文献速递乳腺MR

基于磁共振成像与活检全切片图像机器学习预测乳腺癌新辅助化疗后病理完全缓解的价值:一项双中心研究

The value of machine learning based on magnetic resonance imaging (MRI) and biopsy whole-slide image to predict pathological complete response to breast cancer after neoadjuvant chemotherapy: a two-centre study.
PubMed 文献 · Clin Radiol · 2025-06-04

该双中心回顾性研究纳入331例经病理确诊为浸润性乳腺癌并接受新辅助化疗的患者,基于MRI影像组学特征、活检全切片图像深度学习病理特征及临床病理特征构建联合列线图模型预测病理完全缓解。联合模型在训练、测试及外部验证队列的AUC分别为0.95、0.84和0.83,DeLong检验显示联合模型与其他单一模型AUC差异有统计学意义(p<0.05)。

要点

  • 联合列线图模型在外部验证队列预测病理完全缓解的AUC为0.83。
  • 联合模型与单一模型AUC差异经DeLong检验有统计学意义(p<0.05)。
  • MRI影像组学特征取自增强第一期、DWI及ADC图手工勾画的原发灶感兴趣区。

核心结论

  • 基于MRI与活检全切片图像的联合列线图可预测乳腺癌患者新辅助化疗后的病理完全缓解。

研究设计与样本

  • 双中心回顾性研究,纳入331例病理确诊浸润性乳腺癌并接受新辅助化疗的患者;MRI影像组学基于增强第一期、DWI及ADC图手工勾画原发灶感兴趣区提取,病理深度学习模型基于全切片图像采用AlexNet建立。

诊断标准与关键数值

  • 联合模型列线图在训练队列AUC为0.95、测试队列AUC为0.84、外部验证队列AUC为0.83;DeLong检验显示联合模型与其他单一模型AUC差异有统计学意义(p<0.05)。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

AIM: To develop and validate a combined model based on magnetic resonance imaging (MRI), and whole-slide imaging (WSI) to predict pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) in breast cancer.

MATERIALS AND METHODS: We retrospectively enrolled 331 patients from two institutions, who pathologically confirmed to have invasive breast cancer and underwent NAC. Radiological features from original lesions on MRI and clinicopathological data were analysed using univariate and multivariate logistic analyses, which established the clinical model. Radiomics features were extracted based on the region of interest by manually delineating the primary focus on the sequences of the first phase of enhanced MRI, diffusion weighted imaging (DWI), and apparent diffusion coefficient (ADC) mapping. The radiomics prediction model was established using the Pearson product-moment correlation coefficient, least absolute shrinkage and selection operator (LASSO) regression analysis. A deep learning pathological model (DLPM) was established from WSIs of patients used AlexNet. The combined model was developed based on clinicopathological features, radiomics features, and pathomics deep learning features, and presented as a nomogram.

RESULTS: Five clinical and three deep learning features were screened and combined with the Rad-score to establish a combined model. The results showed that the nomogram had good predictive efficiency, with area under the curve (AUC) values of 0.95, 0.84, and 0.83 in the training, test, and external validation cohorts, respectively. The DeLong's test showed that the difference in AUC values between the combined model and the other single models was statistically significant (p < 0.05).

CONCLUSION: A combined nomogram based on MRI and biopsy WSI can predict pCR to NAC in patients with breast cancer.

来源:Clin Radiol,查看原文。版权归原作者所有。

引用

  1. Liu J, Wang X, Mao N, et al. The value of machine learning based on magnetic resonance imaging (MRI) and biopsy whole-slide image to predict pathological complete response to breast cancer after neoadjuvant chemotherapy: a two-centre study. Clin Radiol. 2025. PMID: 40582270. doi:10.1016/j.crad.2025.106976

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。