🦉 Contributors: Yifei Sun (22' HDU-ITMO Undergraduate), Jincheng Li (23' NTU Undergraduate).

🎓 DeepWiki: Generating GitHub Knowledge Base Documentation in One Click.

📦 Other resources: [1] Paper List for Medical Anomaly Detection, [2] Bone Suppression in Chest X-Rays: A Deep Survey, [3] Paper List for Prototypical Learning, [4] Paper List for Cell Detection, [5] Medical-AI-Guide, [6] Paper List for Medical Reasoning Large Language Models.

Welcome to join us by contacting: szhsxhsyf@hdu.edu.cn.

📇 Contents

• 1. Background
• 2. Related Work
• 3. Datasets
• 4. Metrics

🧑🏻‍🏫 1. Background

Traditional histological staining (e.g., H&E, immunofluorescence) is essential for visualizing tissue structures in pathology but involves complex chemical processes, time-consuming protocols, and labor-intensive workflows. Label-free imaging techniques (e.g., autofluorescence, quantitative phase imaging, Raman microscopy) capture intrinsic optical properties of tissues without exogenous labels, preserving samples and accelerating imaging. However, these methods often lack the diagnostic contrast provided by stains.

Virtual staining leverages Deep Learning (DL) to computationally transform label-free tissue images into virtually stained counterparts that resemble chemically stained images. This approach:

• Eliminates physical staining, reducing costs, preparation time, and chemical waste.
• Enables retrospective analysis of archived label-free data.
• Facilitates multi-stain synthesis from a single scan.
• Preserves tissue integrity for downstream molecular analysis.

Recent advances in DL models (e.g., GANs, U-Nets, diffusion models) have demonstrated remarkable accuracy in predicting stain-specific features directly from label-free inputs, paving the way for scalable, stain-free computational pathology.

✍🏻 2. Related Work

Source	Year	Title	Graph	Code
Nature Communications	2025	Pixel Super-Resolved Virtual Staining of Label-Free Tissue Using Diffusion Models
Nature Communications	2024	Virtual Histological Staining of Unlabeled Autopsy Tissue
Trends in Biotechnology	2024	Virtual Staining for Histology by Deep Learning
Modern Pathology	2024	Virtual Staining of Nonfixed Tissue Histology
Light-Science & Applications	2023	Deep Learning-Enabled Virtual Histological Staining of Biological Samples
Nature Biomedical Engineering	2019	Virtual Histological Staining of Unlabelled Tissue-Autofluorescence Images via Deep Learning

🔢 3. Datasets

💯 4. Metrics

Metric	Full Name	Formula	Purpose
MAE	Mean Absolute Error	$$\frac{1}{n}\sum_{i=1}^{n}|y_i - \hat{y}_i|$$	Low-Level Fidelity
MSE	Mean Squared Error	$$\frac{1}{n}\sum_{i=1}^{n}(y_i - \hat{y}_i)^2$$	Low-Level Fidelity
PSNR	Peak Signal-to-Noise Ratio	$$20 \cdot \log_{10}\left(\frac{\text{MAX}_I}{\sqrt{\text{MSE}}}\right)$$	Low-Level Fidelity
SSIM	Structural Similarity Index Measure	$$\frac{(2\mu_y\mu_{\hat{y}} + C_1)(2\sigma_{y\hat{y}} + C_2)}{(\mu_y^2 + \mu_{\hat{y}}^2 + C_1)(\sigma_y^2 + \sigma_{\hat{y}}^2 + C_2)}$$	Structural Integrity
MS-SSIM	Multi-Scale SSIM		Structural Integrity
FID	Fréchet Inception Distance	$$|\mu_r - \mu_g|^2 + \text{Tr}(\Sigma_r + \Sigma_g - 2\sqrt{\Sigma_r\Sigma_g})$$	Perceptual Realism
IS	Inception Score	$$\exp\left(\mathbb{E}_{\hat{y}} \text{KL}(p(y|\hat{y}) | p(y))\right)$$	Perceptual Realism
LPIPS	Learned Perceptual Image Patch Similarity		Perceptual Realism

💞 Citation

@misc{sun2025,
  author = {Sun, Yifei and Li, Jincheng},
  title = {Deep Learning for Virtual Staining of Label-Free Tissue: A Survey},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/diaoquesang/DL4VS}}
}

🥰 Star History