High-resolution Controllable Prostatic Histology Synthesis using StyleGAN

Gagandeep B. Daroach, Josiah A. Yoder, Kenneth A. Iczkowski, Peter S. LaViolette

2021

Abstract

For use of deep learning algorithms in clinical practice, detailed justification for diagnosis is necessary. Convolutional Neural Networks (CNNs) have been demonstrated to classify prostatic histology using the same diagnostic signals as pathologists. Using the StyleGAN series of networks, we demonstrate that recent advances in high-resolution image synthesis with Generative Adversarial Networks (GANs) can be applied to prostatic histology. The trained network can produce novel histology samples indistinguishable from real histology at 1024x1024 resolution and can learn disentangled representations of histologic semantics that separates at a variety of scales. Through blending of the latent representations, users have the ability to control the projection of histologic semantics onto a reconstructed image. When applied to the medical domain without modification, StyleGAN2 is able to achieve a Fréchet Inception Distance (FID) of 3.69 and perceptual path length (PPL) of 33.25.

Paper Citation

in Harvard Style

Daroach G., Yoder J., Iczkowski K. and LaViolette P. (2021). High-resolution Controllable Prostatic Histology Synthesis using StyleGAN. In Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 2: BIOIMAGING; ISBN 978-989-758-490-9, SciTePress, pages 103-112. DOI: 10.5220/0010393900002865

in Bibtex Style

@conference{bioimaging21,
author={Gagandeep B. Daroach and Josiah A. Yoder and Kenneth A. Iczkowski and Peter S. LaViolette},
title={High-resolution Controllable Prostatic Histology Synthesis using StyleGAN},
booktitle={Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 2: BIOIMAGING},
year={2021},
pages={103-112},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010393900002865},
isbn={978-989-758-490-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 2: BIOIMAGING
TI - High-resolution Controllable Prostatic Histology Synthesis using StyleGAN
SN - 978-989-758-490-9
AU - Daroach G.
AU - Yoder J.
AU - Iczkowski K.
AU - LaViolette P.
PY - 2021
SP - 103
EP - 112
DO - 10.5220/0010393900002865
PB - SciTePress