Surgical oncology · 2026
Ultrasound texture analysis of primary tumor for predicting axillary lymph node metastasis in patients with clinical T1-T2 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.
Liu Xuejing, Hao Yujuan, Sun Shumeng, Xu Yilin, Lu Hong
Аннотация
OBJECTIVE: To evaluate the utility of ultrasound-based texture features from primary T1-T2 breast cancer lesions for predicting axillary lymph node metastasis (ALNM), and to compare its performance with that of expert sonographic assessment.
METHODS: This retrospective study included 198 patients with pathologically confirmed T1-T2 breast cancer who underwent standardized preoperative ultrasound examinations. Randomly divided into training set (n = 159) and testing sets (n = 39) at a 4:1 ratio. Texture features were extracted from the largest cross-sectional image of each lesion. Texture features were extracted from the largest cross-sectional image of each lesion to establish and validate multivariable logistic regression nomogram models as well as MLP models. Its diagnostic performance was compared with that of artificial assessment using ROC curve analysis.
RESULTS: In the testing set, the AUC values for the artificial assessment model, the nomogram model, and the MLP model were 0.638 (95% CI: 0.487-0.789), 0.774 (95% CI: 0.614-0.933), and 0.842 (95% CI: 0.713-0.971), respectively. Compared to the artificial assessment model, the MLP model demonstrated significantly better performance (DeLong test P-value <0.05). When compared with the nomogram model, the MLP model showed higher specificity (0.850 and 0.900) and negative predictive value (0.773 and 0.783). Key predictive features included measures of entropy and texture heterogeneity.
CONCLUSION: Texture analysis of primary tumor ultrasound images demonstrates good performance in predicting ALNM compared to artificial assessment, and can further improve the accuracy of preoperative evaluation, especially in settings where radiological expertise is limited.
METHODS: This retrospective study included 198 patients with pathologically confirmed T1-T2 breast cancer who underwent standardized preoperative ultrasound examinations. Randomly divided into training set (n = 159) and testing sets (n = 39) at a 4:1 ratio. Texture features were extracted from the largest cross-sectional image of each lesion. Texture features were extracted from the largest cross-sectional image of each lesion to establish and validate multivariable logistic regression nomogram models as well as MLP models. Its diagnostic performance was compared with that of artificial assessment using ROC curve analysis.
RESULTS: In the testing set, the AUC values for the artificial assessment model, the nomogram model, and the MLP model were 0.638 (95% CI: 0.487-0.789), 0.774 (95% CI: 0.614-0.933), and 0.842 (95% CI: 0.713-0.971), respectively. Compared to the artificial assessment model, the MLP model demonstrated significantly better performance (DeLong test P-value <0.05). When compared with the nomogram model, the MLP model showed higher specificity (0.850 and 0.900) and negative predictive value (0.773 and 0.783). Key predictive features included measures of entropy and texture heterogeneity.
CONCLUSION: Texture analysis of primary tumor ultrasound images demonstrates good performance in predicting ALNM compared to artificial assessment, and can further improve the accuracy of preoperative evaluation, especially in settings where radiological expertise is limited.