Virtual formalin-fixed and paraffin-embedded staining of fresh brain tissue via stimulated Raman CycleGAN model

Abstract

Intraoperative histology is essential for surgical guidance and decision-making. However, frozen-sectioned hematoxylin and eosin (H&E) staining suffers from degraded accuracy, whereas the gold-standard formalin-fixed and paraffin-embedded (FFPE) H&E is too lengthy for intraoperative use. Stimulated Raman scattering (SRS) microscopy has shown rapid histology of brain tissue with lipid/protein contrast but is challenging to yield images identical to nucleic acid–/protein-based FFPE stains interpretable to pathologists. Here, we report the development of a semi-supervised stimulated Raman CycleGAN model to convert fresh-tissue SRS images to H&E stains using unpaired training data. Within 3 minutes, stimulated Raman virtual histology (SRVH) results that matched perfectly with true H&E could be generated. A blind validation indicated that board-certified neuropathologists are able to differentiate histologic subtypes of human glioma on SRVH but hardly on conventional SRS images. SRVH may provide intraoperative diagnosis superior to frozen H&E in both speed and accuracy, extendable to other types of solid tumors.

In vivo organoid growth monitoring by stimulated Raman histology

“…Due to its chemical sensitivity, virtual histology images based on SRS are not deduced from any AI processes but are generated directly from the chemical spatial differences between CH 2 and CH 3 images on solid physical ground. AI could directly benefit from the recent work that has shown an improvement of SRH images to the level of formalin-fixed and paraffin-embedded 34 .…”

Stain-Free Histopathology with Stimulated Raman Scattering Microscopy

“…Deep-learning models provide improved capability to solve these issues. Liu et al 138 developed a modified CycleGAN model with a combined strongly and weakly supervised strategy that uses unpaired SRS and an H&E image (Figure 9c). With the stimulated Raman CycleGAN (SRC-GAN) model, they were able to generate virtual H&E stains from fresh brain tissue SRS images with formalin-fixed and paraffin-embedded (FFPE) quality to differentiate histologic subtypes of brain tumor.…”

“…Copyright 2023 The Authors.) (c) Stimulated Raman CycleGAN (SRC-GAN) model for virtual FFPE staining of fresh human brain tissues . (Reproduced with permission from Liu, Z.; Chen, L.; Cheng, H.; Ao, J.; Xiong, J.; Liu, X.; Chen, Y.; Mao, Y.; Ji, M. Virtual formalin-fixed and paraffin-embedded staining of fresh brain tissue via stimulated Raman CycleGAN model.…”

“…(Reproduced with permission from Liu, Z.; Chen, L.; Cheng, H.; Ao, J.; Xiong, J.; Liu, X.; Chen, Y.; Mao, Y.; Ji, M. Virtual formalin-fixed and paraffin-embedded staining of fresh brain tissue via stimulated Raman CycleGAN model. Science Advances 2024 , 10 (13), eadn3426 (ref ). Copyright 2024 the Authors.…”

His-MMDM: Multi-domain and Multi-omics Translation of Histopathology Images with Diffusion Models

“…Concurrently, visual language foundation models and copilot systems are also being developed in computational pathology to automate medical visual question answering 16,17, #1774,18,19 . As for image-to-image translation, dedicated models have been developed for various common tasks such as color normalization 20 , stain transfer 21 , virtual staining [21][22][23][24] , and transforming cryosectioned or stimulated Raman images to formalin-fixed paraffin-embedded (FFPE) ones 25,26 .…”