From acquired MRI images, we calculate and analyze quantitative MRI maps, which aim to provide exact measurements of biophysical quantities. For this task, we employ advanced image processing algorithms to account for biases, enhance spatial alignment, and combine different scans into so-called multi-parametric maps. Advanced machine learning methods help us to develop new ways to correct for artifacts that are hard to tackle with conventional image processing methods.
DeepcomplexMRI deep learnig reconstruction has been modified to process multi-echo MRI images. First results for different undersampling strategies suggest that performance is comparable to modern iterative algorithms like ENLIVE while taking only about 5 minutes to reconstruct a full 3D 1mm³/voxel resolved head image stack.
Transverse relaxation parameters are quantified in vivo for different cortical structures of the human brain at ultra-high field strength.
Embedded in the clinical trial NISCI (Nogo inhibition in spinal cord injury: www.nisci-2020.eu), we employ whole brain quantitative imaging at 3 Tesla as a new biomarker for de- and regeneration.
We work on improving reliability of quantitative parametric maps by correcting for rigid head motion and B0-fluctuations measured during acquisition at 7T as well as by employing general function approximators to correct for artifacts of unknown origin at 3T.