Respecting Anatomy

Right: CT-scan with the deformation grid showing both local rigid (green) and non-linear (orange) structures. Left: Intra-session cone-beam CT with transformed target structures.

Sophisticated Image Registration Respecting Anatomy

Clinical Challenge

 

In today's deformable image registration solutions, all available information must be incorporated to achieve comprehensive results and enable smart choices in clinical practice. In addition to elastic, nonlinear deformations, the sophisticated Fraunhofer MEVIS registration framework enables easy and flexible integration of additional, anatomy based information. This information can be obtained directly from available image data or added from other sources. By this, physically plausible deformation behavior can be guaranteed, that enables a new level of individualized diagnosis.

 

Solutions and Features

 

We provide solutions incorporating various constaints in the registration model, e.g.:

  • Sliding motion
  • Local rigidity
  • Locally bounded volume change
  • Volume preservation
  • Mass preservation
  • Landmark matching, either exact orwithin defined tolerances
  • High-dimensional feature matching

Highlights

 

  • Simple: Few plausible parameters
  • Reliable: Constraints can be guaranteed
  • Versatile: Broad range of different constraints possible
  • Robust: Sound mathematical modeling combined with efficient numerics [1,2]

[1] Haber, E., Heldmann, S., & Modersitzki, J. (2009). A computational framework for image-based constrained registration. Linear Algebra Appl., 431(3), 459-470.

[2] Modersitzki, J. (2009). FAIR: flexible algorithms for image registration (Vol. 6). SIAM