Enhancement of Fetal Face Visualization
Praktikum/BA/DA
Sample image of fetal face visualization, obtained from ultrasound data. Initially the face is occluded by other tissues (leftmost image). With automatic method only partial occlusion stays (center image). Face matching or smart sketch-based user interface may reveal the face completely (rightmost image).
Description
In medicine wide variety of scanning procedures is based on ultrasound acquisition technique.
Main reason for that is non-invasive nature of ultrasound (US), contrasted to Computed Tomography or Magnetic Resonance Imaging.
As for prenatal medicine non-invasivity of US allows to use it for fetal 3D volume acquisition.
Fetus is unborn human child.
General concern here is that scanned fetus is located completely inside of mother body.
It is occluded by different kinds of tissues, those should be detected and removed from the scanned volume in order to produce high-quality images of fetus.
Thus several methods were developed to deal with occluding tissues.
One of such a technique for the case of fetus face is developed at Vis-Group of Institute of Computer Graphics.
Although this technique operates reasonably good, we would like to announce topics to enhance the technique with either automatic face matching, or smart user interface.
Open topics
- Topic 1) Face matching on 2D images in case of missing features: design and implementation of algorithms for automatic face matching, matching of its main features and determination of missing features. This also includes localization of regions, where missing features should appear.
- Topic 2) Smart user interface to support interaction during the occluder removal procedure: sketch-based and image-content-driven interface tracks user interaction over 2D image of fetus face to determine the regions, where user wants to cut more of occluder, or leave more face tissue. Main requirement is high efficiency of the interface, as it is intended for routine clinical practice.
What we offer...
- Cooperation with GE Healthcare company
- Supervision by the CG Institute
What we expect...
- Good skills in C++ and software engineering
- Experience in image processing methods and pattern recognition algorithms
- Experience with touch screens and gesture recognition
Further information
If you are interested, just contact us:
Alexey Karimov
Vienna University of Technology - Insitute of Computer Graphics
Tel : +43 1 58801-18645
