Help Privacy Policy Disclaimer
  Advanced SearchBrowse


  Generating 3D TOF-MRA volumes and segmentation labels using generative adversarial networks

Subramaniam, P., Kossen, T., Ritter, K., Hennemuth, A., Hildebrand, K., Hilbert, A., et al. (2022). Generating 3D TOF-MRA volumes and segmentation labels using generative adversarial networks. Medical Image Analysis, 78: 102396. doi:10.1016/j.media.2022.102396.

Item is


show Files
hide Files
Subramaniam_2022.pdf (Publisher version), 4MB
MIME-Type / Checksum:
application/pdf / [MD5]
Technical Metadata:
Copyright Date:
Copyright Info:




Subramaniam, Pooja1, Author
Kossen, Tabea1, 2, Author
Ritter, Kerstin3, 4, Author
Hennemuth, Anja2, 5, 6, Author
Hildebrand, Kristian7, Author
Hilbert, Adam1, Author
Sobesky, Jan8, 9, Author
Livne, Michelle1, Author
Galinovic, Ivana9, Author
Khalil, Ahmed9, 10, 11, 12, Author           
Fiebach, Jochen B.10, Author
Frey, Dietmar1, Author
Madai, Vince I.1, 13, 14, Author
1Charité Lab for Artificial Intelligence in Medicine (CLAIM), Charité University Medicine Berlin, Germany, ou_persistent22              
2Department of Computer Engineering and Microelectronics, TU Berlin, Germany, ou_persistent22              
3Department of Psychiatry and Psychotherapy, Charité University Medicine Berlin, Germany, ou_persistent22              
4Bernstein Center for Computational Neuroscience, Germany, ou_persistent22              
5Institute for Imaging Science and Computational Modelling in Cardiovascular Medicine, Charité University Medicine Berlin, Germany, ou_persistent22              
6Fraunhofer MEVIS, Bremen, Germany, ou_persistent22              
7Department VI Computer Science and Media, Beuth University of Applied Sciences, Berlin, Germany, ou_persistent22              
8Johanna-Etienne-Hospital, Neuss, Germany, ou_persistent22              
9Center for Stroke Research, Charité University Medicine Berlin, Germany, ou_persistent22              
10Department Neurology, MPI for Human Cognitive and Brain Sciences, Max Planck Society, ou_634549              
11MindBrainBody Institute, Berlin School of Mind and Brain, Humboldt University Berlin, Germany, ou_persistent22              
12Berlin Institute of Health (BIH), Germany, ou_persistent22              
13Faculty of Computing, Engineering and the Built Environment, University of Birmingham, United Kingdom, ou_persistent22              
14QUEST-Center for Transforming Biomedical Research, Berlin Institute of Health, Germany, ou_persistent22              


Free keywords: 3D Medical imaging; Anonymization; Brain vessel segmentation; Generative adversarial networks; Mixed precision
 Abstract: Deep learning requires large labeled datasets that are difficult to gather in medical imaging due to data privacy issues and time-consuming manual labeling. Generative Adversarial Networks (GANs) can alleviate these challenges enabling synthesis of shareable data. While 2D GANs have been used to generate 2D images with their corresponding labels, they cannot capture the volumetric information of 3D medical imaging. 3D GANs are more suitable for this and have been used to generate 3D volumes but not their corresponding labels. One reason might be that synthesizing 3D volumes is challenging owing to computational limitations. In this work, we present 3D GANs for the generation of 3D medical image volumes with corresponding labels applying mixed precision to alleviate computational constraints.

We generated 3D Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) patches with their corresponding brain blood vessel segmentation labels. We used four variants of 3D Wasserstein GAN (WGAN) with: 1) gradient penalty (GP), 2) GP with spectral normalization (SN), 3) SN with mixed precision (SN-MP), and 4) SN-MP with double filters per layer (c-SN-MP). The generated patches were quantitatively evaluated using the Fréchet Inception Distance (FID) and Precision and Recall of Distributions (PRD). Further, 3D U-Nets were trained with patch-label pairs from different WGAN models and their performance was compared to the performance of a benchmark U-Net trained on real data. The segmentation performance of all U-Net models was assessed using Dice Similarity Coefficient (DSC) and balanced Average Hausdorff Distance (bAVD) for a) all vessels, and b) intracranial vessels only.

Our results show that patches generated with WGAN models using mixed precision (SN-MP and c-SN-MP) yielded the lowest FID scores and the best PRD curves. Among the 3D U-Nets trained with synthetic patch-label pairs, c-SN-MP pairs achieved the highest DSC (0.841) and lowest bAVD (0.508) compared to the benchmark U-Net trained on real data (DSC 0.901; bAVD 0.294) for intracranial vessels.

In conclusion, our solution generates realistic 3D TOF-MRA patches and labels for brain vessel segmentation. We demonstrate the benefit of using mixed precision for computational efficiency resulting in the best-performing GAN-architecture. Our work paves the way towards sharing of labeled 3D medical data which would increase generalizability of deep learning models for clinical use.


Language(s): eng - English
 Dates: 2022-01-282021-07-132022-02-172022-02-242022-05
 Publication Status: Issued
 Pages: -
 Publishing info: -
 Table of Contents: -
 Rev. Type: -
 Identifiers: DOI: 10.1016/j.media.2022.102396
Other: epub 2022
PMID: 35231850
 Degree: -



Legal Case


Project information

show hide
Project name : -
Grant ID : 031B0154
Funding program : -
Funding organization : Go-Bio

Source 1

Title: Medical Image Analysis
  Other : Med. Image Anal.
Source Genre: Journal
Publ. Info: London : Elsevier
Pages: - Volume / Issue: 78 Sequence Number: 102396 Start / End Page: - Identifier: ISSN: 1361-8415
CoNE: https://pure.mpg.de/cone/journals/resource/954927741859