gms | German Medical Science

The main goal of Medical Image Computing (MIC) is to extract clinically relevant information or knowledge from medical images. MIC focuses on the computational analysis of images using methods such as image segmentation (http://en.wikipedia.org/wiki/Medical_Image_Computing#Segmentation), image registration (http://en.wikipedia.org/wiki/Medical_Image_Computing#Registration), image-based physiological modeling (http://en.wikipedia.org/wiki/Medical_Image_Computing#Image-based_Physiological_Modelling), and visualization. The use of medical images in the context of therapeutic procedures is on the rise. Pre-procedural images are used both for planning and for intra-procedural guidance. Many image modalities are finding use today both in clinical routine and in a variety of research scenarios. Examples of such pre-procedure image data include plain X-rays, CT, a variety of MR techniques and Ultrasound. A variety of MIC techniques are available for preparation of the data, extraction of measures, creation of models and generation of visualizations. These techniques are used to integrate information and present it to clinicians in a way that is useful to them. Many of the fastest and most robust solutions have found their way into commercial medical devices and are approved and available for routine clinical use. Translation of technology in our field is fairly slow. Because of this, many more MIC technologies have yet to make the translation from the laboratory into clinical practice.

In addition to use of MIC technologies in pre-procedural imaging, we see an increasing use of intra-procedureal imaging, most commonly US, X-ray and endoscopy, but increasingly including CT and even MR. Use of MIC technologies on intra-procedural images raises many interesting boundary conditions: typically there are only limited computational resources accessible and only a few seconds or minutes available to perform the computing. On the other hand not everything needs to be automated and the demands for precision vary. In general, the farther away you are from the trajectory of the devices and from the target, the less need for millimeter level accuracy. The MIC community has spent relatively little effort in exploring this very promising constraint space. Accordingly there is a rich potential for new research topics. Another area with need for innovation is on the level of software systems for integration and workflow support. Full-fledged platforms are needed as a start point for this type of research. The traditional approach to research in MIC is that a graduate student is given images from one example case and creates a platform from scratch to enable their research on that data. This approach limits the capabilities and robustness of such research platforms. It is increasingly necessary to change the way problems are approached in order to make progress in the context of actual procedures. In summary, MIC technologies have already a strong presence in the procedure room but there are still significant unsolved problems and a trove of results of technological research that have not yet made it into clinical practice.