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, degenerative spine diseases and inflammatory disorders like multiple sclerosis.
Anduin is a freely available web-based application able to fully automatically segment the full spine in CT scans detect vertebral fractures and measure bone mineral density. Try it out!
Driving forward an interest for fully automated segmentation of the spine in the research community, we publicly released voxel-level-annotated CT data of more than 300 patients and organised the VerSe (Large Scale Vertebrae Segmentation) challenge series MICCAI 2019 and 2020.
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 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.
We investigate different techniques to automatically assess spinal lesion load and its clinical relevance for the individual patient.
Technische Universität München, Ismaninger Str. 22, 81675 München
Technische Universität München, Einsteinstraße 25, 81675 München
Key highlights
Key highlights