Self-learned representation-guided latent diffusion model for breast cancer classification in deep ultraviolet whole surface images

2026-01-16

AI Summary

This study addresses the scarcity of high-quality, deeply annotated ultraviolet pathology images in breast-conserving surgery, which limits the performance of deep learning models for intraoperative margin assessment. To overcome this challenge, the work proposes a novel approach that, for the first time, injects DINO self-supervised representations as semantic guidance into a Latent Diffusion Model (LDM) to generate high-fidelity synthetic images rich in cellular structural semantics. The generated images are combined with real data to fine-tune a Vision Transformer, and whole-slide-level breast cancer classification is achieved through patch-wise prediction aggregation. Evaluated via five-fold cross-validation, the model attains an accuracy of 96.47% and reduces the Fréchet Inception Distance (FID) to 45.72, significantly outperforming class-conditional baselines and demonstrating the effectiveness of the proposed method in enhancing both synthetic image quality and downstream task performance.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Breast-Conserving Surgery (BCS) requires precise intraoperative margin assessment to preserve healthy tissue. Deep Ultraviolet Fluorescence Scanning Microscopy (DUV-FSM) offers rapid, high-resolution surface imaging for this purpose; however, the scarcity of annotated DUV data hinders the training of robust deep learning models. To address this, we propose an Self-Supervised Learning (SSL)-guided Latent Diffusion Model (LDM) to generate high-quality synthetic training patches. By guiding the LDM with embeddings from a fine-tuned DINO teacher, we inject rich semantic details of cellular structures into the synthetic data. We combine real and synthetic patches to fine-tune a Vision Transformer (ViT), utilizing patch prediction aggregation for WSI-level classification. Experiments using 5-fold cross-validation demonstrate that our method achieves 96.47 % accuracy and reduces the FID score to 45.72, significantly outperforming class-conditioned baselines.

Problem

• breast cancer classification
• intraoperative margin assessment
• Deep Ultraviolet Fluorescence Scanning Microscopy
• annotated data scarcity
• whole surface imaging

Innovation

• Self-Supervised Learning
• Latent Diffusion Model
• DINO
• Vision Transformer
• Synthetic Data Generation