GraphDerm: Fusing Imaging, Physical Scale, and Metadata in a Population-Graph Classifier for Dermoscopic Lesions

2025-09-14

AI Summary

Skin lesion classification from dermoscopic images is limited by neglecting patient metadata (e.g., age, sex, anatomical site) and physical scale information. Method: This paper proposes a population graph modeling framework that jointly integrates dermoscopic imagery, millimeter-scale measurements, and clinical metadata. It is the first to apply graph neural networks (GNNs) to large-scale ISIC data: U-Net segments lesions and rulers; a 1D-CNN regresses pixel-to-millimeter ratios; EfficientNet-B3 extracts node features; and a weighted graph is constructed based on geometric and metadata similarity, followed by spectral graph convolution for semi-supervised classification. Results: The model achieves an AUC of 0.9812; its sparse variant (25% edges) retains 0.9788 AUC—significantly outperforming unimodal image baselines (0.9440). Lesion and ruler segmentation achieves Dice > 0.90, and scale regression error is low.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Computer Vision: Segmentation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Large-scale security measurements

Abstract

Introduction. Dermoscopy aids melanoma triage, yet image-only AI often ignores patient metadata (age, sex, site) and the physical scale needed for geometric analysis. We present GraphDerm, a population-graph framework that fuses imaging, millimeter-scale calibration, and metadata for multiclass dermoscopic classification, to the best of our knowledge the first ISIC-scale application of GNNs to dermoscopy. Methods. We curate ISIC 2018/2019, synthesize ruler-embedded images with exact masks, and train U-Nets (SE-ResNet-18) for lesion and ruler segmentation. Pixels-per-millimeter are regressed from the ruler-mask two-point correlation via a lightweight 1D-CNN. From lesion masks we compute real-scale descriptors (area, perimeter, radius of gyration). Node features use EfficientNet-B3; edges encode metadata/geometry similarity (fully weighted or thresholded). A spectral GNN performs semi-supervised node classification; an image-only ANN is the baseline. Results. Ruler and lesion segmentation reach Dice 0.904 and 0.908; scale regression attains MAE 1.5 px (RMSE 6.6). The graph attains AUC 0.9812, with a thresholded variant using about 25% of edges preserving AUC 0.9788 (vs. 0.9440 for the image-only baseline); per-class AUCs typically fall in the 0.97-0.99 range. Conclusion. Unifying calibrated scale, lesion geometry, and metadata in a population graph yields substantial gains over image-only pipelines on ISIC-2019. Sparser graphs retain near-optimal accuracy, suggesting efficient deployment. Scale-aware, graph-based AI is a promising direction for dermoscopic decision support; future work will refine learned edge semantics and evaluate on broader curated benchmarks.

Problem

• Classifying dermoscopic lesions using multimodal data fusion
• Integrating imaging, physical scale, and patient metadata
• Improving accuracy beyond image-only AI approaches

Innovation

• Population-graph framework fuses imaging, scale, metadata
• U-Nets segment lesions and rulers for scale calibration
• Spectral GNN performs semi-supervised node classification