On the usability of synthetic data for improving the robustness of deep learning-based segmentation of cardiac magnetic resonance images.
Yasmina Al KhalilSina AmirrajabCristian LorenzJürgen WeeseJosien P. W. PluimMarcel BreeuwerPublished in: Medical Image Anal. (2023)
Keyphrases
- synthetic data
- magnetic resonance images
- deep learning
- brain mri
- cardiac mri
- medical images
- intra subject
- partial volume
- brain scans
- real image data
- mr imaging
- mri data
- brain tumors
- tissue segmentation
- medical imaging
- mri images
- mr images
- brain structures
- myocardial infarction
- myocardial motion
- bias field
- magnetic resonance imaging
- unsupervised learning
- brain tissue
- white matter
- real world
- left ventricular
- data sets
- diffusion tensor
- manual segmentation
- weakly supervised
- high resolution
- image segmentation
- brain tumor segmentation
- accurate segmentation
- machine learning
- magnetic resonance
- mental models
- image analysis
- feature selection
- diffusion weighted
- segmentation method
- pattern recognition
- deformable models
- image data
- segmentation algorithm
- multiscale
- left ventricle
- feature space
- model based segmentation