🤖 AI Summary
Modeling long-range spatial dependencies in multi-label thoracic CT abnormality classification remains challenging due to the computational inefficiency of 3D CNNs and the heavy reliance of Vision Transformers (ViTs) on large-scale pretraining.
Method: We propose a lightweight, graph-structured 2.5D approach: modeling CT volumes as weighted graphs whose nodes are axial slice triplets; employing spectral graph convolution to capture inter-slice long-range dependencies; and designing dedicated node representation learning, edge weighting, and multi-strategy graph aggregation mechanisms—eliminating the need for 3D convolutions or domain-specific large-scale pretraining.
Contribution/Results: Our method achieves competitive classification performance with mainstream visual encoders across three independent multi-institutional datasets, significantly improving cross-center generalization. It further demonstrates strong transferability to radiology report generation and abdominal CT analysis, validating its effectiveness, robustness, and scalability.
📝 Abstract
With the growing volume of CT examinations, there is an increasing demand for automated tools such as organ segmentation, abnormality detection, and report generation to support radiologists in managing their clinical workload. Multi-label classification of 3D Chest CT scans remains a critical yet challenging problem due to the complex spatial relationships inherent in volumetric data and the wide variability of abnormalities. Existing methods based on 3D convolutional neural networks struggle to capture long-range dependencies, while Vision Transformers often require extensive pre-training on large-scale, domain-specific datasets to perform competitively. In this work, we propose a 2.5D alternative by introducing a new graph-based framework that represents 3D CT volumes as structured graphs, where axial slice triplets serve as nodes processed through spectral graph convolution, enabling the model to reason over inter-slice dependencies while maintaining complexity compatible with clinical deployment. Our method, trained and evaluated on 3 datasets from independent institutions, achieves strong cross-dataset generalization, and shows competitive performance compared to state-of-the-art visual encoders. We further conduct comprehensive ablation studies to evaluate the impact of various aggregation strategies, edge-weighting schemes, and graph connectivity patterns. Additionally, we demonstrate the broader applicability of our approach through transfer experiments on automated radiology report generation and abdominal CT data.\ This work extends our previous contribution presented at the MICCAI 2025 EMERGE Workshop.