Improve Glioma Grading via Graphs of High-Centrality Cells in Pathological Imaging
Poster No:
1101
Submission Type:
Abstract Submission
Authors:
Lin Zhang1, Bowen Xin2, Xiuying Wang3
Institutions:
1the University of Sydney and Australian e-Health Research Centre, CSIRO, Sydney, Australia, 2Australian e-Health Research Centre, CSIRO, Sydney, Australia, 3School of Computer Science, The University of Sydney, Sydney, Australia
First Author:
Co-Author(s):
Introduction:
Glioblastoma (GBM), the leading cause of cancer death, is the most common primary cancer in the central nervous system (Ma et al., 2023). Quantifying cell distribution within biological tissues spatially is crucial to understanding tumour growth (Jiang et al., 2024). Analyzing centrality in a biological network can identify which nodes and interactions influence the network most. Nodes with high centrality are the most important biological signals (Uthamacumaran & Craig, 2022). Computational pathology has great potential for assisting with grading to improve cancer pathology. Although several computational pathology-based studies investigate the interactions among all-cells and cells with individual types (Abbas et al., 2023), the effect of high-centrality (HC) cell interaction has not been addressed. In this study, we hypothesize that we can improve the glioma grading by removing noise edges in the all-cell network using graphs based on HC cells only, and we present a novel framework for HC cell graph analysis.
Methods:
In the proposed HC cell analysis, we first segment cells and extract cell morphologic features in Whole Slides Images (WSI) using QuPath(Bankhead et al., 2017) (Fig 1A). Secondly, we build a graph using the radius-based method, where all nodes within the distance threshold are considered connected. Then, we compute the eigenvector centrality of each cell, which measures the influence of the cell in the network. Only the cells with top 20% HC values in the WSI are preserved in the graph (Fig 1B). Lastly, we extract HC cell interactions using the graph attention network and aggregate the morphologic features for grading (Fig 1C).
Results:
The proposed framework was evaluated on 200 GBM and 200 low-grade glioma (LGG) randomly selected from TCGA public dataset (https://www.cancer.gov/tcga). The dataset was split into training (60%), validation (20%) and test (20%). Our HC graph framework is compared with all-cell graph frameworks with common GNN models, including GraphSAGE (Hamilton et al., 2017), GCN (Kipf & Welling, 2016) and GAT (Veliˇckov´c et al., 2017). Our framework outperformed all all-cell graph frameworks, as shown in Table I.
Conclusions:
This research proposes an HCCSA framework that identifies HC cells, removes noise edges in the all-cell network, and constructs graphs based on HC cells only to improve the glioma grading. The HCCSAF construct HC graph performs better than all-cell graph for the GBM grading.
Modeling and Analysis Methods:
Classification and Predictive Modeling 1
Connectivity (eg. functional, effective, structural) 2
Keywords:
Cellular
Data analysis
Machine Learning
Structures
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
Bankhead, P., et al. (2017). Qupath: Open source software for digital pathology image analysis. Scientific reports, 7(1), 1–7.
Hamilton, W., et al. (2017). Inductive representation learning on large graphs. Advances in neural information processing systems, 30.
Jiang, J., et al. (2024). Meti: Deep profiling of tumor ecosystems by integrating cell morphology and spatial transcriptomics. Nature Communications, 15(1), 7312.
Kipf, T. N., et al. (2016). Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907.
Ma, J., et al. (2023). Tumor microenvironment targeting system for glioma treatment via fusion cell membrane coating nanotechnology. Biomaterials, 295, 122026.
Uthamacumaran, A., & Craig, M. (2022). Algorithmic reconstruction of glioblastoma network complexity. Iscience, 25(5).
Veliˇckovi´c, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., & Bengio, Y. (2017). Graph attention networks. arXiv preprint arXiv:1710.10903.