AUC 0.79) models (both p < 0.05, DeLong test). The combined model outperformed other models in both the internal (AUC 0.91, 95% confidence intervals [CI] 0.86-0.96) and external validation (AUC 0.87, 95% CI, 0.81-0.94) cohorts (both p < 0.05). In the external validation cohort, DCA indicated favorable clinical utility. SHAP analysis revealed that ITH contributed most substantially to model predictions (value = 0.420). For burden stratification, the ITH achieved the highest discriminatory performance in the external validation cohort, yielding a macro-average accuracy of 0.72 and a micro-average accuracy of 0.82.
CONCLUSION: The multi-scale radiomics model demonstrates strong potential for noninvasively predicting ALNM and burden stratification in BC patients.