Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Pouya Afshin, David Helminiak, Tianling Niu, Julie M. Jorns, Tina Yen, Bing Yu, Dong Hye Ye

TL;DR

This work tackles the challenge of scarce annotated deep ultraviolet WSI data for breast cancer margin assessment in Breast-Conserving Surgery by introducing an SSL-guided latent diffusion model. Using embeddings from a fine-tuned DINO teacher, the diffusion model generates high-fidelity synthetic DUV patches that augment real training data, which are then used to fine-tune a Vision Transformer for patch- and WSI-level classification. The SSL-guided synthesis yields more realistic textures and cellular structures, reflected in a lower Fréchet Inception Distance (FID = 45.72) and superior WSI accuracy (96.47%), sensitivity (96.46%), and specificity (96.36%) across 5-fold cross-validation, outperforming class-conditioned baselines. This approach provides a data-efficient pathway to robust breast cancer classification on DUV WSIs, with tangible potential to improve intraoperative margin assessment and reduce re-excision rates.

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.

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

TL;DR

This work tackles the challenge of scarce annotated deep ultraviolet WSI data for breast cancer margin assessment in Breast-Conserving Surgery by introducing an SSL-guided latent diffusion model. Using embeddings from a fine-tuned DINO teacher, the diffusion model generates high-fidelity synthetic DUV patches that augment real training data, which are then used to fine-tune a Vision Transformer for patch- and WSI-level classification. The SSL-guided synthesis yields more realistic textures and cellular structures, reflected in a lower Fréchet Inception Distance (FID = 45.72) and superior WSI accuracy (96.47%), sensitivity (96.46%), and specificity (96.36%) across 5-fold cross-validation, outperforming class-conditioned baselines. This approach provides a data-efficient pathway to robust breast cancer classification on DUV WSIs, with tangible potential to improve intraoperative margin assessment and reduce re-excision rates.

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.
Paper Structure (11 sections, 9 equations, 2 figures, 1 table)

This paper contains 11 sections, 9 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methodology
  3. SSL Embedding Extraction
  4. SSL-Guided LDM Training and Synthesis
  5. Patch-to-WSI Classification with ViT
  6. Experiment
  7. Dataset and Experimental Setup
  8. Visual Inspection
  9. Classification Performance
  10. Conclusion
  11. Compliance with Ethical Standards

Figures (2)

  • Figure 1: System model overview for DUV WSI classification with SSL-guided LDM augmentation. (a) Real training patches are encoded into SSL embeddings using a fine-tuned DINO teacher model. (b) LDM uses these embeddings to guide its U-Net denoiser via cross-attention during denoising, generating realistic synthetic patches to enrich the training set. (c) A ViT is fine-tuned on both real and synthetic patches for patch-level classification, and WSI-level predictions are obtained by aggregating patch predictions and applying a binary threshold.
  • Figure 2: Visualization of feature distributions for real and synthetic DUV WSI patches. SSL-guided synthesis better resembles the true feature distribution, producing sharper, more detailed images, whereas class-conditioned synthesis yields blurrier, less intricate results. Moreover, its feature space resembles the real feature space better than class-guided synthetic features.