MAGE: Color-Invariant and Spatial Knowledge Distillation for Gastric Neoplasm Classification

2026-07-14

AI Summary

This study addresses the challenge of distinguishing gastric adenomas from early-stage carcinomas in endoscopic images, where high morphological similarity and ambiguous boundaries lead to segmentation errors, lack of global context, and spurious correlations based on color or texture. To overcome these limitations, the authors propose the MAGE framework, which employs achromatic masked images to guide a local expert branch, compelling the model to focus on structural features. By integrating dual-objective knowledge distillation—leveraging both classification logits and spatial attention maps—the method injects morphological priors into the backbone network. Notably, MAGE operates without requiring masks during inference, preserving original color information for clinical interpretability while enhancing robustness and explainability. Evaluated on clinical gastroscopic datasets, MAGE significantly outperforms detectors such as YOLO and classification models like Swin-Transformer, and produces reliable attention visualizations.

Abstract

Accurate differentiation between gastric adenoma and carcinoma during endoscopy is critical for clinical decision-making. Yet, this task is highly challenging due to high inter-class similarity and ambiguous boundaries between the two classes. Existing ROI-based classification methods often suffer from detection/segmentation error propagation and loss of surrounding global context. In contrast, full-image classification lacks the necessary spatial focus. Furthermore, we observe that deep neural networks gravitate towards domain-specific texture biases(e.g. bleeding, lighting artifacts), often causing models to predict based on spurious correlations instead of intrinsic morphological features. To address these limitations, we propose a novel framework, Masked Achromatic Guidance Expert (MAGE). During training, we introduce an auxiliary local expert branch trained on masked achromatic views of the neoplasm. By suppressing background context and color, this branch is forced to learn highly discriminative, purely structural features. We then employ a dual-objective distillation strategy, transferring both classification logits and spatial attention maps to provide implicit spatial supervision to the main branch that receives full WLI as input. This dual-objective distillation forces the model to ground its predictions in morphology rather than relying on shortcuts, while still retaining clinically relevant color cues. At inference time, our deployable model operates on images without annotated masks, ensuring real-time deployability . Extensive experiments on a clinical gastric endoscopy dataset show that our method significantly outperforms existing detection-based methodologies (e.g. YOLO) and classification-based methodologies (e.g. Swin-Transformer), providing not only superior classification performance but also interpretable attention maps for clinical reliability.

Problem

• gastric neoplasm classification
• inter-class similarity
• spurious correlations
• spatial context
• color bias

Innovation

• knowledge distillation
• color-invariant learning
• spatial attention
• gastric neoplasm classification
• morphological feature learning