At the Parco 99 Conference in Delft, Dr. Bart M. ter Haar Romeny gave an overview of the advanced applications in medicine which are actively being used at the Utrecht University Hospital. Dr. ter Haar Romeny received an M.S. in Applied Physics from Delft University of Technology in 1978, and a Ph.D. from Utrecht University in 1983. After being the principal physicist of the Utrecht University Hospital Radiology Department, he joined in 1989 the Image Sciences Institute of Utrecht University as an associate professor. Housed within the Radiology Department of Utrecht University Hospital, the Image Sciences Institute (ISI) is engaged in fundamental computer vision as well as in clinical research. An intense collaboration with the industry has amounted in several innovative imaging projects.
Established in 1989, ISI counts more than 50 researchers, half of which has studied physics whereas the other half has a computer science or mathematical background. The Institute has introduced High Performance Computing (HPC) for medical imaging at the 1000-bed University Hospital and is funded by the Utrecht University, the University Medical Centre, the Dutch National Science Foundation, the Ministry of Economic Affairs, and the European Commission. The advanced imaging equipment consists of 3 CT scanners with high resolution, 4 whole body MR scanners, 6 high field MR scanners, and a fully digital Nuclear Medicine Department. The PACS or Picture Archival Computer System is able to store about 10 Tbytes of patient images per year.
Priorities for ISI are to gain a fundamental insight in the image structure, to develop new workstation applications for companies, and to design tools for computer assisted radiology and surgery. The ISI researchers built a highly versatile 3D Volume Visualization Software to perform 3D matching of multi-modality data sets, fused image segmentation and quantitative analysis. The images are derived from different types of scanners to acquire an integrated view of the organ structures. This allows the ISI team to explore the complex relationship between computational vision and perception theory. A second hot topic is the image content-based database search. Clinical applications are focused on trabecular bone structure analysis for strength calculations in order to stimulate bone growth; the fast rubber-sheet matching of motion artefacts in digital subtraction angiography, applied in brain visualization; and local brain perfusion, imaged via SPECT and MRI.
Four or five companies work together with the Institute and pay $50.000 a year for software development by ISI. As a result, ISI assists Philips to build interventional MR equipment; Shell to perform seismic data analysis; Zeiss to design operating theatre microscopes; and Nucletron to develop X-thorax systems for digital chest scanning. Two ISI projects with the industry are of particular interest. The first one enables to visualize data directly next to the scanner for MR intravascular interventions. The second one allows to treat patients with stroke suffering from Abdominal Aorta Aneurysm (AAA). The visualization software is helping the surgeon to introduce a stent within the aorta. The advantages of MR guided vascular interventions over X-ray are that MR provides an excellent soft-tissue contrast, offers a fine selection of the scan planes, and safeguards the patient against ionizing radiation.
Together with Nucletron, which delivers digital X-ray chest scanners to third world countries, ISI is fighting the global tuberculosis epidemic, which kills 3 million people each year. The disease, affecting patients between 15 and 50, who form the economic work power of a nation, however can be cured. Since films are too expensive to use, the digital chest units will help to detect the contagious disease of which several multi-resistant forms are emerging. The Nucletron programme has shipped 35 mobile systems to Brazil. ISI has to do an image selection at the source with an automatic analysis software to screen the images which are sent by satellite at the reading station. In this way, the suspect images can be traced in the large collection of data that are just normal. The radiologists tell the ISI researchers what indications to look for in the thorax and vessel structures.
For the Abdominal Aorta Aneurysm (AAA) project, more and more images are needed to guide the minimally invasive surgery procedures. The treatment to repair AAA is called Trans-femoral Endovascular Aneurysm (TEAM). ISI has designed a fast automatic method to calculate slices, derived from a CT scan. The task is to let specialized companies produce by means of the aorta measurements data a highly individualized stent for each patient. ISI helps young radiologists to perform this complicated procedure with computer-simulated training. To this purpose, a phantom or closed box with a model of the aorta aneurysm is used for the testing and training, which acts in fact as a simulator. Even the deformations due to insertion of the stent can be calculated and rendered.
The University Hospital has equally embarked on a programme of computer-intensive visualization work in the operating theatre for the next five years to support image guided surgery. The interactive placement of electrodes in the skull of epileptic patients forms one of the applications. During the operation these electrodes need to be tracked. The advantage is that only a small hole has to be bored into the patient's skull. Neuro-navigation is a second useful application to register the real patient data to pre-scanned MR and CT data with an infrared localization and optical tracking system. The problem of the brain shift is solved with local warping. The ISI team has designed a special Java user interface for the surgeon to manipulate.
Surgical robotics constitutes one of the future ISI projects. The surgeon will no longer be standing next to the patient but is using a remote manipulator while a robot arm on the patient performs exactly the same finger and arm motions but 10 to 20 times smaller, to execute the highly accurate minimally invasive interventions. In conclusion, Dr. ter Haar Romeny stresses the vital importance of parallel computing and high computational power for image assisted surgery in the operation theatre of the future. Next to this, the MR physics and physiological MR research field is exploding with a potential of 26 different imaging modalities per patient, supported by the computational vision programme.
You can find more information on neuro-navigation in the VMW June 1999 article Neuro-navigation techniques visualize boundaries between sick and healthy tissue in the brain-pan. Please, also consult the Image Sciences Institute home page for more details.