Development of the “EPID” megavolt X-ray camera in the NKI

Epid prototype.jpg 

This small image detector from 1984 is the first working prototype of the Electronic Portal Imaging Device (EPID) developed in the Netherlands Cancer Institute (NKI).

The photo shows the artificial head of the "Alderson" test phantom. With the syringe (right) the space between the printed  circuit boards is filled with trimethylpentane. Excess liquid flows into the bottle.


1st EPID air ionchambers.jpg 

The heart of the prototype image detector is a flat sandwich of two printed circuit boards.  On the boards a pattern of 30 copper strips of equal width is etched. The photo shows an earlier prototype of the detector which is filled with air.  The final model is filled with a fluid, increasing the sensitivity considerably. A short-lived memory effect occurs by which the measured signals in the pulsed radiation of the accelerator can be better processed.

The boards are turned at 90 degrees to each other so that a matrix is formed of 30x30 crossovers between the upper and lower copper strips. Each crossing forms an ionisation chamber of approximately two cubic millimetres.The total image surface is 7.5 cm x 7.5 cm. 

The photon radiation from the accelerator generates a small electric current in the ion chambers. The current is proportional to the intensity of the radiation at the location of each chamber. Within a few seconds the strips on one board are connected to the high voltage one after the other and the current strengths are measured from each individual strip on the opposing board. From the 900 measurement values an electronic megavolt image is calculated. 

The photo shows an earlier prototype of the detector which is filled with air.  The final model is filled with a fluid, increasing the sensitivity considerably. A short-lived memory effect occurs by which the measured signals in the pulsed radiation of the accelerator can be better processed. 

   The importance of online megavoltage image detection.                 In the clinical practice of radiotherapy in the 1980's megavolt portal film imaging becomes an important part of the daily quality control of radiation treatments. Treatment planning software and computer systems for dose calculations are improved and the clinic is able to design increasingly complex dose geometries. The daily number of portal image checks becomes so large that the processing of the films by hand cannot keep up. There is an urgent need for rapid and automatic portal image checks on the actual treatment process. Research groups worldwide work on the design of a megavolt image detector. This is often achieved by means of a video camera which, with the use of a mirror, records the image from a fluorescent plate, or by means of a scanning row of detectors. The Netherlands Cancer Institute (NKI) choses a radical approach by designing a flat detector which is just as easy to use as the traditional X-ray film cassette.

Development of the EPID megavolt X-ray camera.                     Physicist Harm Meertens is the first in the NKI to experiment with a home-built wire chamber. Here the ionisation of the air due to the radiation beam is measured between metal wires which cross each other. By placing two rows of wires at right angles to each other, each set of two crossing wires formes a small ionisation chamber and the combined current measurements from this array of ion chambers resulted in a blurred megavolt photo . The principle does work but it is not sufficiently sensitive for practical application.

In 1982 the development of the idea of Harm Meertens gets new impetus.  The young physicist in training, Marcel van Herk, is appointed to further develop the electronic X-ray camera. In the NKI, Marcel and electronic engineer Jan de Gans form a development team with high potential. Both are inspired technical developers and are able to turn their ideas into working constructions. They find each other also in their rejection of bureaucracy and create in the physics wing of the radiotherapy department a work climate in which ideas can ripen and what appear to be impossible is achieved.

1988 Jan de Gans.jpg

Prof. dr. Marcel van Herk (li)                                            ing. Jan de Gans (re)

With limited resources, the principle of a matrix radiation detector with readout by a computer is realised. With the help of the department's own instrument workshop and with spare parts from electronics shops, Marcel develops a sensitive matrix detector with fluid-filled ionisation chambers; first a prototype with 30x30 chambers, then a large matrix of 128x128 and finally a high resolution EPID of 256x256 image points. 

1986 EPID clinical.jpg 


128x128 matrix ionisation chamber EPID, image breadth 320 mm x 320 mm. Prototype built in 1986 in the NKI-RT workshops.


In 1986, for the first time, megavolt portal images are obtained with a completed EPID during an irradiation. In 1988 the experience with this EPID is reported [69].

1986 first patient image HN.jpg 

In 1986 the first image with the EPID is obtained in the clinic. At first sight there is little detail to be seen.  After processing with advanced software, the image provides important information about the geometry of the radiation treatment.  Together with the development of the EPID, the development of software for image processing is an important activity of the radiotherapy research group in the Netherlands Cancer Institute.

The EPID computer system. 

1986 epid system schematic.jpg 

System drawings of the complete system for online portal imaging of the NKI.

In order to have a read-out of the megavolt image, the copper strips of the upper printed circuit board are alternately connected to a voltage of 300 V.  Under high voltage, the collected charge of each of the 128 crossing strips is read out in 20 ms.  This process is repeated 128 times until the whole image is formed.  The total time for the formation of the megavolt portal image and making it visible is 3.1 s at most. 

Complete computers for this reading out of the image information were still too large and expensive for this experiment.  PCs did not exist at that time and the first computer chips were only just appearing on the market. With these first chips,  Jan de Gans develops the necessary fast computer systems and also designs the electronic switching of the matrix chamber.

JandeGans werkplaats web.jpg 

Jan de Gans in the electronics workshop of the Department of Radiotherapy.

The EPID computer system is designed and built by Jan himself. The printed circuit boards are of his own design. The empty circuit boards are manufactured externally. In case of urgent need, the circuit boards can be etched in the workshop. Computer chips and other components are added in the NKI. 

Epid Prototype CPU 286.jpg 

Microprocessor card for the EPID computer system.

This computer circuitry is also designed and manufactured by electronic engineer Jan de Gans. This system computer consists of an Intel 80286 16-bit microprocessor, 256 kbyte RAM, a fast 12-bit A/D converter and a hard disk.

The operating software is programmed bit by bit. Reasonably priced development software for the new configuration is not available. Jan and Marcel build their own software development system and program the software modules with which the EPID system is constructed.

The EPID development is extensively described in the doctoral theses, of Harm Meertens [63] in 1989, and Marcel van Herk [67] in  1992.

Bronnen & Publicaties

  • [63] “On-line acquisition and analysis of Portal images”, Harm Meertens. Doctoral thesis, University of Amsterdam 1989, supervisor Prof. Dr. J. Strackee. ,
  • [67] “An electronic portal imaging device; physics, development and application” by M. van Herk. Doctoral thesis, University of Amsterdam 1992, supervisors prof . dr. J. Strackee and prof. dr.H. Bartelink. ,
  • [69] Harm Meertens, Presentation during the 30th Annual Meeting of the American Society of Therapeutic Radiology and Oncology (ASRO), New Orleans, October 1988 ,