Multi-scale GANs for Memory-efficient Generation of High Resolution Medical Images

Hristina Uzunova*, Jan Ehrhardt, Fabian Jacob, Alex Frydrychowicz, Heinz Handels

*Korrespondierende/r Autor/-in für diese Arbeit

Abstract

Currently generative adversarial networks (GANs) are rarely applied to medical images of large sizes, especially 3D volumes, due to their large computational demand. We propose a novel multi-scale patch-based GAN approach to generate large high resolution 2D and 3D images. Our key idea is to first learn a low-resolution version of the image and then generate patches of successively growing resolutions conditioned on previous scales. In a domain translation use-case scenario, 3D thorax CTs of size 512 3 and thorax X-rays of size 2048 2 are generated and we show that, due to the constant GPU memory demand of our method, arbitrarily large images of high resolution can be generated. Moreover, compared to common patch-based approaches, our multi-resolution scheme enables better image quality and prevents patch artifacts.

OriginalspracheEnglisch
TitelMICCAI 2019: Medical Image Computing and Computer Assisted Intervention – MICCAI 2019
Redakteure/-innenDinggang Shen, Tianming Liu, Terry M. Peters, Lawrence H. Staib, Caroline Essert, Sean Zhou, Pew-Thian Yap, Ali Khan
Seitenumfang9
Band11769
ErscheinungsortCham
Herausgeber (Verlag)Springer Verlag
Erscheinungsdatum10.10.2019
Seiten112-120
ISBN (Print)978-3-030-32225-0
ISBN (elektronisch)978-3-030-32226-7
DOIs
PublikationsstatusVeröffentlicht - 10.10.2019
Veranstaltung22nd International Conference on Medical Image Computing and Computer-Assisted Intervention - Shenzhen, China
Dauer: 13.10.201917.10.2019
Konferenznummer: 232939

Strategische Forschungsbereiche und Zentren

  • Forschungsschwerpunkt: Biomedizintechnik

Fingerprint

Untersuchen Sie die Forschungsthemen von „Multi-scale GANs for Memory-efficient Generation of High Resolution Medical Images“. Zusammen bilden sie einen einzigartigen Fingerprint.

Zitieren