A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides
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.
Authors
- Gary Royle (ORCID: https://orcid.org/0000-0002-8548-7684)
- M. Hawkins (ORCID: https://orcid.org/0000-0002-6669-0628)
- Katherine Trevers (ORCID: https://orcid.org/0000-0002-1254-9469)
- Eleanna Kara (ORCID: https://orcid.org/0000-0001-5428-6743)
- Mikaël Simard (ORCID: https://orcid.org/0000-0003-0610-6319)
- Fernanda Amary (ORCID: https://orcid.org/0000-0002-6707-8246)
- Adrienne M. Flanagan (ORCID: https://orcid.org/0000-0002-2832-1303)
- Ali Al Khader (ORCID: https://orcid.org/0000-0001-8453-3629)
- Thomas Butters
- Douglas Brand (ORCID: https://orcid.org/0000-0003-1130-4520)
- Paul Cool (ORCID: https://orcid.org/0000-0002-4985-3085)
- Simon Haefliger (ORCID: https://orcid.org/0000-0002-4227-6051)
- Charles‐Antoine Collins‐Fekete (ORCID: https://orcid.org/0000-0003-2464-8795)
- Vanghelita Andrei (ORCID: https://orcid.org/0000-0002-2162-2082)
- Roberto Tirabosco
- Fatine Oumlil
- Zhuoyan Shen (ORCID: https://orcid.org/0009-0002-8806-1379)
Institutions
- Rutgers, The State University of New Jersey (US)
- University College London Hospitals NHS Foundation Trust (GB)
- Robert Jones and Agnes Hunt Orthopaedic Hospital (GB)
- University Hospital of Basel (CH)
- Royal National Orthopaedic Hospital (GB)
- London Cancer (GB)
- University College London (GB)
- Keele University (GB)
Publication Details
- Journal
- Communications Biology
- Published
- 2024-12-19
- DOI
- https://doi.org/10.1038/s42003-024-07398-6
- Citations
- 8
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 2.78
Funders
- Sarcoma UK
- UK Research and Innovation
- Cancer Research UK
- National Institute for Health and Care Research
- Pathological Society of Great Britain and Ireland
- Jean Shanks Foundation
- University College London Hospitals NHS Foundation Trust
- Engineering and Physical Sciences Research Council