Evaluation of neoadjuvant chemotherapy response for breast cancer with artificial intelligence assisted imaging ZHANG Nan1, ZHANG Yafang2, ZHOU　Qi3

张楠; 张亚芳; 周琪

doi:10.19732/j.cnki.2096-6210.2026.02.006

您当前的位置：

首页 >

文章列表页 >

Evaluation of neoadjuvant chemotherapy response for breast cancer with artificial intelligence assisted imaging ZHANG Nan¹, ZHANG Yafang², ZHOU　Qi³

Article | 更新时间：2026-05-07

- Evaluation of neoadjuvant chemotherapy response for breast cancer with artificial intelligence assisted imaging ZHANG Nan¹, ZHANG Yafang², ZHOU　Qi³
- Oncoradiology Vol. 35, Issue 2, Pages: 264-272(2026)
- 作者机构：
  
  1.唐山市人民医院乳腺外科，河北唐山 063000
  2.唐山市人民医院核磁室，河北唐山 063000
  3.唐山市人民医院乳腺三科，河北唐山 063000
- 作者简介：
- 基金信息：
  
  Hebei Provincial Health Commission Medical Scientific Research Project(20251218)
- DOI：10.19732/j.cnki.2096-6210.2026.02.006
  CLC： R737.9;R445.2
- Received：27 November 2025，
  
  Revised：2026-01-17，
  
  Published：28 April 2026
- 稿件说明：
移动端阅览
张　楠, 张亚芳, 周　琪. 人工智能辅助影像学评价乳腺癌患者新辅助化疗效果［J］. 肿瘤影像学, 2026, 35(2): 264-272.

ZHANG N, ZHANG Y F, ZHOU Q. Citation:Evaluation of neoadjuvant chemotherapy for breast cancer with artificial intelligence assisted imaging［J］. Oncoradiology, 2026, 35(2): 264-272.
张　楠, 张亚芳, 周　琪. 人工智能辅助影像学评价乳腺癌患者新辅助化疗效果［J］. 肿瘤影像学, 2026, 35(2): 264-272. DOI： 10.19732/j.cnki.2096-6210.2026.02.006.

ZHANG N, ZHANG Y F, ZHOU Q. Citation:Evaluation of neoadjuvant chemotherapy for breast cancer with artificial intelligence assisted imaging［J］. Oncoradiology, 2026, 35(2): 264-272. DOI： 10.19732/j.cnki.2096-6210.2026.02.006.

摘要

目的

探讨人工智能辅助影像学在乳腺癌新辅助化疗（neoadjuvant chemotherapy，NAC）效果评价中的应用价值。

方法

从公共数据集中下载乳腺癌患者的临床资料进行回顾性分析，按照2∶1比例随机分为训练集和验证集。另回顾并分析2023年3月—2025年5月于唐山市人民医院接受治疗的乳腺癌患者的临床资料作为测试集。所有患者术前均行磁共振成像检查，且术后均有明确的病理学检查结果。利用分层成像维度累加结合锚定注意力框和不同深度学习模型（convnext、efficientnet、swin、vit）、不同机器学习算法（SVM、Ranger）组合构建8个机器学习模型，基于训练集数据筛选最佳机器学习模型。使用验证集对最佳机器学习模型进行优化，并利用测试集评价该模型预测乳腺癌NAC效果的效能。

结果

本研究共构建8个机器学习模型，其中基于CropNor-Substract-convnext-Ranger的机器学习模型预测乳腺癌NAC效果的灵敏度、特异度、准确度、曲线下面积均高于其他类型机器学习模型；使用验证集对CropNor-Substract-convnext-Ranger机器学习模型进行优化，优化终止时交叉熵损失为1.012，灵敏度为90.91%、特异度为92.67%、准确度为91.97%；CropNor-Substract-convnext-Ranger机器学习模型预测测试集乳腺癌NAC效果的灵敏度为90.48%、特异度为92.81%、准确度为91.86%、曲线下面积为0.924，其与病理学检查结果的一致性Kappa指数为0.832（

0.001）。

结论

CropNor-Substract-convnext-Ranger机器学习模型预测乳腺癌NAC效果的效能较好，具有较好的临床应用价值。

Abstract

Objective

To explore the application value of artificial intelligence assisted imaging in the evaluation of neoadjuvant chemotherapy (NAC) response for breast cancer.

Methods

The clinical data of breast cancer patients downloaded from the public data set were retrospectively analyzed

and they were randomly divided into a training set and a validation set according to the 2∶1 ratio. In addition

the clinical data of breast cancer patients treated in Tangshan People's Hospital from March 2023 to May 2025 were analyzed retrospectively as the test set. All patients underwent magnetic resonance imaging examination before surgery

and all patients had definitive postoperative pathological results. Eight machine learning models were constructed by combining layered imaging dimension accumulation with anchored attention boxes and different deep learning models (convnext

efficientnet

swin

and vit) and different machine learning algorithms (SVM and Ranger)

and the best machine learning model was selected based on the training set data. The best machine learning model was optimized using the validation set

and the effectiveness of the model in predicting the curative effect of breast cancer NAC was evaluated using the test set.

Results

Eight machine learning models were constructed in this study

and the sensitivity

specificity

accuracy and area under curve of the machine learning model based on CropNor-Subtracts-convnext-Ranger were higher than those of other machine learning models in predicting the efficacy of NAC in breast cancer. The CropNor-Subtracts-convnext-Ranger machine learning model was optimized using the validation set

with a cross entropy loss of 1.012 at the end of the optimization

sensitivity of 90.91%

specificity of 92.67% and accuracy of 91.97%. The sensitivity

specificity

accuracy

area under curve of CropNor-Subtract-convnext-Ranger machine learning model to predict the efficacy of breast cance

r NAC in the test set were 90.48%

92.81%

91.86%

0.924

and the consistency Kappa index with the pathological results was 0.832 (

0.001).

Conclusion

CropNor-Subtract-convnext-Ranger machine learning model shows good performance in predicting NAC response in breast cancer

and it has good clinical application value.

关键词

Keywords

references

SUN K X , ZHANG B L , LEI S Y , et al . Incidence, mortality, and disability-adjusted life years of female breast cancer in China, 2022 ［J］. Chin Med J , 2024 , 137 ( 20 ): 2429 - 2436 .

张恒乐 , 董雪珂 , 殷雨来 , 等 . 三阴性乳腺癌TP方案新辅助化疗疗效及预后因素分析［J］. 中国新药与临床杂志 , 2023 , 42 ( 10 ): 664 - 669 .

ZHANG H L /Y) , DONG X K , YIN Y L , et al . Analysis of efficacy and prognostic factors of neoadjuvant chemotherapy with TP regimen in triple negative breast cancer ［J］. Chin J New Drugs Clin Remedies , 2023 , 42 ( 10 ): 664 - 669 .

LIU M , XIANG Q , DAI F S , et al . Comparison of the pathological complete response rate and survival between HER2-low and HER2-zero breast cancer in neoadjuvant chemotherapy setting: a systematic review and meta-analysis ［J］. Clin Breast Cancer , 2024 , 24 ( 7 ): 575 - 584.e1 .

ZHENG G Y , PENG J X , SHU Z Y , et al . Predicting pathological complete response to neoadjuvant chemotherapy in breast cancer patients: use of MRI radiomics data from three regions with multiple machine learning algorithms ［J］. J Cancer Res Clin Oncol , 2024 , 150 ( 3 ): 147 .

FENG X D , SHI Y , WU M , et al . Predicting the efficacy of neoadjuvant chemotherapy in breast cancer patients based on ultrasound longitudinal temporal depth network fusion model ［J］. Breast Cancer Res , 2025 , 27 ( 1 ): 30 .

罗云昭 , 蒋宏传 , 徐　峰 . 基于深度学习模型辅助穿刺病理图像预测乳腺癌新辅助治疗疗效的研究［J］. 中国全科医学 , 2025 , 28 ( 19 ): 2407 - 2413 .

LUO Y Z , JIANG H C , XU F . Predicting response to neoadjuvant therapy in breast cancer using deep learning on primary core needle biopsy slides ［J］. Chin Gen Pract , 2025 , 28 ( 19 ): 2407 - 2413 .

ZHOU Q , PENG F , PANG Z Y , et al . Development and validation of an interpretable machine learning model for diagnosing pathologic complete response in breast cancer ［J］. Comput Methods Programs Biomed , 2025 , 267 : 108803 .

中国抗癌协会乳腺癌专业委员会 . 中国抗癌协会乳腺癌诊治指南与规范（2021年版）［J］. 中国癌症杂志 , 2021 , 31 ( 10 ): 367 - 417 .

The Society of Breast Cancer China Anti-Cancer Association, Breast Oncology Group of the Oncology Branch of the Chinese Medical Association . Guidelines and norms for diagnosis and treatment of breast cancer of China Anti-Cancer Association (2021 edition) ［J］. China Oncol , 2021 , 31 ( 10 ): 367 - 417 .

杨学宁 , 吴一龙 . 实体瘤治疗疗效评价标准—RECIST ［J］. 循证医学 , 2004 , 4 ( 2 ): 85 - 90 .

YANG X N , WU Y L . RECIST-New guidelines to evaluate the response to treatment in solid tumors ［J］. J Evid Based Med , 2004 , 4 ( 2 ): 85 - 90 .

LI J , CHEN C , NIE J J , et al . Changes in the disease burden of breast cancer along with attributable risk factors in China from 1990 to 2019 and its projections: an analysis of the global burden of disease study 2019 ［J］. Cancer Med , 2023 , 12 ( 2 ): 1888 - 1902 .

周昭悦 , 李　越 , 尚玲敏 , 等 . 乳腺癌新辅助化疗后病理学完全缓解的因素分析［J］. 现代肿瘤医学 , 2023 , 31 ( 1 ): 86 - 90 .

ZHOU Z Y , LI Y , SHANG L M , et al . Analysis of factors associated with pathological complete remission after neoadjuvant chemotherapy for breast cancer ［J］. J Mod Oncol , 2023 , 31 ( 1 ): 86 - 90 .

ZHANG J Q , WU Q , YIN W , et al . Development and validation of a radiopathomic model for predicting pathologic complete response to neoadjuvant chemotherapy in breast cancer patients ［J］. BMC Cancer , 2023 , 23 ( 1 ): 431 .

徐子良 , 郑敏文 . 影像人工智能在医学领域的时代创新与挑战［J］. 山东大学学报（医学版） , 2023 , 61 ( 12 ): 7 - 12 .

XU Z L , ZHENG M W . Innovation and challenge of imaging artificial intelligence in medical field ［J］. J Shandong Univ Health Sci , 2023 , 61 ( 12 ): 7 - 12 .

李秀涛 , 谭理连 , 邱绍清 , 等 . 人工智能在早期胸部CT肋骨骨折的临床应用价值及局限性分析［J］. 中国CT和MRI杂志 , 2023 , 21 ( 10 ): 115 - 118 .

LI X T , TAN L L , QIU S Q , et al . Clinical application value and limitation analysis of artificial intelligence in early thoracic CT for rib fractures ［J］. Chin J CT MRI , 2023 , 21 ( 10 ): 115 - 118 .

WANG C C , WU Z Z , SUN M L , et al . Instance segmentation based on global-local attention and local Chan-Vese model ［J］. Signal Image Video Process , 2025 , 19 ( 9 ): 746 .

韩　磊 , 裴溪源 , 温军玲 . 基于ConvNeXt模型的胸部X线图像的疾病分类与可视化［J］. 北京生物医学工程 , 2024 , 43 ( 4 ): 346 - 351 .

HAN L , PEI X Y , WEN J L . Disease classification and visualization on chest X-ray images based on the ConvNeXt model ［J］. Beijing Biomed Eng , 2024 , 43 ( 4 ): 346 - 351 .

Views

下载量

CNKI被引量

Alert me when the article has been cited

提交

Tools

Publicity Resources

Automated classification of breast MRI background parenchymal enhancement using deep learning and thresholding segmentation method

Early prediction of response to neoadjuvant chemotherapy using quantitative dynamic contrast-enhanced magnetic resonance imaging in locally advanced breast cancer

Magnetic resonance imaging with apparent diffusion coef f icient histogram analysis: evaluation of luminal type breast cancer prior to neoadjuvant chemotherapy

Application and prospect of artificial intelligence in breast imaging diagnosis

From structure to function: evolution of magnetic resonance technology and transformation of imaging diagnosis

Related Author

MA Mingming

ZHANG Yaofeng

WANG Xiangpeng

ZHANG Xiaodong

QIN Naishan

WANG Xiaoying

SUN Haixin

ZHANG Renzhi

Related Institution

Department of Radiology, Peking University First Hospital

Beijing Smart-Imaging Technology Co. Ltd.

Department of Radiology, Beijing Chaoyang Hospital, Capital Medical University

Department of Radiology, Cancer Hospital Chinese Academy of Medical Sciences

Department of Radiology, Renji Hospital, School of Medicine, Shanghai Jiao Tong University

AI脑图

AI问答

⁰