Image Registration

Elastix Image Quantitative MRI CAD

In the Image Registration group we develop and evaluate novel image registration methods and use these in several applications. These applications are mainly focusing on quantitative MRI analysis and computer aided diagnosis. Our research can thus be roughly divided into three (interconnected) research lines:

  • Elastix: This research line covers all basic research on novel image registration methods. New methods are integrated in the open source elastix software. Recent developments have focused on groupwise registration methods, optimization techniques, multimodal registration, and semi-automated registration.

  • Quantitative MRI: This theme brings together research on MRI acquisition, reconstruction, and postprocessing. We have developed a comprehensive pipeline for motion compensation (based on Elastix) and model fitting, which is applicable to any quantitative MRI technique, such as T1&T2-relaxometry, diffusion tensor imaging, and dynamic contrast-enhanced MRI. We also have developed the “SPARCK” software for quantitative analysis of knee MRI, which is used extensively by the Musculoskeletal Imaging Research Group within Erasmus MC.

  • Compter Aided Diagnosis: Highlights in this theme include the organisation of a grand challenge on computer-aided diagnosis of dementia and the development of an interactive website showing a spatiotemporal model of the average aging brain. Advanced pattern recognition and machine learning methods play an important role in this research.

Some examples of previous projects where image registration played an important role are described below.

Neuro image analysis:

  • Automated segmentation of brain structures on MRI (applied to 4500 images from the Rotterdam Scan Study). More details…
  • Parameterization of fiber tracts in the cingulum. [More details…]
  • Early diagnosis of dementia by a novel pattern recognition method that uses nonrigid image registration to measure “anatomical distance” between subjects. More details…

Cardiovascular image analysis:

  • Automated segmentation of 7 heart segments on CTA (applied to 1400 images from the Department of Cardiology). Developed in the context of the Heart in 3D project.
  • Motion compensation in dynamic imaging data (e.g. cardiac time series). An integrated method for nonrigid registration of time series was developed by [Coert Metz]. Click here, here and [here] for example applications.
  • Shape models for cardiac motion estimation. Click [here], here and here for more details.
  • Alignment of 3-D CTA with 2-D X-ray data for improved guidance in coronary interventions. More details…
  • Registration of histology of human atherosclerotic carotid plaque to in-vivo imaging modalities. More details…
  • Registration of US, MR, and CT images of human atherosclerotic carotid plaque. More details…

In many projects we make use of the publicly available elastix software, and we extend the package with our newly developed methods.

Projects

Associated People

Chair

Group Leader

Postdocs

PhD Students

Technical Staffs

Trainees

  • Kees van Hespen
  • Quentin Couturier
  • Tarik Luisman
  • Jim Smit