bart ter haar romeny


Part II: 1979-1992

A visual story.

This period covers:
- PhD at Utrecht University
- Clinical Physicist at Radiology
- Picture Archiving & Communication System
- 3D visualization
- Medical Image Analysis

PhD at Utrecht University

In June 1979 I started a PhD project in neurophysiology at Utrecht University, guided by prof. Jan Denier van der Gon. My thesis described and was titled the 'Organization and Control of Motor Units in Human Upper Arm Muscles' [Haar1983]. My project was funded by an NWO grant.

Invasive single fibre EMG recording experiments on myself.

As EMG was far too noisy, I decided to work with 25 μm bipolar wire electrodes in my own muscles (4 wires per needle insertion, in 2-3 muscles). It was fascinating to study the task-driven non-linear recruitment policies [Haar1982], and listen to the muscle fibre spikes that I myself generated. Compared to today it was primitive, but fascinating electrophysiology with myself (and my students) as subjects.

Anatomy, electrophysiology and electronics

I needed to know much more of the anatomy of human upper arm muscles. I decided to ask for help at the Anatomy Department of Utrecht University in my first months as a PhD student. It was agreed, and I remember the first day that I arrived for my lessons that I had to wait. The assistant went to the basement, and I heard the sound of a circular saw. Moments later he came back with a detached arm in a stainless steel container. It gave me 2 fascinating weeks dissecting it to the very nitty-gritty. How incredibly well-organized nature can be!

A wall of recording and display equipment

As experiments lasted long (typically a whole day + evening of experiments on a single electrode bundle injection) I made a switchable wiring patch panel, so experiments could be quickly interchanged. I programmed 12 coupled microprocessor systems (in assembly language) for the realtime signal- and data analysis. 

Dr . Stan Gielen (later professor in Biophysics in Nijmegen, and president of the national Dutch Science Foundation) built me a real-time spike discriminator, based on principal component analysis (PCA) of the cacophony of the many motor units recorded simultaneously.

I graduated on 3 July 1983.
The ceremony was in the beautiful Academy Room of Utrecht University.

Prof.dr.dr.Jan Koenderink

The Department Medical and Physiological Physics, section 'Physics of Man', in the Buys Ballot building at Utrecht University campus where I worked was headed by prof. Jan J. Koenderink. I was captivated by his keen sense of physics to discover the secrets of visual perception, and much enjoyed his lectures and discussions.
He has par excellence pioneered the introduction of (differential geometry) in modern vision research. Jan Koenderink was awarded an honorary doctorate by the Catholic University of Leuven in 1987, and the Azriel Rosenfeld Lifetime Achievement Award at ICCV 2013 in Sydney, Australia (acceptance speech).

The Koenderink Prize recognises fundamental contributions in computer vision. It is awarded each year at the European Conference on Computer Vision (ECCV, one of the most prestigious conferences in the field) for a paper published ten years ago at that conference which has withstood the test of time.

During my PhD period I was a member of the Dutch Society for Biophysics and Biomedical Engineering (BIOPM). Later I became its President (1998-2002).

On 28-10-1988 I married with Hetty van Eldijk, my incredible and energizing life partner, in Utrecht.

I travelled the world with her, an amazing series of adventurous journeys: see

She owns a wonderful shop in specialized necklaces and beads, 'KettingKunst' in the historic center of Utrecht.

I never learned so much about imaging equipment as those early years, where we opened every scanner and X-ray system and developed image quality control phantoms. I was teaching courses on Radiation Protection, and Physics of Medical Imaging.

Another major activity was the interaction with Philips Healthcare, our primaty equipment provider, through Walter Blount, the sympathetic and highly professional Philips account manager. I was involved in the purchase, acceptance testing, quality control and maintenance of new equipment. A grand acquisition was the 1.5 T MRI scanner, one of the first 4 MRI scanners in the Netherlands.

In 1994 I co-organized a nice symposium 'Physics in Medicine' at TU/e, showcasing the role of the emerging field of clinical physics (PDF).

The Utrecht University Hospital was the major test site for Philips' CT scanners. Such as the Philips Tomoscan CT500, a first-of-a-kind 4th generation CT helical scanner with moving focal spot X-ray tube on a slip ring.

Before such complex systems went to the market, they were first fully clinically tested for one year by the radiologists in our department. This involved many different local research projects. 
This was great. It meant that we often had the newest equipment, and a very special relation with this big industrial partner in Eindhoven (Philips is in the world's top 3 of medical imaging equipment manufacturers, with GE and Siemens).

Philips Tomoscan 500

I came back from the US with the strong advice to set up digital PACS in our hospital. It would be the first in the Netherlands, one of the first in Europe. At the same time, BAZIS, the leading Hospital Information System (HIS) developer in the Netherlands (led by prof. Ab Bakker in Leiden) and Philips Medical Systems were starting work to build a PACS (the 'Dutch PACS Project'), and we joined forces, funded by the Dutch Ministry of Health. I became the project leader of the radiological PACS experiment at Utrecht University (with a team of 7 people), while my respected colleague Jan Peter de Valk was HIS-PACS project leader in Leiden.

We managed to build the first digital reading room in Europe, and fully digitized one clinic of the Internal Medicine Department. It was a nice pioneering time, with a high quality PACS team, carrying out a careful clinical design, a clinical evaluation and a thorough efficiency study. We paid  frequent visits to the SPIE Medical Imaging conferences in San Diego in California (the brainstorm discussions in the big jacuzzi...), and collaborated with many early PACS research groups worldwide.
We also carried out extensive studies on the image logisticssavings and costs of the new digital system, and its diagnostic efficiency. The conclusions were very positive.

We founded the EuroPACS organization in those days. The 6th International EuroPACS Meeting was held April 25–26 1988 in Utrecht and Leiden. In Utrecht we demonstrated the digital reading room to the international PACS community, and had a memorable conference dinner in Museum 'van Speeldoos tot Pierement' in front of the big Dutch street organs playing.

A new career

From archive to ICT

I was lucky to find Andre Achterberg, archive worker, who was eager to participate in the PACS team. His knowledge of all procedures, workflow and adminsitrative hurdles was indispensible. We made great progress with the comparison analog-digital, and published a lot of papers. Andre later acquired ICT diplomas and endeavored on a new ICT career.

3D Visualization

Visualization of the skull thickness of a newborn, showing the vulnerable thin areas [Source].

Once the radiologists had digital workstations, they asked for 3D visualization in-house. At that time (1989), a just finished MSc student from Twente University contacted me, if he was allowed to work in our hospital as a substitute for his obligatory military service. I talked to him, we went into town for a beer, and talked till 02:00 AM. I was impressed, found funding and hired him: Karel Zuiderveld. He finally did his PhD with us on medical 3D visualization. He turned out to be a extraordinary programmer and 3D specialist [GScholar], and became later Director of Technology R&D at Vital Images Inc., one of the leading radiological 3D visualization workstation developers in the US.

The software package that was developed in our 3D Computer Vision group was called VROOM (Volume Rendering by Object Oriented Methods).
Below: Closest vessel projection [Zuiderveld1994], far superior to maximum intensity projection.


Prof. Paul van Waes with 3D printed models. Applications included the construction of STL models for missing parts of the skull from CT scans of the intact symmetric part. Work in collaboration with prof. Frans Zonneveld (Philips, Utrecht University).

During the celebration of the 100th birthday of Röntgen's X-rays in 1995 the Dutch Society organized a major event in the grand Circus Theater in Scheveningen, and I was invited as a keynote to speak about 3D visulization for 2000(!) visitors.

At that time, prof. Karl-Heinz Höhne at Eppendorf University Hospital in Germany was experimenting with beautiful 3D visualizations. We were lucky to collaborate.

The grant with UNC was extended by Philips Medical Systems in 1992 [pdf], which enabled even more regular mutual visits between Utrecht and Chapel Hill. Together we wrote a highly cited paper (cited > 3200 times on Google Scholar) on Adaptive Histogram Equalization [Pizer1987].

A) Original image; (B) interactively windowed result; (C) unclipped AHE result; (D) clipped AHE result; CT abdomen. From [Pizer1987].

It was very inspiring to see the lab at UNC, experience the many innovations (head mounted displays, vibrating mirror 3D, ' PixelPlanes', shape statistics) and meet many extraordinary people, as prof. Henry Fuchs.

The Medical Image Display and Analysis Group (MIDAG) is the oldest and most important medical image group in the US.

In 1989 the hospital 'Academisch Ziekenhuis Utrecht' (AZU) moved from the location in the old city to Utrecht University campus 'De Uithof', 6 km east of Utrecht, to a splendid new (one billion euro) building. Its new name was 'Utrecht University Medical Center' UMCU. 
I was involved in the planning of the new large radiology department, a project of several years, focusing on radiation safety, MRI magnetic safety, workflow management, digitization (phasing out the large-scale conventional photography film systems), integration with modern computer vision, PACS integration etc.
A very learnable period.  

Today the ​UMC Utrecht Center for Image Sciences (CIS ) is one of the most modern hospitals in the Netherlands in the field of medical imaging, with 13 MRI systems (0.5-7T), 4 MRI LINAC (MRI linear accelerator) systems, HIFU (High Intensity Focused Ultrasound), PET-CT, PET-MRI etc.

NVKF Lu​stru​m 1998