Medical image analysis
- Project
- 17021 IMPACT
- Type
- New standard
- Description
- Fast and automatic segmentation of brain tumors.
- Generation of synthetic images from noise (progressive GAN 3D) or from other images (3D CycleGAN).
- Analysis of brain activity in white matter by combining functional MRI and diffusion MRI.
- Contact
- Anders Eklund
- Anders.eklund@liu.se
- Technical features
Input(s):
- MR images.
- fMRI images.
- dMRI images.
Main feature(s):
- Segmentation of brain tumors.
- Generation of synthetic images.
- Analysis of brain activity in white matter.
Output(s):
- Segmentations.
- Synthetic images.
- Brain activity maps.
- Integration constraints
Needs deep learning packages such as Tensorflow and/or Keras installed, see each repository.
- Targeted customer(s)
Medical imaging researchers.
- Conditions for reuse
Code is open source and available at different github repositories
- Confidentiality
- Public
- Publication date
- 12-09-2021
- Involved partners
- Linköping University (SWE)
Links
- https://github.com/IulianEmilTampu/bts_anatomical_context_info
- https://github.com/mdciri/3D-augmentation-techniques
- https://github.com/mdciri/Vox2Vox
- https://github.com/DavidAbramian/DSS
- https://github.com/DavidAbramian/CycleGAN
- https://github.com/wanderine/ProgressiveGAN3D
- https://github.com/wanderine/BrainTumourSegmentationqMRI