Godson, L., Alemi, N., Nsengimana, J. et al. (6 more authors) (2024) Multi-level Graph Representations of Melanoma Whole Slide Images for Identifying Immune Subgroups. In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). International Conference on Medical Image Computing and Computer-Assisted Intervention, 08/10/2023 - 12/10/2023, Vancouver, Canada. Springer Nature , pp. 85-96. ISBN 978-3-031-55087-4
Abstract
Stratifying melanoma patients into immune subgroups is important for understanding patient outcomes and treatment options. Current weakly supervised classification methods often involve dividing digitised whole slide images into patches, which leads to the loss of important contextual diagnostic information. Here, we propose using graph attention neural networks, which utilise graph representations of whole slide images, to introduce context to classifications. In addition, we present a novel hierarchical graph approach, which leverages histopathological features from multiple resolutions to improve on state-of-the-art (SOTA) multiple instance learning (MIL) methods. We achieve a mean test area under the curve metric of 0.80 for classifying low and high immune melanoma subtypes, using multi-level and 20x patch graph representations of whole slide images, compared to 0.77 when using SOTA MIL methods. Our experimental results comprehensively show how our whole slide image graph representation is a valuable improvement on the MIL paradigm and could help to determine early-stage prognostic markers and stratify melanoma patients for effective treatments. Code is available at https://github.com/lucyOCg/graph_mil_project/.
Metadata
Item Type: | Proceedings Paper |
---|---|
Authors/Creators: |
|
Copyright, Publisher and Additional Information: | © 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG. This is an author produced version of a conference paper published in Lecture Notes in Computer Science. Uploaded in accordance with the publisher's self-archiving policy. |
Dates: |
|
Institution: | The University of Leeds |
Academic Units: | The University of Leeds > Faculty of Medicine and Health (Leeds) > School of Medicine (Leeds) > Leeds Institute of Medical Research (LIMR) > Division of Haematology and Immunology The University of Leeds > Faculty of Medicine and Health (Leeds) > School of Medicine (Leeds) > Leeds Institute of Cancer and Pathology (LICAP) The University of Leeds > Faculty of Engineering & Physical Sciences (Leeds) > School of Computing (Leeds) > Artificial Intelligence The University of Leeds > Faculty of Medicine and Health (Leeds) > School of Medicine (Leeds) > Leeds Institute of Medical Research (LIMR) > Division of Pathology and Data Analytics The University of Leeds > Faculty of Engineering & Physical Sciences (Leeds) > School of Computing (Leeds) |
Funding Information: | Funder Grant number Alan Turing Institute No Ext Ref MRC (Medical Research Council) MR/S001530/1 |
Depositing User: | Symplectic Publications |
Date Deposited: | 05 Jul 2024 14:27 |
Last Modified: | 10 Mar 2025 01:13 |
Published Version: | http://dx.doi.org/10.1007/978-3-031-55088-1_8 |
Status: | Published |
Publisher: | Springer Nature |
Identification Number: | 10.1007/978-3-031-55088-1_8 |
Open Archives Initiative ID (OAI ID): | oai:eprints.whiterose.ac.uk:214306 |