Diffusion Attention Expert Model for Predicting and Semi-automatic Localizing STAS in Lung Cancer Histopathological Images

2026-05-14

AI Summary

This study addresses the challenges of time-consuming diagnosis and frequent misdiagnosis or missed detection of spread through air spaces (STAS) in lung adenocarcinoma histopathological images by proposing a Diffusion Attention Expert Model (DAEM). DAEM introduces, for the first time, a diffusion attention mechanism into STAS detection, integrating a dual-branch multi-scale feature fusion architecture with quantitative analysis of the tumor microenvironment (TME). The model enables high-accuracy detection, semi-automatic localization, and quantification of the distance between STAS lesions and the primary tumor in both frozen and paraffin-embedded sections. Evaluated on an internal dataset, DAEM achieves AUCs of 0.8946 and 0.9112, respectively, and demonstrates strong generalizability across eight external centers. Moreover, it identifies several interpretable TME biomarkers, supporting precise intraoperative and postoperative decision-making and risk stratification.

Abstract

Accurate intraoperative and postoperative diagnosis of spread through air spaces (STAS) is essential for guiding surgical decisions and postoperative management in lung cancer. However, histopathological assessment is labor-intensive and is prone to missed or incorrect diagnoses. We propose a Diffusion Attention Expert Model (DAEM) to detect STAS in frozen sections (FSs) and paraffin sections (PSs). Its diffusion attention expert module leverages full attention aggregation to learn multi-scale features from histopathological images, while a dual-branch architecture strengthens multi-scale feature representation. On an internal dataset, DAEM achieves AUCs of 0.8946 for FSs and 0.9112 for PSs. Validation on external multi-center datasets from eight institutions demonstrates strong generalizability and interpretability. Using tumor microenvironment (TME) features in PSs, we further enable semi-automatic measurement of STAS location and its distance from the primary tumor. Several quantitative TME metrics are identified as potential biomarkers for STAS, including micropapillary-type STAS. Overall, DAEM offers a clinically actionable framework for STAS assessment by enabling accurate and interpretable detection on FSs and PSs, supporting postoperative risk stratification through quantitative TME-based analysis.

Problem

• STAS
• lung cancer
• histopathological images
• diagnosis
• tumor microenvironment

Innovation

• Diffusion Attention
• Multi-scale Feature Representation
• Semi-automatic Localization
• Tumor Microenvironment (TME)
• STAS Detection