Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study

Abstract Objective Phyllodes tumors (PTs) are rare breast tumors with high recurrence rates, current methods relying on post-resection pathology often delay detection and require further surgery. We propose a deep-learning-based Phyllodes Tumors Hierarchical Diagnosis Model (PTs-HDM) for preoperativ...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuqi Yan, Yuanzhen Liu, Yao Wang, Tian Jiang, Jiayu Xie, Yahan Zhou, Xin Liu, Meiying Yan, Qiuqing Zheng, Haifei Xu, Jinxiao Chen, Lin Sui, Chen Chen, RongRong Ru, Kai Wang, Anli Zhao, Shiyan Li, Ying Zhu, Yang Zhang, Vicky Yang Wang, Dong Xu
Format: Article
Language:English
Published: BMC 2025-05-01
Series:Cancer Imaging
Subjects:
Online Access:https://doi.org/10.1186/s40644-025-00879-9
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850190639103213568
author Yuqi Yan
Yuanzhen Liu
Yao Wang
Tian Jiang
Jiayu Xie
Yahan Zhou
Xin Liu
Meiying Yan
Qiuqing Zheng
Haifei Xu
Jinxiao Chen
Lin Sui
Chen Chen
RongRong Ru
Kai Wang
Anli Zhao
Shiyan Li
Ying Zhu
Yang Zhang
Vicky Yang Wang
Dong Xu
author_facet Yuqi Yan
Yuanzhen Liu
Yao Wang
Tian Jiang
Jiayu Xie
Yahan Zhou
Xin Liu
Meiying Yan
Qiuqing Zheng
Haifei Xu
Jinxiao Chen
Lin Sui
Chen Chen
RongRong Ru
Kai Wang
Anli Zhao
Shiyan Li
Ying Zhu
Yang Zhang
Vicky Yang Wang
Dong Xu
author_sort Yuqi Yan
collection DOAJ
description Abstract Objective Phyllodes tumors (PTs) are rare breast tumors with high recurrence rates, current methods relying on post-resection pathology often delay detection and require further surgery. We propose a deep-learning-based Phyllodes Tumors Hierarchical Diagnosis Model (PTs-HDM) for preoperative identification and grading. Methods Ultrasound images from five hospitals were retrospectively collected, with all patients having undergone surgical pathological confirmation of either PTs or fibroadenomas (FAs). PTs-HDM follows a two-stage classification: first distinguishing PTs from FAs, then grading PTs into benign or borderline/malignant. Model performance metrics including AUC and accuracy were quantitatively evaluated. A comparative analysis was conducted between the algorithm’s diagnostic capabilities and those of radiologists with varying clinical experience within an external validation cohort. Through the provision of PTs-HDM’s automated classification outputs and associated thermal activation mapping guidance, we systematically assessed the enhancement in radiologists’ diagnostic concordance and classification accuracy. Results A total of 712 patients were included. On the external test set, PTs-HDM achieved an AUC of 0.883, accuracy of 87.3% for PT vs. FA classification. Subgroup analysis showed high accuracy for tumors < 2 cm (90.9%). In hierarchical classification, the model obtained an AUC of 0.856 and accuracy of 80.9%. Radiologists’ performance improved with PTs-HDM assistance, with binary classification accuracy increasing from 82.7%, 67.7%, and 64.2–87.6%, 76.6%, and 82.1% for senior, attending, and resident radiologists, respectively. Their hierarchical classification AUCs improved from 0.566 to 0.827 to 0.725–0.837. PTs-HDM also enhanced inter-radiologist consistency, increasing Kappa values from − 0.05 to 0.41 to 0.12 to 0.65, and the intraclass correlation coefficient from 0.19 to 0.45. Conclusion PTs-HDM shows strong diagnostic performance, especially for small lesions, and improves radiologists’ accuracy across all experience levels, bridging diagnostic gaps and providing reliable support for PTs’ hierarchical diagnosis.
format Article
id doaj-art-1c46d382b80f42bbbc40bef5f2b1ec9f
institution OA Journals
issn 1470-7330
language English
publishDate 2025-05-01
publisher BMC
record_format Article
series Cancer Imaging
spelling doaj-art-1c46d382b80f42bbbc40bef5f2b1ec9f2025-08-20T02:15:12ZengBMCCancer Imaging1470-73302025-05-0125111310.1186/s40644-025-00879-9Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center studyYuqi Yan0Yuanzhen Liu1Yao Wang2Tian Jiang3Jiayu Xie4Yahan Zhou5Xin Liu6Meiying Yan7Qiuqing Zheng8Haifei Xu9Jinxiao Chen10Lin Sui11Chen Chen12RongRong Ru13Kai Wang14Anli Zhao15Shiyan Li16Ying Zhu17Yang Zhang18Vicky Yang Wang19Dong Xu20Department of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Ultrasound, Lishui People’s HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalPostgraduate training base Alliance of Wenzhou Medical University (Zhejiang Cancer Hospital)Wenling Institute of Big Data and Artificial Intelligence Institute in MedicineDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Ultrasound, Lishui People’s HospitalDepartment of Ultrasound, Lishui People’s HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalDepartment of Ultrasound, Zhejiang Xiaoshan HospitalDepartment of Ultrasound, The Affiliated Dongyang Hospital of Wenzhou Medical UniversityDepartment of Ultrasound, The Affiliated Dongyang Hospital of Wenzhou Medical UniversityDepartment of Ultrasound in Medicine, Affiliated Sir Run Run Shaw Hospital of Zhejiang University School of MedicineDepartment of Ultrasound, Ruijin Hospital, Shanghai Jiaotong University School of MedicineWenling Institute of Big Data and Artificial Intelligence Institute in MedicineWenling Institute of Big Data and Artificial Intelligence Institute in MedicineDepartment of Diagnostic Ultrasound Imaging & Interventional Therapy, Zhejiang Cancer HospitalAbstract Objective Phyllodes tumors (PTs) are rare breast tumors with high recurrence rates, current methods relying on post-resection pathology often delay detection and require further surgery. We propose a deep-learning-based Phyllodes Tumors Hierarchical Diagnosis Model (PTs-HDM) for preoperative identification and grading. Methods Ultrasound images from five hospitals were retrospectively collected, with all patients having undergone surgical pathological confirmation of either PTs or fibroadenomas (FAs). PTs-HDM follows a two-stage classification: first distinguishing PTs from FAs, then grading PTs into benign or borderline/malignant. Model performance metrics including AUC and accuracy were quantitatively evaluated. A comparative analysis was conducted between the algorithm’s diagnostic capabilities and those of radiologists with varying clinical experience within an external validation cohort. Through the provision of PTs-HDM’s automated classification outputs and associated thermal activation mapping guidance, we systematically assessed the enhancement in radiologists’ diagnostic concordance and classification accuracy. Results A total of 712 patients were included. On the external test set, PTs-HDM achieved an AUC of 0.883, accuracy of 87.3% for PT vs. FA classification. Subgroup analysis showed high accuracy for tumors < 2 cm (90.9%). In hierarchical classification, the model obtained an AUC of 0.856 and accuracy of 80.9%. Radiologists’ performance improved with PTs-HDM assistance, with binary classification accuracy increasing from 82.7%, 67.7%, and 64.2–87.6%, 76.6%, and 82.1% for senior, attending, and resident radiologists, respectively. Their hierarchical classification AUCs improved from 0.566 to 0.827 to 0.725–0.837. PTs-HDM also enhanced inter-radiologist consistency, increasing Kappa values from − 0.05 to 0.41 to 0.12 to 0.65, and the intraclass correlation coefficient from 0.19 to 0.45. Conclusion PTs-HDM shows strong diagnostic performance, especially for small lesions, and improves radiologists’ accuracy across all experience levels, bridging diagnostic gaps and providing reliable support for PTs’ hierarchical diagnosis.https://doi.org/10.1186/s40644-025-00879-9Deep learningUltrasoundBreastPhyllodes tumorsFibroadenoma
spellingShingle Yuqi Yan
Yuanzhen Liu
Yao Wang
Tian Jiang
Jiayu Xie
Yahan Zhou
Xin Liu
Meiying Yan
Qiuqing Zheng
Haifei Xu
Jinxiao Chen
Lin Sui
Chen Chen
RongRong Ru
Kai Wang
Anli Zhao
Shiyan Li
Ying Zhu
Yang Zhang
Vicky Yang Wang
Dong Xu
Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
Cancer Imaging
Deep learning
Ultrasound
Breast
Phyllodes tumors
Fibroadenoma
title Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
title_full Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
title_fullStr Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
title_full_unstemmed Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
title_short Hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images: a retrospective multi-center study
title_sort hierarchical diagnosis of breast phyllodes tumors enabled by deep learning of ultrasound images a retrospective multi center study
topic Deep learning
Ultrasound
Breast
Phyllodes tumors
Fibroadenoma
url https://doi.org/10.1186/s40644-025-00879-9
work_keys_str_mv AT yuqiyan hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT yuanzhenliu hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT yaowang hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT tianjiang hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT jiayuxie hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT yahanzhou hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT xinliu hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT meiyingyan hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT qiuqingzheng hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT haifeixu hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT jinxiaochen hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT linsui hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT chenchen hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT rongrongru hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT kaiwang hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT anlizhao hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT shiyanli hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT yingzhu hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT yangzhang hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT vickyyangwang hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy
AT dongxu hierarchicaldiagnosisofbreastphyllodestumorsenabledbydeeplearningofultrasoundimagesaretrospectivemulticenterstudy