Related Experiment Video
Updated: Jun 20, 2026
A Static Self-Directed Method for Generating Brain Organoids from Human Embryonic Stem Cells
Published on: March 4, 2020
A Multi-Perspective Self-Supervised Generative Adversarial Network for FS to FFPE Stain Transfer
Abstract:
In clinical practice, frozen section (FS) images can be utilized to obtain the immediate pathological results of the patients in operation due to their fast production speed. However, compared with the formalin-fixed and paraffin-embedded (FFPE) images, the FS images greatly suffer from poor quality. Thus, it is of great significance to transfer the FS image to the FFPE one, which enables pathologists to observe high-quality images in operation. However, obtaining the paired FS and FFPE images is quite hard, so it is difficult to obtain accurate results using supervised methods. Apart from this, the FS to FFPE stain transfer faces many challenges. Firstly, the number and position of nuclei scattered throughout the image are hard to maintain during the transfer process. Secondly, transferring the blurry FS images to the clear FFPE ones is quite challenging. Thirdly, compared with the center regions of each patch, the edge regions are harder to transfer. To overcome these problems, a multi-perspective self-supervised GAN, incorporating three auxiliary tasks, is proposed to improve the performance of FS to FFPE stain transfer. Concretely, a nucleus consistency constraint is designed to enable the high-fidelity of nuclei, an FFPE guided image deblurring is proposed for improving the clarity, and a multi-field-of-view consistency constraint is designed to better generate the edge regions. Objective indicators and pathologists' evaluation for experiments on the five datasets across different countries have demonstrated the effectiveness of our method. In addition, the validation in the downstream task of microsatellite instability prediction has also proved the performance improvement by transferring the FS images to FFPE ones. Our code link is https://github.com/linyiyang98/Self-Supervised-FS2FFPE.git.
Insights
This study introduces a novel self-supervised Generative Adversarial Network (GAN) to enhance frozen section (FS) pathology images, improving clarity and nucleus fidelity for better surgical diagnosis. The method effectively converts low-quality FS images to high-quality formalin-fixed and paraffin-embedded (FFPE) equivalents.
Area of Science:
- Digital Pathology
- Medical Image Analysis
- Artificial Intelligence in Medicine
Background:
- Frozen section (FS) images provide rapid intraoperative pathological insights but suffer from poor quality compared to formalin-fixed and paraffin-embedded (FFPE) images.
- The scarcity of paired FS and FFPE images hinders supervised learning for stain transfer, a crucial step for improving image quality during surgery.
- Existing FS to FFPE stain transfer methods face challenges in maintaining nucleus integrity, deblurring images, and accurately generating edge regions.
Purpose of the Study:
- To develop an effective method for transferring FS images to FFPE equivalents, enabling pathologists to access high-quality images during operations.
- To address the limitations of supervised methods by proposing a self-supervised approach for FS to FFPE stain transfer.
- To improve nucleus consistency, image clarity, and the generation of edge regions in the transferred images.
Main Methods:
- A multi-perspective self-supervised Generative Adversarial Network (GAN) was designed, incorporating three auxiliary tasks.
- A nucleus consistency constraint was implemented to ensure high-fidelity nuclei preservation during image transfer.
- An FFPE-guided image deblurring technique and a multi-field-of-view consistency constraint were utilized to enhance image clarity and edge region generation.
Main Results:
- Objective metrics and pathologist evaluations across five international datasets confirmed the effectiveness of the proposed method.
- The method successfully improved nucleus fidelity, image clarity, and the quality of edge regions in FS to FFPE stain transfer.
- Validation in a downstream task of microsatellite instability prediction demonstrated performance improvements attributed to the enhanced image quality.
Conclusions:
- The proposed multi-perspective self-supervised GAN effectively addresses the challenges in FS to FFPE stain transfer, yielding high-quality pathological images.
- This approach offers a significant advancement for intraoperative pathological diagnosis by providing clearer, more reliable image data.
- The method's utility is further validated by its positive impact on downstream clinical prediction tasks.
Related Concept Videos
01:21Association Areas of the Cortex
Prefrontal Association Area: This area is located in the frontal lobe and is involved in planning, decision-making, and moderating social behavior. It connects with primary motor areas,...
01:22Biasing of FET
In an N-channel JFET, the structure consists of N-type material forming the channel on a P-type substrate, with the gate...