Browsing by Author "Gietzen, Thomas"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    A Multilinear Model for Bidirectional Craniofacial Reconstruction
    (The Eurographics Association, 2018) Achenbach, Jascha; Brylka, Robert; Gietzen, Thomas; Hebel, Katja zum; Schömer, Elmar; Schulze, Ralf; Botsch, Mario; Schwanecke, Ulrich; Puig Puig, Anna and Schultz, Thomas and Vilanova, Anna and Hotz, Ingrid and Kozlikova, Barbora and Vázquez, Pere-Pau
    We present a bidirectional facial reconstruction method for estimating the skull given a scan of the skin surface and vice versa estimating the skin surface given the skull. Our approach is based on a multilinear model that describes the correlation between the skull and the facial soft tissue thickness (FSTT) on the one hand and the head/face surface geometry on the other hand. Training this model requires to densely sample the Cartesian product space of skull shape times FSTT variation, which cannot be obtained by measurements alone. We generate this data by enriching measured data-volumetric computed tomography scans and 3D surface scans of the head-by simulating statistically plausible FSTT variations. We demonstrate the versatility of our novel multilinear model by estimating faces from given skulls as well as skulls from given faces within just a couple of seconds. To foster further research in this direction, we will make our multilinear model publicly available.
  • No Thumbnail Available
    Item
    Segmenting Teeth from Volumetric CT Data with a Hierarchical CNN-based Approach
    (The Eurographics Association, 2018) Macho, Philipp Marten; Kurz, Nadja; Ulges, Adrian; Brylka, Robert; Gietzen, Thomas; Schwanecke, Ulrich; {Tam, Gary K. L. and Vidal, Franck
    This paper addresses the automatic segmentation of teeth in volumetric Computed Tomography (CT) scans of the human skull. Our approach is based on a convolutional neural network employing 3D volumetric convolutions. To tackle data scale issues, we apply a hierarchical coarse-to fine approach combining two CNNs, one for low-resolution detection and one for highresolution refinement. In quantitative experiments on 40 CT scans with manually acquired ground truth, we demonstrate that our approach displays remarkable robustness across different patients and device vendors. Furthermore, our hierarchical extension outperforms a single-scale segmentation, and network size can be reduced compared to previous architectures without loss of accuracy.