We study new imaging techniques in CT and MRI for quantitative imaging of the spine. We develop automated evaluation algorithms including artificial intelligence, deep learning and biomechanic modeling to predict disease progression and get new pathophysiologic insights. Currently we focus on osteoporosis, back pain and degenerative spine diseases.
We optimize both CT and MRI protocols to derive quantitative tissue parameters of the spine. Recently we investigated novel iterative CT reconstruction techniques and sparse sampling acquisition to lower radiation dose in bone mineral density measurements. In MRI we demonstrated that diffusion tensor imaging (DTI) correlates well with muscle strength.
We develop vision- and learning-based algorithms capable of locating, identifying, and segmenting vertebrae in CT and MR scans as well as in radiographs. We intend to open-source our processing algorithms to the computer science and medical communities.
We develop automated techniques for opportunistic BMD measurements in any CT dataset and patient specific evaluation of local bone loss using voxel based morphometry (VBM). We also implemented automatic fracture detection algorithms using artificial intelligence.
To better understand back pain and surgical outcome, we combine quantitative parameters from different imaging modalities (X-rays, MRI, CT) in one biomechanical model, based on multibody simulations (MBS) on a large scale and finite element models (FEM/FCM) on a small scale.
Technische Universität München, Ismaninger Str. 22, 81675 München
Technische Universität München, Einsteinstraße 25, 81675 München
We are organising the Large Scale Vertebrae Segmentation Challenge (VerSe2019) in conjunction with MICCAI 2019, Shenzen, China.
Key highlights