Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)

Background and objective: Diagnosis of breast intraductal proliferative lesions (BIDPLs) in hematoxylin-eosin (HE) images remains a time-consuming and intractable topic because of subjective processes and subtle morphological differences. Convolutional neural networks (CNNs) show great potential for...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiangmin Li, Jiamei Chen, Bo Luo, Minyan Xia, Xu Zhang, Hangjia Zhu, Yutian Zhang-Cai, Yongshun Chen, Yang Yang, Yaofeng Wen
Format: Article
Language:English
Published: Elsevier 2025-05-01
Series:Heliyon
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2405844025011375
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849694873295257600
author Xiangmin Li
Jiamei Chen
Bo Luo
Minyan Xia
Xu Zhang
Hangjia Zhu
Yutian Zhang-Cai
Yongshun Chen
Yang Yang
Yaofeng Wen
author_facet Xiangmin Li
Jiamei Chen
Bo Luo
Minyan Xia
Xu Zhang
Hangjia Zhu
Yutian Zhang-Cai
Yongshun Chen
Yang Yang
Yaofeng Wen
author_sort Xiangmin Li
collection DOAJ
description Background and objective: Diagnosis of breast intraductal proliferative lesions (BIDPLs) in hematoxylin-eosin (HE) images remains a time-consuming and intractable topic because of subjective processes and subtle morphological differences. Convolutional neural networks (CNNs) show great potential for providing objective analysis strategies for HE images. In this study, we proposed a novel knowledge distillation (KD) framework using CNN-based nuclei segmentation-assisted classification (KDCNN-NSAC). Methods: The diagnosis of BIDPLs is treated as multiple class classification tasks in the BReAst Carcinoma Subtyping dataset. The KDCNN-NSAC fully leveraged the epithelial and stromal nuclei-level features in training phases and performed region-of-interest (ROI)-level classifications in predicting phases. Then, the whole slide image (WSI) was diagnosed based on the risk ratings of the ROIs within it, instead of processing a WSI. Results: The principal results showed that in ROI-level classifications, KDCNN-NSAC outperformed the state-of-the-art methods for 7-class classification with an average F1 score of 63.26 % and achieves F1 score of 98.36 % and 94.21 %, respectively, in distinguishing BIDPLs from invasive cancer and normal tissue. The WSI-level predictions obtained a high degree of consistency with the pathologists’ annotation (kappa value of 0.88). Ablation experiments showed that nuclei segmentation and classification components improve the performance of the baseline model in KDCNN-NSAC by 3 %. Conclusions: The KDCNN-NSAC makes the model focus on important cellular information and predicts the WSI in accordance with the pathologists’ diagnostic thinking, thus improving model explainability. Moreover, the introduce of KDCNN-NSAC will help achieve superior performance in diagnosing BIDPLs.
format Article
id doaj-art-e4ba64ebec3f4c0ab9ab2cd214e2fb40
institution DOAJ
issn 2405-8440
language English
publishDate 2025-05-01
publisher Elsevier
record_format Article
series Heliyon
spelling doaj-art-e4ba64ebec3f4c0ab9ab2cd214e2fb402025-08-20T03:19:56ZengElsevierHeliyon2405-84402025-05-011110e4275610.1016/j.heliyon.2025.e42756Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)Xiangmin Li0Jiamei Chen1Bo Luo2Minyan Xia3Xu Zhang4Hangjia Zhu5Yutian Zhang-Cai6Yongshun Chen7Yang Yang8Yaofeng Wen9School of Mechanical Engineering, Shanghai Dianji University, Shanghai, 201306, ChinaCenter of Oncology, Renmin Hospital of Wuhan University, Wuhan, 430060, ChinaDepartment of Pathology, The Central Hospital of Wuhan, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430014, ChinaSchool of Mechanical Engineering, Shanghai Dianji University, Shanghai, 201306, ChinaSchool of Mechanical Engineering, Shanghai Dianji University, Shanghai, 201306, ChinaCenter of Oncology, Renmin Hospital of Wuhan University, Wuhan, 430060, ChinaCenter of Oncology, Renmin Hospital of Wuhan University, Wuhan, 430060, ChinaCenter of Oncology, Renmin Hospital of Wuhan University, Wuhan, 430060, China; Corresponding author.School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200030, China; Shanghai Lanhui Medical Technology Co, Shanghai, 200063, China; Corresponding author. School of Biomedical Engineering, Shanghai Jiao Tong University Shanghai Lanhui Medical Technology Co, Shanghai, 200063, China.School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200030, China; Shanghai Lanhui Medical Technology Co, Shanghai, 200063, China; Corresponding author. School of Biomedical Engineering, Shanghai Jiao Tong University Shanghai Lanhui Medical Technology Co, Shanghai, 200063, China.Background and objective: Diagnosis of breast intraductal proliferative lesions (BIDPLs) in hematoxylin-eosin (HE) images remains a time-consuming and intractable topic because of subjective processes and subtle morphological differences. Convolutional neural networks (CNNs) show great potential for providing objective analysis strategies for HE images. In this study, we proposed a novel knowledge distillation (KD) framework using CNN-based nuclei segmentation-assisted classification (KDCNN-NSAC). Methods: The diagnosis of BIDPLs is treated as multiple class classification tasks in the BReAst Carcinoma Subtyping dataset. The KDCNN-NSAC fully leveraged the epithelial and stromal nuclei-level features in training phases and performed region-of-interest (ROI)-level classifications in predicting phases. Then, the whole slide image (WSI) was diagnosed based on the risk ratings of the ROIs within it, instead of processing a WSI. Results: The principal results showed that in ROI-level classifications, KDCNN-NSAC outperformed the state-of-the-art methods for 7-class classification with an average F1 score of 63.26 % and achieves F1 score of 98.36 % and 94.21 %, respectively, in distinguishing BIDPLs from invasive cancer and normal tissue. The WSI-level predictions obtained a high degree of consistency with the pathologists’ annotation (kappa value of 0.88). Ablation experiments showed that nuclei segmentation and classification components improve the performance of the baseline model in KDCNN-NSAC by 3 %. Conclusions: The KDCNN-NSAC makes the model focus on important cellular information and predicts the WSI in accordance with the pathologists’ diagnostic thinking, thus improving model explainability. Moreover, the introduce of KDCNN-NSAC will help achieve superior performance in diagnosing BIDPLs.http://www.sciencedirect.com/science/article/pii/S2405844025011375Breast intraductal proliferative lesionsBreast cancerKnowledge distillationConvolutional neural networkNuclei segmentationClassification
spellingShingle Xiangmin Li
Jiamei Chen
Bo Luo
Minyan Xia
Xu Zhang
Hangjia Zhu
Yutian Zhang-Cai
Yongshun Chen
Yang Yang
Yaofeng Wen
Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
Heliyon
Breast intraductal proliferative lesions
Breast cancer
Knowledge distillation
Convolutional neural network
Nuclei segmentation
Classification
title Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
title_full Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
title_fullStr Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
title_full_unstemmed Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
title_short Classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network-based nuclei-segmentation-assisted classification (KDCNN-NSAC)
title_sort classifying breast intraductal proliferative lesions via a knowledge distillation framework using convolutional neural network based nuclei segmentation assisted classification kdcnn nsac
topic Breast intraductal proliferative lesions
Breast cancer
Knowledge distillation
Convolutional neural network
Nuclei segmentation
Classification
url http://www.sciencedirect.com/science/article/pii/S2405844025011375
work_keys_str_mv AT xiangminli classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT jiameichen classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT boluo classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT minyanxia classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT xuzhang classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT hangjiazhu classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT yutianzhangcai classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT yongshunchen classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT yangyang classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac
AT yaofengwen classifyingbreastintraductalproliferativelesionsviaaknowledgedistillationframeworkusingconvolutionalneuralnetworkbasednucleisegmentationassistedclassificationkdcnnnsac