Home | english  | Impressum | Datenschutz | Sitemap | KIT

Thesis or Internship: Multimodal computer aided breast cancer detection based on 3D-2D breast image registration

Thesis or Internship: Multimodal computer aided breast cancer detection based on 3D-2D breast image registration
Stellenausschreibung: Links:
Stellennummer:

IPE 03-17

Institut:

Institute for Data Processing and Electronics

Eintrittstermin:

On appointment

Bewerbungsfrist:

December 12, 2017

Kontaktperson:

Torsten Hopp

Multimodal computer aided breast cancer detection based on 3D-2D breast image registration

Foto 1 IPE 03-17
Foto 2 IPE 03-17
Background
In earlier work an image registration framework for 3D-2D registration of MRI and X-ray mammography was built up. This offers new possibilities in computer aided detection and diagnosis as by the known spatial correspondence two complementary modalities can be combined. In a first prototype implementation we demonstrated the feasibility of automated multimodal computer aided detection of breast cancer (see exemplary automated detection in MRI and X-ray mammogram on the right). However the number of features, the classification algorithms as well as the clinical parameters and images integrated into this prototype did not yet use available modalities to the full extent.

Aim
The aim of this work would be to restructure and extend the computer aided detection framework allowing for the extraction of features in 3D MRI volumes and at corresponding positions in multi view 2D X-ray mammograms.

Task description (topics will be adapted to available time frame)
 • Extension of the framework for the extraction of features, e.g. including mammograms from different viewing angles,
   adding more features, speeding up the mapping computation, extension to a sliding window approach.
 • Implementation of the methods
 • Development of methods for evaluation, in-depth evaluation with clinical datasets Qualifications
 • Preferably basic knowledge about medical imaging and medical image processing, biomechanics, pattern recognition,
    machine learning
 • Programming knowledge required, predominantly in MATLAB, partly C.
 • basic mathematics, statistics

Contact
Torsten Hopp, torsten hoppJqt4∂kit edu