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

TL;DR

This paper introduces a self-supervised latent diffusion model guided by rich semantic embeddings to generate synthetic data, enhancing breast cancer classification accuracy in deep ultraviolet surface images.

Contribution

The study presents a novel SSL-guided LDM that produces high-quality synthetic training data, improving deep learning performance in breast cancer margin assessment.

Findings

Achieved 96.47% accuracy in classification

Reduced FID score to 45.72, indicating high-quality synthetic data

Outperformed class-conditioned baselines significantly

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.