Automatic segmentation of the aortic valve using deep learning

The content of the project is to provide an automatic valve segmentation with features for manual post-processing in order to optimally support the physician in planning and performing surgery. The automatic valve segmentation will be performed using current methods of “deep learning”. According to the current state of research, these methods provide excellent results in the field of image segmentation. Quantifications of the valve geometry can be generated on a patient-specific basis after completion of the project. This allows for a more accurate and comprehensive characterization of the presenting clinical picture.




Robert Kreher, Thomas Groscheck, Kristjan Qarri, Bernhard Preim, Alexander Schmeisser, Thomas Rauwolf, Rüdiger Christian Braun-Dullaeus, Sandy Engelhardt

A Novel Calibration Phantom for Combining Echocardiography with Electromagnetic Tracking Inproceedings

Proc. of CURAC, 2020.

BibTeX | Links: