Open Ground Truth Training Network – Simulated MRI for medical image analysis research
News regarding openGTN
On 20 April 2023, Sina Amirrajab successfully defended his PhD thesis at the Eindhoven University of Technology (TU/e). Congratulations to Sina! He is currently employed as Postdoc at TU/e in the ITEA4 funded Spectralligence project.
On 20 April 2023 Sina Amirrajab will defend his PhD thesis entitled “Simulation and Synthesis for Cardiac Magnetic Resonance Image Analysis”.
Location: TU Eindhoven, Building Atlas, room 0.710 (map)
Time: 13:30 – 15:30
The defense is public, you are welcome to attend!
openGTN PhD researcher Sina Amirrajab has investigated physics-based simulation and deep-learning based synthesis of cardiac magnetic resonance image (MRI) data. Together with openGTN PhD researchers Aymen Ayaz and Yasmina Al Khalil he showed that deep-learning based image segmentation can be significantly improved by augmenting real human MRI data with the simulated and synthesized data. His research has resulted in ample publications. His PhD thesis can be downloaded here,
On 2 November 2022 the openGTN project held its Final Symposium. The meeting took place at partner University Medical Center Utrecht (UMCU) and was hosted by the Fall Meeting of the Dutch Society of Pattern Recognition and Image Processing (in Dutch: NVPHBV). A total of 100 persons registered. After a short introduction about openGTN by project coordinator prof. Marcel Breeuwer, keynote speaker prof. Sebastian Kozerke gave his view on the importance, status and potential of image simulation and synthesis. Thereafter, the 3 involved PhD researcher Sina Amirrajab, Aymen Ayaz and Yasmina Al Khalil presented their research achievements. All in all, a very successful meeting, with many interesting discussion during the breaks.
The full program of the openGTN final symposium (2 November 2022, hosted by the NVPHBV Fall Meeting) is now available here.
Registration is still open!
The 25th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) conference was held 18-22 September 2022 in Singapore. At the conference, the openGTN PhD researchers Yasmina Al Khalil and Sina Amirrajab received two awards from the MICCAI CMRxMotion Challenge. They participated in two tasks: “Robust CMR Segmentation” and “Image Quality Assessment”, ended up 3rd in both! Ample synthesized image data was used to train the deep-learning neural network solutions that they implemented.
at the University Medical Center Utrecht, The Netherlands
2 November 2022
The final public symposium of the Open Ground Truth Training Network (openGTN) project will take place on Wednesday 2 November, hosted by the Fall Meeting of the Dutch Society for Pattern Recognition and Image Processing (NVPHBV). The morning will be fully dedicated to openGTN: exciting new methods of MRI simulation and synthesis and their application to image segmentation will be presented by the involved PhD researchers. Keynote presentation will be given by prof. dr. Sebastian Kozerke from ETH Zurich. In the afternoon, more broader applications of simulation and synthesis will be presented, also outside the medical imaging domain.
You can register for the symposium via the website of the NVPHBV
After a secondment of more than 2 years at Philips Research Hamburg, Germany, PhD researcher Yasmina Al Khalil returned to the Eindhoven University of Technology (TU/e), The Netherlands, to finalize her PhD thesis. The above picture shows the 3 openGTN PhD’s united in their room at TU/e.
Finally, after 2 years of repeated COVID lock-downs and attending online-only conferences, the 3 openGTN PhD researchers Aymen Ayaz, Yasmina Al Khalil and Sina Amirrajab could all attend and present their achievements at the ISMRM 2022 conference. The left picture shows the team at the ISMRM exhibit booth, together with prof. Marcel Breeuwer.
6 abstracts have been accepted for presentation at the coming ISMRM 2022 conference:
S. Amirrajab, C. Lorenz, J. Weese, J. Pluim, M. Breeuwer. “Intra- and intersubject synthesis of cardiac MR images using a VAE and GAN”
S. Amirrajab, Y. Al Khalil, C. Lorenz, J. Weese, J. Pluim, M. Breeuwer. “sim2real: Cardiac MR image simulation-to-real translation via unsupervised GANs”
Y. Al Khalil, A. Ayaz, C. Lorenz, J. Weese, J. Pluim, M. Breeuwer. “Reducing the impact of texture on deep-learning brain tissue segmentation networks trained with simulated MR images”
Y. Al Khalil, S. Amirrajab, C. Lorenz, J. Weese, J. Pluim, M. Breeuwer, “Late feature fusion and GAN-based augmentation for generalizable cardiac MRI segmentation”
A. Ayaz, K. Lukassen, C. Lorenz, J. Weese, M. Breeuwer. “Brain MR image super resolution using simulated data to perform in real-world MRI”
A. Ayaz, R. Jong, S. Abbasi-Sureshjani, S. Amirrajab, C. Lorenz, J. Weese, M. Breeuwer. “3D brain MRI synthesis utilizing 2D SPADE-GAN and 3D CNN architecture”
See also our publication list
We have recently added simulated MRI with pathology to our openGTN public database. Please visit opengtn.eu/database for more information.
