IMPORTANT LEGAL NOTICE - The information on this site is subject to a disclaimer and a copyright notice.

European Flag

   EuropaThe European CommissionResearchPress releases

Building a bionic body? - Biomaterials research for healthcare
Media Briefing - Brussels and Leuven, 13 April 2000


PHIDIAS: Laser PHotopolymerisation models based on medical Imaging, a Development Improving the Accuracy of Surgery

Fried Vancraen - Materialise NV, Belgium

See also: images


The PHIDIAS1 project set out to build three-dimensional (3D) medical models derived from tomography-based medical images (CT, MR). By laser photopolymerisation (stereolithography) even complex models can be generated very fast in low-toxicity materials.

As the ultimate 3D representation, solid medical models can be very helpful during the preparation of complex surgery. Their main uses are as follows.

They provide visual and tactile documentation for diagnosis, therapy planning and didactic purposes. The medical model facilitates communication between surgical team members and patients.

They enable accurate planning and simulation of interventions. For instance, in facial surgery, life-size individual models of the patient's skull are used to rehearse osteotomies, enabling to measure displacements in advance.

They serve as a master or negative for prosthesis or implant production.

In the last few years, several (mainly cranio-facial) surgeons have started to use medical models in the preparation of complex surgery. The results are very promising and in some cases even spectacular. However, all these activities are the isolated efforts of different research groups, which do not result in a global solution. The Phidias project aimed to investigate the full process of making medical models and to streamline the process from the patient in the scanner through to preparing the operation.

One of the first objectives of the project was to investigate the requirements for the different applications and to meet these needs with new technological developments.

The request for lower costs and shorter lead times has led to the development of faster and more reliable CT imaging techniques, easy segmentation tools and faster production machines.

In order to use the models in the operating theatre, low-toxicity resins were developed with various sterilisation possibilities.

To differentiate between different structures in one model (e.g. tumour and skull base) a new process was developed: Colour Stereolithography.

Partners in the Phidias Project

Materialise (Belgium) is a software developer for Rapid Prototyping applications with research activities in the medical applications of Rapid Prototyping: together with the Dept. of Mechanical Engineering from the University of Leuven, it has years of experience on resin experiments, laser drawing strategies, for Laser Photopolymerisation and data preparation for medical applications.

Materialise is integrating the research results as co-ordinator of the project. Being the rapid prototyping knowledge centre within the project, it is concentrating on software for the technical data preparation and the laser photopolymerisation process on a dedicated experimental set-up. It supplies test parts and medical models, the ultimate materialisation of the research results, for validation to the different medical centres.

Siemens Medical Systems (Germany), the major CT-scanner manufacturer of Europe, leads the investigation to generate images with less distortions and artefacts.

Zeneca Specialties (UK), having many years of experience with UV-curable polymers in the medical field, is developing the resins that are suited for the medical models produced by laser photopolymerisation and develops the new Colour Stereolithography resins.

The Laboratory for Medical Imaging Research of the KU-Leuven (Belgium) is working on tools for efficient image segmentation and flexible 3D visualisation and manipulation that can be interactively controlled by the responsible radiologist. The Radiology department of the academic hospital Gasthuisberg of the KU-Leuven is charged with the user requirement analysis and validation by different surgeons.

Technical Achievements

The Phidias project team has investigated the full process of making medical models and streamlined the process from the patient in the scanner until the operation preparation.

Image generation

In order to produce high quality and high accuracy models, the image generation is extremely important. If artefacts are introduced during the image generation it is impossible to correct them later on.

Siemens Medical Systems concentrated its work on developing new spiral scanning techniques which reduce the artefacts and give better spatial resolution.

The work resulted in a patented interpolation approach for image generation. This approach reduces stepping artefacts caused by inconsistencies in the interpolated data and 'Flying pixels' due to inhomogeneity of noise and spatial resolution. Also homogeneity artefacts are investigated. Motion artefacts can be considered minimised by the use of Spiral CT where the short scanning periods reduce the risk of moving patients during scanning.

Image Segmentation

The standard procedure to select bone out of CT images is thresholding. All pixels which have a value higher than the threshold value are assumed to belong to the bone. This binary thresholding is only successful for large bone structures which have a high density. Very tiny structures or bone structures with low density will have a value under this threshold (under-segmentation). Other structures will be so close to each other that the area in between will have values above the threshold (over-segmentation).

Primary solutions for both problem areas have been developed by KU Leuven (Lab. for Medical Imaging Research) via image enhancement algorithms. These algorithms will take into account both the contrast and the gradation in the images. They enable a much higher degree of automation in the segmentation process, thus reduce the effort required from the radiologist. In addition, the resulting surface quality is better.

In a second phase the 2D results were coupled to an environment for 3D interaction in shaded images based on existing advanced visualisation and interaction equipment.

Support for 3D interaction in regions of interest pointed out on the 3D surfaces has been implemented by a virtual sculpting approach. In this method not only models derived from medical imaging data, but also 3D representations of prostheses and implants can be manipulated in the same 3D space. This development can be seen as a step towards custom-made fabrication of implants

Another advantage of virtual image-based sculpting is that surgical procedures can be simulated beforehand in a non-destructive way. New applications requiring models can be investigated before actually using (and often destroying) the physical models. Materialise has transferred most of the results obtained at KUL to the Medical Modeller Workstation.

Medical Laser Photopolymerisation resin

In terms of the resin development there were two main objectives:

- a non toxic resin, which can be used in theatre room.

- a resin system which can be selectively coloured.

Both of these aims were fulfilled in the Phidias project. Patents were filed to protect the developments.

Zeneca colourable and clear resins are formulated with raw materials which have minimal toxicity and are TSCA and EINECS registered so that the final resin when processed or used by the end users such as surgeons would have a good clean safety profile. A number of samples were submitted for Ames (genotoxicity), skin sensitisation, acute oral and dermal and eye irritation testing. These involved extensive testing of liquid resin to anticipate the handling of liquid at Service Bureaux and recommending handling procedures for SL resin. Selected resin samples including clear resin and colourable resins were submitted for testing. Results to-date indicated that Zeneca resins have low skin sensitisation, oral, dermal, and eye irritation with negative limited Ames (genotoxicity) results.

Furthermore, FDA approved USP class VI tests were carried out on selected Zeneca clear and colourable resin. Both resins passed when tested for USP class VI, indicating the safe nature of the Zeneca resins in the cured state so that models will be safe in close contact to the patient during operating procedures.

After study of both the chemical background (by Zeneca) and the machine implications (by Materialise) it was decided to work with a one Laser System instead of a two Laser System. The same Laser will be used for curing and for coloration. The principle is that the critical energy to start the coloration process must be sufficiently higher than the critical curing energy.

The laser photopolymerisation process

Materialise developed software that interfaces from the Medical Modeller Workstation to any kind of Rapid Prototyping system. Where the Workstation is used to define the regions to be processed and to display the segmentation result, this software interpolates the data and interfaces directly to the RP machine. Because of this direct interface and because of the higher order interpolation algorithms it produces the most accurate models in a very short time (scanner images to machine in 1 hour).

Materialise has implemented algorithms for cubic interpolation in Z, X and Y direction. Because of the morphological forms of the medical models, we know in advance more ore less how they will look like: human organs or structures do not have sharp edges or very irregular forms. We can take benefit of this knowledge by performing a cubic spline convolution between four adjacent scans.

The system also includes contour based support generation. A graphical user interface has been added to work on the layer based support structures. 3D visualisation of the contour information is made possible. The editing of the support structures to add or remove elements is implemented in a user friendly environment that allows complex manipulations, such as trimming on stacks of layers.

A new support building style was introduced: perforated supports. (patent pending). This reduces the support building time and eases of support removal.

On one hand, most of the technical achievements related to the laser photopolymerisation process are usable on advanced, but standard, stereolithography machines. Special support building and scanning algorithms, as well as the resins with low toxicity and even colour, can run on standard machines.

On the other hand the important objective of speed in medical applications was hard to achieve due to the slowness of standard recoating systems. For this purpose a lot of effort has been invested in the development of a new recoating system and in this aspect the prototype Medical Modeller Machine is very specific. A curtain coating based recoating system was introduced and patented.


Validation of the results has been performed on an technical and clinical level.

The accuracy of medical models is limited by the accuracy of the medical images and the equipment used for scanning. However, it is possible to overcome the resolution limitations of the scanner with intelligent interpolation approaches. For specific applications, accuracies up to a few tenths of a millimetre can be reached. This reassures that implants can be based on medical models.

The clinical validation in the Phidias project has been the first study that goes beyond individual case reports. 25 surgeons have been using models for validation in the treatment of 48 patients in craniofacial, orthodontic and orthopaedic applications

The models were used for visualisation, for surgical planning and implant preparation. Models proved to be useful in the communication with colleagues and patients. SL-model lead to different decisions in many of these patients (Fig. 7).

The decision to operate or not changed in 11 patients, the composition of the surgical team changed in 6 cases, skin incision altered (n=8), patient's position for surgery changed (n=3), the choice of osteosynthesis material altered (n=13), the implant site of osteosynthesis devices varied in 13 cases, surgery sequence changed for 16 patients and the number of imaging studies altered in 16 cases. This proves there substantial impact on surgery and improvement potential.

Overall assessment of usefulness of SL models based on a validation of 48 patients by 25 surgeons

Overall assessment of usefulness of SL models based on a validation

of 48 patients by 25 surgeons.


The overall assessment of the models by 25 surgeons proves that the initial thesis of the project is true. Medical models help reconstructive surgeons to make their patients again resemble the ideal human being of the Greek sculptor Phidias. During the project, six patents have been filed by the industrial partners. This shows the interest of the partners in the research results and their belief in the commercialisation potential.

The Greek sculptor PHIDIAS -fourth century BC- is known for the technical and artistic quality of his representations of the human being, full of dignity and nobility. His conserved masterpiece, the frieze of the Parthenon, is still today a great symbol of European culture. The medical models resulting from this project should contribute to make disabled, injured or ill persons resemble once again the ideal human beings of PHIDIAS.

Search Top
PRESS RELEASE OF 27.03.2000 | 18.04.2000