A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides

Abstract Mitotic activity is an important feature for grading several cancer types. However, counting mitotic figures (cells in division) is a time-consuming and laborious task prone to inter-observer variation. Inaccurate recognition of MFs can lead to incorrect grading and hence potential suboptim...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhuoyan Shen, Mikaël Simard, Douglas Brand, Vanghelita Andrei, Ali Al-Khader, Fatine Oumlil, Katherine Trevers, Thomas Butters, Simon Haefliger, Eleanna Kara, Fernanda Amary, Roberto Tirabosco, Paul Cool, Gary Royle, Maria A. Hawkins, Adrienne M. Flanagan, Charles-Antoine Collins-Fekete
Format: Article
Language:English
Published: Nature Portfolio 2024-12-01
Series:Communications Biology
Online Access:https://doi.org/10.1038/s42003-024-07398-6
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850102771827605504
author Zhuoyan Shen
Mikaël Simard
Douglas Brand
Vanghelita Andrei
Ali Al-Khader
Fatine Oumlil
Katherine Trevers
Thomas Butters
Simon Haefliger
Eleanna Kara
Fernanda Amary
Roberto Tirabosco
Paul Cool
Gary Royle
Maria A. Hawkins
Adrienne M. Flanagan
Charles-Antoine Collins-Fekete
author_facet Zhuoyan Shen
Mikaël Simard
Douglas Brand
Vanghelita Andrei
Ali Al-Khader
Fatine Oumlil
Katherine Trevers
Thomas Butters
Simon Haefliger
Eleanna Kara
Fernanda Amary
Roberto Tirabosco
Paul Cool
Gary Royle
Maria A. Hawkins
Adrienne M. Flanagan
Charles-Antoine Collins-Fekete
author_sort Zhuoyan Shen
collection DOAJ
description Abstract Mitotic activity is an important feature for grading several cancer types. However, counting mitotic figures (cells in division) is a time-consuming and laborious task prone to inter-observer variation. Inaccurate recognition of MFs can lead to incorrect grading and hence potential suboptimal treatment. This study presents an artificial intelligence-based approach to detect mitotic figures in digitised whole-slide images stained with haematoxylin and eosin. Advances in this area are hampered by the small size and variety of datasets available. To address this, we create the largest dataset of mitotic figures (N = 74,620), combining an in-house dataset of soft tissue tumours with five open-source datasets. We then employ a two-stage framework, named the Optimised Mitoses Generator Network (OMG-Net), to identify mitotic figures. This framework first deploys the Segment Anything Model to automatically outline cells, followed by an adapted ResNet18 that distinguishes mitotic figures. OMG-Net achieves an F1 score of 0.84 in detecting pan-cancer mitotic figures, including human breast carcinoma, neuroendocrine tumours, and melanoma. It outperforms previous state-of-the-art models in hold-out test sets. To summarise, our study introduces a generalisable data creation and curation pipeline and a high-performance detection model, which can largely contribute to the field of computer-aided mitotic figure detection.
format Article
id doaj-art-862ddc50bb82412983a0b201e5558c21
institution DOAJ
issn 2399-3642
language English
publishDate 2024-12-01
publisher Nature Portfolio
record_format Article
series Communications Biology
spelling doaj-art-862ddc50bb82412983a0b201e5558c212025-08-20T02:39:41ZengNature PortfolioCommunications Biology2399-36422024-12-017111110.1038/s42003-024-07398-6A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slidesZhuoyan Shen0Mikaël Simard1Douglas Brand2Vanghelita Andrei3Ali Al-Khader4Fatine Oumlil5Katherine Trevers6Thomas Butters7Simon Haefliger8Eleanna Kara9Fernanda Amary10Roberto Tirabosco11Paul Cool12Gary Royle13Maria A. Hawkins14Adrienne M. Flanagan15Charles-Antoine Collins-Fekete16Department of Medical Physics and Biomedical Engineering, University College LondonDepartment of Medical Physics and Biomedical Engineering, University College LondonDepartment of Medical Physics and Biomedical Engineering, University College LondonResearch Department of Pathology, University College London Cancer InstituteResearch Department of Pathology, University College London Cancer InstituteCellular and Molecular Pathology, Royal National Orthopaedic Hospital NHS Foundation TrustResearch Department of Pathology, University College London Cancer InstituteResearch Department of Pathology, University College London Cancer InstituteResearch Department of Pathology, University College London Cancer InstituteDepartment of Neurology, Rutgers Biomedical and Health SciencesResearch Department of Pathology, University College London Cancer InstituteResearch Department of Pathology, University College London Cancer InstituteDepartment of Orthopaedics, The Robert Jones and Agnes Hunt Orthopaedic HospitalDepartment of Medical Physics and Biomedical Engineering, University College LondonDepartment of Medical Physics and Biomedical Engineering, University College LondonResearch Department of Pathology, University College London Cancer InstituteDepartment of Medical Physics and Biomedical Engineering, University College LondonAbstract Mitotic activity is an important feature for grading several cancer types. However, counting mitotic figures (cells in division) is a time-consuming and laborious task prone to inter-observer variation. Inaccurate recognition of MFs can lead to incorrect grading and hence potential suboptimal treatment. This study presents an artificial intelligence-based approach to detect mitotic figures in digitised whole-slide images stained with haematoxylin and eosin. Advances in this area are hampered by the small size and variety of datasets available. To address this, we create the largest dataset of mitotic figures (N = 74,620), combining an in-house dataset of soft tissue tumours with five open-source datasets. We then employ a two-stage framework, named the Optimised Mitoses Generator Network (OMG-Net), to identify mitotic figures. This framework first deploys the Segment Anything Model to automatically outline cells, followed by an adapted ResNet18 that distinguishes mitotic figures. OMG-Net achieves an F1 score of 0.84 in detecting pan-cancer mitotic figures, including human breast carcinoma, neuroendocrine tumours, and melanoma. It outperforms previous state-of-the-art models in hold-out test sets. To summarise, our study introduces a generalisable data creation and curation pipeline and a high-performance detection model, which can largely contribute to the field of computer-aided mitotic figure detection.https://doi.org/10.1038/s42003-024-07398-6
spellingShingle Zhuoyan Shen
Mikaël Simard
Douglas Brand
Vanghelita Andrei
Ali Al-Khader
Fatine Oumlil
Katherine Trevers
Thomas Butters
Simon Haefliger
Eleanna Kara
Fernanda Amary
Roberto Tirabosco
Paul Cool
Gary Royle
Maria A. Hawkins
Adrienne M. Flanagan
Charles-Antoine Collins-Fekete
A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
Communications Biology
title A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
title_full A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
title_fullStr A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
title_full_unstemmed A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
title_short A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
title_sort deep learning framework deploying segment anything to detect pan cancer mitotic figures from haematoxylin and eosin stained slides
url https://doi.org/10.1038/s42003-024-07398-6
work_keys_str_mv AT zhuoyanshen adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT mikaelsimard adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT douglasbrand adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT vanghelitaandrei adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT alialkhader adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT fatineoumlil adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT katherinetrevers adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT thomasbutters adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT simonhaefliger adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT eleannakara adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT fernandaamary adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT robertotirabosco adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT paulcool adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT garyroyle adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT mariaahawkins adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT adriennemflanagan adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT charlesantoinecollinsfekete adeeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT zhuoyanshen deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT mikaelsimard deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT douglasbrand deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT vanghelitaandrei deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT alialkhader deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT fatineoumlil deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT katherinetrevers deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT thomasbutters deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT simonhaefliger deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT eleannakara deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT fernandaamary deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT robertotirabosco deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT paulcool deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT garyroyle deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT mariaahawkins deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT adriennemflanagan deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides
AT charlesantoinecollinsfekete deeplearningframeworkdeployingsegmentanythingtodetectpancancermitoticfiguresfromhaematoxylinandeosinstainedslides