Master Thesis or Internship Development and evaluation of automated segmentation methods in multi-modal 3D breast image registration

  • job offer:

Development and evaluation of automated segmentation methods in multi-modal 3D breast image registration

Breast cancer is the most common cancer in women worldwide: more than 1.6 Mio. cancers are diagnosed per year according to WHO cancer statistics. At Karlsruhe Institute of Technology an automated method for multi-modal image registration is developed. The registration allows localizing suspicious tissue structures in all mo-dalities at a glance. Using this approach the advantages of modalities may be combined. Our method therefore contributes to multimodal diagnosis of breast cancer. The challenges for the image registration are the severe differences in image acquisition of different modalities such as patient positioning and compression state of the breast during examination. In order to overcome these challeng-es we apply sophisticated patient-specific biomechanical models of the breast to simulate the tissue deformation when subjected to compres-sion. Digital breast tomosynthesis (DBT) is an emerging technology which provides 3D information of the compressed breast anatomy. In a joint HEiKA research project ( in collabora-tion with the Medical Faculty Mannheim of the University Heidelberg, we apply this new modality to gain a deeper knowledge of the complex breast deformation and increase the accuracy of image registration methods to a new clinically applicable level. The project develops a fundamental method for the registration of MRI and DBT in order to identify more precise deformation models in general and provide a tool for easier, faster and more intuitive multimodal diagnosis of MRI and Tomosynthesis. The aim of this work is to contribute to the development of an automated workflow for the proposed patient-specific image registration method by means of developing fundamental algorithms for image processing and segmentation of the breast. The current methods (e.g. image segmentation) should be analyzed in terms of robustness and extended if necessary. Furthermore, the influence of material parameters and different model complexities should be evalu-ated by selected clinical datasets.

Task description (adaptable to desired time frame)
  •  Starting point of the work will be the analysis of the current state of the registration method and development of a work-plan.
  •  Implementation of image segmentation methods for breast MRI into the existing image registration workflow and their evaluation by selected clinical datasets.
  •  Extension of biomechanical model in terms of model complexities and evaluation of most influencing parameters for selected clinical datasets in terms of registration accuracy.

  •  Programming skills in MATLAB required.
  •  Interest in medical imaging and medical image processing, in particular image segmentation and image registration.
  •  Basic knowledge in (bio-)mechanical simulations / Finite Element simulations beneficial.  Basic mathematics, statistics.


Torsten Hopp,  torsten.hopp@kit.e
Patricia Cotič Smole