Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis

Abstract Based on the expertise of pathologists, the pixelwise manual annotation has provided substantial support for training deep learning models of whole slide images (WSI)-assisted diagnostic. However, the collection of pixelwise annotation demands massive annotation time from pathologists, lead...

Full description

Saved in:
Bibliographic Details
Main Authors: Tianhang Nan, Song Zheng, Siyuan Qiao, Hao Quan, Xin Gao, Jun Niu, Bin Zheng, Chunfang Guo, Yue Zhang, Xiaoqin Wang, Liping Zhao, Ze Wu, Yaoxing Guo, Xingyu Li, Mingchen Zou, Shuangdi Ning, Yue Zhao, Wei Qian, Hongduo Chen, Ruiqun Qi, Xinghua Gao, Xiaoyu Cui
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-025-60307-1
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849238519228137472
author Tianhang Nan
Song Zheng
Siyuan Qiao
Hao Quan
Xin Gao
Jun Niu
Bin Zheng
Chunfang Guo
Yue Zhang
Xiaoqin Wang
Liping Zhao
Ze Wu
Yaoxing Guo
Xingyu Li
Mingchen Zou
Shuangdi Ning
Yue Zhao
Wei Qian
Hongduo Chen
Ruiqun Qi
Xinghua Gao
Xiaoyu Cui
author_facet Tianhang Nan
Song Zheng
Siyuan Qiao
Hao Quan
Xin Gao
Jun Niu
Bin Zheng
Chunfang Guo
Yue Zhang
Xiaoqin Wang
Liping Zhao
Ze Wu
Yaoxing Guo
Xingyu Li
Mingchen Zou
Shuangdi Ning
Yue Zhao
Wei Qian
Hongduo Chen
Ruiqun Qi
Xinghua Gao
Xiaoyu Cui
author_sort Tianhang Nan
collection DOAJ
description Abstract Based on the expertise of pathologists, the pixelwise manual annotation has provided substantial support for training deep learning models of whole slide images (WSI)-assisted diagnostic. However, the collection of pixelwise annotation demands massive annotation time from pathologists, leading to a high burden of medical manpower resources, hindering to construct larger datasets and more precise diagnostic models. To obtain pathologists’ expertise with minimal pathologist workloads then achieve precise diagnostics, we collect the image review patterns of pathologists by eye-tracking devices. Simultaneously, we design a deep learning system: Pathology Expertise Acquisition Network (PEAN), based on the collected visual patterns, which can decode pathologists’ expertise and then diagnose WSIs. Eye-trackers reduce the time required for annotating WSIs to 4%, of the manual annotation. We evaluate PEAN on 5881 WSIs and 5 categories of skin lesions, achieving a high area under the curve of 0.992 and an accuracy of 96.3% on diagnostic prediction. This study fills the gap in existing models’ inability to learn from the diagnostic processes of pathologists. Its efficient data annotation and precise diagnostics provide assistance in both large-scale data collection and clinical care.
format Article
id doaj-art-22493a42682c4ae9a91c53aada7beb60
institution Kabale University
issn 2041-1723
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Nature Communications
spelling doaj-art-22493a42682c4ae9a91c53aada7beb602025-08-20T04:01:35ZengNature PortfolioNature Communications2041-17232025-07-0116111410.1038/s41467-025-60307-1Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosisTianhang Nan0Song Zheng1Siyuan Qiao2Hao Quan3Xin Gao4Jun Niu5Bin Zheng6Chunfang Guo7Yue Zhang8Xiaoqin Wang9Liping Zhao10Ze Wu11Yaoxing Guo12Xingyu Li13Mingchen Zou14Shuangdi Ning15Yue Zhao16Wei Qian17Hongduo Chen18Ruiqun Qi19Xinghua Gao20Xiaoyu Cui21College of Medicine and Biological Information Engineering, Northeastern UniversityDepartment of Dermatology, The First Hospital of China Medical UniversityCollege of Computer Science and Technology, Fudan UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityComputer Science Program, Computer, Electrical and Mathematical Sciences and Engineering Division, King Abdullah University of Science and Technology (KAUST)Department of Dermatology, General Hospital of Northern Theater CommandCollege of Medicine and Biological Information Engineering, Northeastern UniversityDepartment of Dermatology, Shenyang Seventh People’s HospitalDepartment of Dermatology, Shengjing hospital of China Medical UniversityCenter of Excellence on Generative AI, King Abdullah University of Science and TechnologyDepartment of Dermatology, Zhongyi Northeast International HospitalComputer Science Program, Computer, Electrical and Mathematical Sciences and Engineering Division, King Abdullah University of Science and Technology (KAUST)Department of Dermatology, The First Hospital of China Medical UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityDepartment of Dermatology, The First Hospital of China Medical UniversityDepartment of Dermatology, The First Hospital of China Medical UniversityDepartment of Dermatology, The First Hospital of China Medical UniversityCollege of Medicine and Biological Information Engineering, Northeastern UniversityAbstract Based on the expertise of pathologists, the pixelwise manual annotation has provided substantial support for training deep learning models of whole slide images (WSI)-assisted diagnostic. However, the collection of pixelwise annotation demands massive annotation time from pathologists, leading to a high burden of medical manpower resources, hindering to construct larger datasets and more precise diagnostic models. To obtain pathologists’ expertise with minimal pathologist workloads then achieve precise diagnostics, we collect the image review patterns of pathologists by eye-tracking devices. Simultaneously, we design a deep learning system: Pathology Expertise Acquisition Network (PEAN), based on the collected visual patterns, which can decode pathologists’ expertise and then diagnose WSIs. Eye-trackers reduce the time required for annotating WSIs to 4%, of the manual annotation. We evaluate PEAN on 5881 WSIs and 5 categories of skin lesions, achieving a high area under the curve of 0.992 and an accuracy of 96.3% on diagnostic prediction. This study fills the gap in existing models’ inability to learn from the diagnostic processes of pathologists. Its efficient data annotation and precise diagnostics provide assistance in both large-scale data collection and clinical care.https://doi.org/10.1038/s41467-025-60307-1
spellingShingle Tianhang Nan
Song Zheng
Siyuan Qiao
Hao Quan
Xin Gao
Jun Niu
Bin Zheng
Chunfang Guo
Yue Zhang
Xiaoqin Wang
Liping Zhao
Ze Wu
Yaoxing Guo
Xingyu Li
Mingchen Zou
Shuangdi Ning
Yue Zhao
Wei Qian
Hongduo Chen
Ruiqun Qi
Xinghua Gao
Xiaoyu Cui
Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
Nature Communications
title Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
title_full Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
title_fullStr Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
title_full_unstemmed Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
title_short Deep learning quantifies pathologists’ visual patterns for whole slide image diagnosis
title_sort deep learning quantifies pathologists visual patterns for whole slide image diagnosis
url https://doi.org/10.1038/s41467-025-60307-1
work_keys_str_mv AT tianhangnan deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT songzheng deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT siyuanqiao deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT haoquan deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT xingao deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT junniu deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT binzheng deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT chunfangguo deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT yuezhang deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT xiaoqinwang deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT lipingzhao deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT zewu deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT yaoxingguo deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT xingyuli deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT mingchenzou deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT shuangdining deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT yuezhao deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT weiqian deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT hongduochen deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT ruiqunqi deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT xinghuagao deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis
AT xiaoyucui deeplearningquantifiespathologistsvisualpatternsforwholeslideimagediagnosis