Context-aware Skin Cancer Epithelial Cell Classification with Scalable Graph Transformers

Lucas Sancéré; Noémie Moreau; Katarzyna Bozek

cs.CVFeb 17, 2026

Context-aware Skin Cancer Epithelial Cell Classification with Scalable Graph Transformers

Authors: Lucas Sancéré, Noémie Moreau, Katarzyna Bozek

Organizations: Faculty of Mathematics and Natural Sciences, University of Cologne, Cologne, North Rhine-Westphalia, Germany · Institute for Biomedical Informatics, Faculty of Medicine and University Hospital Cologne, University of Cologne, North Rhine-Westphalia, Germany · Center for Molecular Medicine Cologne (CMMC), Faculty of Medicine and University Hospital Cologne, University of Cologne, North Rhine-Westphalia, Germany · Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, North Rhine-Westphalia, Germany

Abstract

Whole-slide images (WSIs) from cancer patients contain rich information that can be used for medical diagnosis or to follow treatment progress. To automate their analysis, numerous deep learning methods based on convolutional neural networks and Vision Transformers have been developed and have achieved strong performance in segmentation and classification tasks. However, due to the large size and complex cellular organization of WSIs, these models rely on patch-based representations, losing vital tissue-level context. We propose using scalable Graph Transformers on a full-WSI cell graph for classification. We evaluate this methodology on a challenging task: the classification of healthy versus tumor epithelial cells in cutaneous squamous cell carcinoma (cSCC), where both cell types exhibit very similar morphologies and are therefore difficult to differentiate for image-based approaches. We first compared image-based and graph-based methods on a single WSI. Graph Transformer models SGFormer and DIFFormer achieved balanced accuracies of $85.2 \pm 1.5$ ( $\pm$ standard error) and $85.1 \pm 2.5$ in 3-fold cross-validation, respectively, whereas the best image-based method reached $81.2 \pm 3.0$ . By evaluating several node feature configurations, we found that the most informative representation combined morphological and texture features as well as the cell classes of non-epithelial cells, highlighting the importance of the surrounding cellular context. We then extended our work to train on several WSIs from several patients. To address the computational constraints of image-based models, we extracted four $2560 \times 2560$ pixel patches from each image and converted them into graphs. In this setting, DIFFormer achieved a balanced accuracy of $83.6 \pm 1.9$ (3-fold cross-validation), while the state-of-the-art image-based model CellViT256 reached $78.1 \pm 0.5$ .

Figures & tables

Figure 1 : Description of WSI-Graph. (a) Steps to generate WSI-Graph. First a WSI from cSCC patient is segmented using SCC Hovernet and tumor regions are annotated by an expert to refine segmentation. From this segmentation map we build a graph, simplify it around anchor nodes and optionally split it with K-means on centroid coordinate features. (b) Zoom into the graph, edges generated with threshold distance $r_{0}=50\operatorname{pixels}$ corresponding to $r_{0}\approx$11.5\text{\,}\mathrm{\SIUnitSymbolMicro m}$$ are shown. (c) Number of edges, nodes, node features and instances of given cell classes before and after simplification (here $k=3$ max-hops simplification). — Figure 1 : Description of WSI-Graph. (a) Steps to generate WSI-Graph. First a WSI from cSCC patient is segmented using SCC Hovernet and tumor regions are annotated by an expert to refine segmentation. From this segmentation map we build a graph, simplify it around anchor nodes and optionally split it with K-means on centroid coordinate features. (b) Zoom into the graph, edges generated with threshold distance $r_{0}=50\operatorname{pixels}$ corresponding to $r_{0}\approx$ 11.5\text{,}\mathrm{\SIUnitSymbolMicro m}$$ are shown. (c) Number of edges, nodes, node features and instances of given cell classes before and after simplification (here $k=3$ max-hops simplification).

Figure 2 : Description of TILE-Graphs. (a) Steps to generate TILE-Graphs. First, patches from cSCC patient sample extracted from tumor epithelial and healthy epithelial regions are segmented using SCC Hovernet. Then from these segmentation maps 372 graphs are built. (b) TILE-Graphs dataset statistics. It includes 372 patches from 93 samples from 84 patients. The resulting 372 graphs are then split keeping graphs of the same patients in the same split during cross-validation.

Method	Subgraphs	Random Nodes
DIFFormer	85.2 $\pm$ 1.5	91.1 $\pm$ 0.1
SGFormer	85.1 $\pm$ 2.5	94.9 $\pm$ 0.2
SIGN	80.5 $\pm$ 2.8	84.8 $\pm$ 0.6
GAT	80.4 $\pm$ 1.0	73.3 $\pm$ 2.0
SGCMLP	80.4 $\pm$ 2.7	79.3 $\pm$ 0.4
NodeFormer	79.0 $\pm$ 0.5	85.0 $\pm$ 1.4

Table 1 : Comparison of GNN models on binary epithelial node classification performance on simplified WSI-Graph. Balanced accuracy (%) is reported as mean ± standard error over 3-fold cross-validation under subgraph-based and random node evaluation protocols and

k=3

max-hops simplification for the graph ( WSI-Graph (3) ).

Method	Training set	3-fold crossval
Hovernet	all patches	73.7 $\pm$ 1.4
	epithelial only	79.3 $\pm$ 2.3
CellViT256	all patches	76.5 $\pm$ 1.9
	epithelial only	81.2 $\pm$ 3.0
CellViT-SAM-B	all patches	OOM
	epithelial only	OOM

Table 2 : Comparison of image-based models on binary epithelial cell classification performance on WSI images. Balanced accuracy (%) is reported as mean ± standard error over 3-fold cross-validation for different methods and training set configurations, using image-based representations.

Node Features	z-score norm	Subgraphs	Random Nodes
morphology	no	67.8 $\pm$ 3.6	92.6 $\pm$ 0.3
morphology	yes	79.6 $\pm$ 2.4	94.0 $\pm$ 0.5
morphology & texture	no	70.6 $\pm$ 9.0	94.0 $\pm$ 0.9
morphology & texture	yes	84.2 $\pm$ 1.6	94.3 $\pm$ 0.6
morphology & cell class	no	73.6 $\pm$ 3.7	93.7 $\pm$ 0.5
morphology & cell class	yes	84.0 $\pm$ 2.8	94.5 $\pm$ 0.4

Table 3 : Impact of node features on binary node classification performance on simplified WSI-Graph. Balanced accuracy (%) is reported as mean ± standard error over 3-fold cross-validation for different node features under subgraph and random node evaluation protocols with SGFormer model and WSI-Graph (3) .

Graph Simplifications	Subgraphs	Random Nodes
No simplification	82.2 $\pm$ 2.9	94.6 $\pm$ 0.1
50 max-hops	83.3 $\pm$ 3.3	94.9 $\pm$ 0.4
10 max-hops	86.6 $\pm$ 2.2	95.0 $\pm$ 0.2
5 max-hops	82.5 $\pm$ 1.9	94.6 $\pm$ 0.5
4 max-hops	82.1 $\pm$ 4.3	95.0 $\pm$ 0.4
3 max-hops	85.1 $\pm$ 2.5	94.9 $\pm$ 0.2

Table 4 : Impact of graph simplification on binary node classification performance on simplified WSI-Graph. Balanced accuracy (%) is reported as mean ± standard error over 3-fold cross-validation for different maximum hop thresholds under subgraph and random node evaluation protocols with SGFormer model.

Method	3-fold crossval
CellViT256	78.1 $\pm$ 0.5
DIFFormer	83.6 $\pm$ 1.9
NodeFormer	78.7 $\pm$ 4.4
SGCMLP	66.4 $\pm$ 1.1
SIGN	66.3 $\pm$ 1.0
SGC	63.8 $\pm$ 1.2

Table 5 : Comparison of graph-based and image-based models on binary epithelial cell classification performance on TILE-Graphs and baseline dataset. Balanced accuracy (%) is reported as mean ± standard error over 3-fold cross-validation for different methods and training set configurations.

Explore similar work

Geometry-Aware Superpixel Graph Transformer with Metadata for Skin Lesion Classification
Jun 18, 2026Muhammad Azeem, Tanveer Hussain, Amr Ahmed +1Skin Lesion Classification Melanoma
Edge Hill University, Ormskirk, Lancashire, L39 4QP, United Kingdom
From Multi-Resolution Cells to Gigapixel Whole Slide Images Foundation Model for Computational Pathology
Aug 4, 2026Basit Alawode, Moshira Ali Abdalla, Dwarikanath Mahapatra +2Whole-Slide Images Computational Pathology
Khalifa University of Science and Technology, UAE · University of Western Australia, Australia
CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction
Mar 3, 2025Rustin Soraki, Huayu Wang, Sitong Liu +2Whole-Slide Images Computational Pathology
University of Washington, Seattle, WA · University of California, Los Angeles, CA

Jun 18, 2026, cs.CVJ/K move · Enter open · S save

Geometry-Aware Superpixel Graph Transformer with Metadata for Skin Lesion Classification

Muhammad Azeem, Tanveer Hussain, Amr Ahmed, Ardhendu Behera

Edge Hill University, Ormskirk, Lancashire, L39 4QP, United Kingdom

Skin Lesion Classification Melanoma Lesion Multimodal Fusion Metadata

Open paper AskPDF arXiv

Automated skin cancer classification from dermoscopic images remains challenging due to heterogeneous lesion structure, strong intra-class variability, and subtle visual differences between benign and malignant cases. Existing CNN/ViT pipelines typically rely on global or patch-level features and often combine patient metadata via late fusion, which limits spatially grounded multimodal reasoning. We present a novel region-based graph learning framework that explicitly models lesions as graphs of spatially coherent superpixel regions represented as frozen CNN features. To capture fine-grained lesion arrangements, we encode inter-regional geometry as edge attributes and introduce a dedicated metadata context node connected to all regions, providing structured integration of demographic/clinical variables within the same relational space. Node representations are updated using our edge-aware graph transformer followed by attention-driven propagation, and a final graph-level embedding for benign-malignant classification. Experiments on four public benchmarks demonstrate that explicit region-level relational modeling and graph-native multimodal fusion yield consistent gains over the state-of-the-art. Consequently, we establish a new graph-centric perspective in which CNN features are modeled as relational nodes and improved through contextual integration, yielding more expressive and robust classifications.

Abstract

Explore similar work

Geometry-Aware Superpixel Graph Transformer with Metadata for Skin Lesion Classification

From Multi-Resolution Cells to Gigapixel Whole Slide Images Foundation Model for Computational Pathology

CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction