Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Jan 6:13:7.
doi: 10.4103/jpi.jpi_59_20. eCollection 2022.

Histo-Fetch - On-the-Fly Processing of Gigapixel Whole Slide Images Simplifies and Speeds Neural Network Training

Brendon Lutnick  1 , Leema Krishna Murali  2 , Brandon Ginley  1 , Avi Z Rosenberg  3 , Pinaki Sarder  1   2
Affiliations

Histo-Fetch - On-the-Fly Processing of Gigapixel Whole Slide Images Simplifies and Speeds Neural Network Training

Brendon Lutnick et al. J Pathol Inform. .

Abstract

Background: Training convolutional neural networks using pathology whole slide images (WSIs) is traditionally prefaced by the extraction of a training dataset of image patches. While effective, for large datasets of WSIs, this dataset preparation is inefficient.

Methods: We created a custom pipeline (histo-fetch) to efficiently extract random patches and labels from pathology WSIs for input to a neural network on-the-fly. We prefetch these patches as needed during network training, avoiding the need for WSI preparation such as chopping/tiling.

Results & conclusions: We demonstrate the utility of this pipeline to perform artificial stain transfer and image generation using the popular networks CycleGAN and ProGAN, respectively. For a large WSI dataset, histo-fetch is 98.6% faster to start training and used 7535x less disk space.

Keywords: Convolutional neural network; generative adversarial network; tensorflow; whole slide images.

PubMed Disclaimer

Conflict of interest statement

There are no conflicts of interest.

Figures

Fig. 1
Fig. 1
(a) The traditional method uses the CPU to chop whole slide images into patches which are saved to disk before convolutional neural network training. These patches are read and fed to the graphics processing unit for training. (b) Histo-fetch randomly selects indices containing tissue on the fly. These are processed on the CPU and supplied to the graphics processing unit. (c) Efficiency comparison of the two approaches using ProGAN, highlighting preprocessing time and additional disk space required using a dataset of 151 human biopsy whole slide images. The average training step time does not significantly change.
Fig. 2
Fig. 2
(a) Shows results from two CycleGAN networks, which take hematoxylin and eosin or silver stained input patches and transform them to in silico periodic acid-schiff stains. (b) Shows synthetic tissue patches generated using ProGAN trained on 1331 human biopsy (765 GB) whole slide images with various histological stains.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Al-Janabi S., Huisman A., Van Diest P.J. Digital pathology: current status and future perspectives. Histopathology. 2012;61:1–9. - PubMed
    1. Goode A., Gilbert B., Harkes J., Jukic D., Satyanarayanan M. OpenSlide: A vendor-neutral software foundation for digital pathology. J Pathol Inform. 2013;4:27. - PMC - PubMed
    1. Cruz-Roa A., Gilmore H., Basavanhally A., et al. High-throughput adaptive sampling for whole-slide histopathology image analysis (HASHI) via convolutional neural networks: Application to invasive breast cancer detection. PLoS One. 2018;13 - PMC - PubMed
    1. Lutnick B., Ginley B., Govind D., et al. An integrated iterative annotation technique for easing neural network training in medical image analysis. Nat Mach Intell. 2019;1:112–119. - PMC - PubMed
    1. Cruz-Roa A., Basavanhally A., González F., et al. Medical Imaging 2014: Digital Pathology. International Society for Optics and Photonics; 2014. Automatic detection of invasive ductal carcinoma in whole slide images with convolutional neural networks.

LinkOut - more resources