Home | english  | Impressum | Datenschutz | Sitemap | Kontakt | KIT | UNI HD

HEiKA Geschäftsstelle

Dr. Regine Kleber and

María García

infoUzz2∂heika-research de

Coregistration of breast MRI and tomosynthesis: can biomechanical modeling of the breast be improved by factor two?

Coregistration of breast MRI and tomosynthesis: can biomechanical modeling of the breast be improved by factor two?

Dr. Clemens Kaiser, Medical Faculty Mannheim, University of Heidelberg.

Dr., Torsten Hopp, Institute for Data Processing and Electronics, KIT.


Medical Technology for Health





Multimodal imaging is essential for early breast cancer detection as imaging techniques vary in their diagnostic features. The challenge in combining modalities is the different patient position and especially deformation of the breast. To overcome this challenge, image registration techniques automatically estimate the spatial correspondence. Especially in dense breasts, combining the information of mammography and MRI is substantial for a more conclusive diagnosis. However, the registration of MRI and mammography faces the problem of complex nonlinear breast deformation in 3D, which is only depicted in a 2D projection image.
While researchers have developed registration methods with increasingly complex deformation models, their accuracy is still limited to more than 10 mm, which dramatically limits their clinical applicability. Digital breast tomosynthesis (DBT) is an emerging technology which provides 3D information of the compressed breast anatomy.
We propose to apply this new modality to gain a deeper knowledge of the complex breast deformation and raise the accuracy of image registration methods to a new clinically applicable level. The proposed HEiKA project aims at developing and evaluating the fundamental method for registration of MRI and DBT in order to 1) identify more precise deformation models, 2) enhance the registration accuracy by approx. factor two and 3) provide a striking tool for easier, faster and more intuitive multimodal diagnosis. The method will build on our earlier work using a patientspecific biomechanical model to simulate the compression of the breast during mammography acquisition.
The HEiKA project will lay the foundations for a subsequent joint DFG grant. In a larger context we would like to initiate research on establishing multimodal image registration to integrate a variety of imaging modalities based on a generic transformation model and translate the methods to final application in clinical routine.