Deep learning-enabled realistic virtual histology with ultraviolet photoacoustic remote sensing microscopy

The goal of oncologic surgeries is complete tumor resection, yet positive margins are frequently found postoperatively using gold standard H&E-stained histology methods. Frozen section analysis is sometimes performed for rapid intraoperative margin evaluation, albeit with known inaccuracies. Her...

Full description

Saved in:
Bibliographic Details
Published in:Nature communications 2023-09, Vol.14 (1), p.5967-5967, Article 5967
Main Authors: Martell, Matthew T., Haven, Nathaniel J. M., Cikaluk, Brendyn D., Restall, Brendon S., McAlister, Ewan A., Mittal, Rohan, Adam, Benjamin A., Giannakopoulos, Nadia, Peiris, Lashan, Silverman, Sveta, Deschenes, Jean, Li, Xingyu, Zemp, Roger J.
Format: Article
Language:English
Subjects:
631/1647/245/2160
631/1647/245/2226
631/67/1347
631/67/2321
692/4028/546
Breast
Deep learning
Generative adversarial networks
Histology
Histopathology
Humanities and Social Sciences
Labels
Medical imaging
multidisciplinary
Prostate
Remote sensing
Scattering
Science
Science (multidisciplinary)
Tumors
Ultraviolet reflection
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The goal of oncologic surgeries is complete tumor resection, yet positive margins are frequently found postoperatively using gold standard H&E-stained histology methods. Frozen section analysis is sometimes performed for rapid intraoperative margin evaluation, albeit with known inaccuracies. Here, we introduce a label-free histological imaging method based on an ultraviolet photoacoustic remote sensing and scattering microscope, combined with unsupervised deep learning using a cycle-consistent generative adversarial network for realistic virtual staining. Unstained tissues are scanned at rates of up to 7 mins/cm 2 , at resolution equivalent to 400x digital histopathology. Quantitative validation suggests strong concordance with conventional histology in benign and malignant prostate and breast tissues. In diagnostic utility studies we demonstrate a mean sensitivity and specificity of 0.96 and 0.91 in breast specimens, and respectively 0.87 and 0.94 in prostate specimens. We also find virtual stain quality is preferred ( P  = 0.03) compared to frozen section analysis in a blinded survey of pathologists. Oncologic tumour resection is not fully accurate. Here the authors report a label-free virtual histological imaging method based on a non-contact, reflection-mode ultraviolet photoacoustic remote sensing and scattering microscope, combined with unsupervised deep learning using a cycle-consistent GAN.
ISSN:2041-1723
2041-1723
DOI:10.1038/s41467-023-41574-2