Minimum Data, Maximum Impact: 20 annotated samples for explainable lung nodule classification

2025-08-01

AI Summary

To address the severe scarcity of pathology-oriented visual attribute annotations (e.g., spiculation, lobulation, vacuole) hindering the performance of interpretable models for pulmonary nodule classification, this work proposes a few-shot attribute-conditional diffusion model for medical image synthesis. Leveraging only 20 real annotated CT scans, we construct a diffusion generative model capable of precisely controlling clinically relevant pathological attributes while producing high-fidelity, semantically consistent nodule images. The synthesized data augment training of an interpretable classification model, preserving decision transparency while substantially improving performance: attribute prediction accuracy increases by 13.4%, and benign/malignant classification accuracy improves by 1.8%. To our knowledge, this is the first study to apply ultra-low-shot conditional diffusion modeling to generate radiologist-defined clinical visual attributes. Our approach establishes a novel paradigm for deploying few-shot interpretable AI in medical imaging.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transparent, Interpretable, Explainable ML
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Classification models that provide human-interpretable explanations enhance clinicians' trust and usability in medical image diagnosis. One research focus is the integration and prediction of pathology-related visual attributes used by radiologists alongside the diagnosis, aligning AI decision-making with clinical reasoning. Radiologists use attributes like shape and texture as established diagnostic criteria and mirroring these in AI decision-making both enhances transparency and enables explicit validation of model outputs. However, the adoption of such models is limited by the scarcity of large-scale medical image datasets annotated with these attributes. To address this challenge, we propose synthesizing attribute-annotated data using a generative model. We enhance the Diffusion Model with attribute conditioning and train it using only 20 attribute-labeled lung nodule samples from the LIDC-IDRI dataset. Incorporating its generated images into the training of an explainable model boosts performance, increasing attribute prediction accuracy by 13.4% and target prediction accuracy by 1.8% compared to training with only the small real attribute-annotated dataset. This work highlights the potential of synthetic data to overcome dataset limitations, enhancing the applicability of explainable models in medical image analysis.

Problem

• Lack of large medical datasets with annotated visual attributes
• Need for explainable AI models aligning with clinical reasoning
• Improving accuracy with minimal real annotated lung nodule samples

Innovation

• Attribute-conditioned Diffusion Model for data synthesis
• 20 annotated samples for explainable classification
• Synthetic data boosts accuracy by 13.4%