PHIDIAS brought together an SME,
an interdisciplinary university research group and two of Europe's
largest companies to adapt the latest 'rapid prototyping' technologies
from manufacturing industry to surgery.
As a result, surgeons can now have highly accurate models of their
patients' internal organs just a few hours after performing CT or
MR scans. The models improve both diagnosis and surgical planning.
They can also be used as masters for surgical implants or prothesis.
By mid-1996 the SME leading the project had sold 20 licenses world-wide
and opened a sales office in the USA, while both large companies have
improved their product range. The three companies have applied for
at least seven patents between them.
3D medical imaging systems, such as Computerised
Tomogaphy (CT) scanners, have been allowing surgeons to 'see' inside
their patients and plan appropriate treatments since the 1970s.
For surgeons planning complex operations, however, a solid model
of their patients' organs and bones would offer many advantages
over the images displayed on the scanners' computer screens.
To begin with, a model would make interpreting the scan data far
easier, ensuring that the surgeons are better informed on what they
will see on the operating table. Having a model to work with also
improves communication within the medical team and between doctor
and patient, and allows the surgeons the luxury of a 'dry run' before
Until recently, unfortunately, there was no way of creating such
models. In the late 1980s, however, manufacturing industry commercialised
rapid prototyping, a new technique with intriguing possibilities.
The BRITE-EURAM project PHIDIAS was launched to apply rapid prototyping
to surgical applications.
Manufacturers had always either carved their design models out
of solid blocks ('milling') or made them by hand. Rapid prototyping
(RP), by contrast, builds models upwards, layer by layer. There
are several techniques - one, for example, involves laying down
a series of carefully cut pieces of paper - and all translate computer-generated
3D images into solid objects in a few hours.
The technique can, unlike milling, create objects with internal
spaces - such as the sinus cavities in a human skull. The Belgian
SME Materialise NV, recognising the potential for medical applications,
bought a stereo photolithography RP system from 3D-Systems in 1990.
The machine essentially builds each layer of the model from a resin
which is 'cured', or solidified, by a laser.
After first confirming their idea's basic feasibility, Materialise
launched the PHIDIAS project in 1993 with Siemens Medical Systems,
Europe's largest producer of medical scanners; Zeneca, a world supplier
of stereo lithography resins, and the interdisciplinary medical
imaging research group of the Katholieke Universiteit Leuven (KUL),
a Belgian university.
One of the first steps was to find out what surgeons wanted from
the models, so the KUL surveyed 30 surgeons from Europe and the
USA. As expected, while some requirements were universal (non-toxic
resins, reasonable accuracy, hardness), different specialists had
different demands - dental surgeons need sub-millimetre accuracy,
for example, while emergency spinal or skull surgery often requires
the models within 12 hours.
Higher resolution and accuracy
The project structure was straightforward, with one partner responsible
for each stage of the process. Siemens, for example, focused on
data quality, as solid models require much higher resolution and
accuracy than normal scans.
They concentrated on 'spiral CT' imaging, where the CT scanner is
moved continuously around the patient in a spiral pattern. Unlike
normal scans, which requires the patient to lie motionless for up
to 20 minutes, a spiral CT scan only takes around 30 seconds, significantly
reducing inaccuracies caused by patient movement.
The disadvantage is that, unlike normal scans, spiral CT data has
to be heavily processed into a series of 2D images, each representing
a thin slice of the patient. In addition, Siemens had to develop
a suite of mathematical techniques to 'fill in the gaps' between
the data points.
By the end of the project, Siemens had successfully developed high-power
algorithms which both improved the resolution of each slice by a
factor of 100 and filled in the gaps between each slice, producing
the smooth curves required for medical models. Apart from making
an essential contribution to the production of medical models, their
work has improved their entire product range and resulted in two
new patent applications.
Apart from coordinating the project and marketing its results,
Materialise developed the software for the Medical Modelling Workstation.
This equipment takes the data, displays it to the radiologist and
surgeon, helps them select the objects (such as bones, muscles and
organs) they want produced and prepares the data for the rapid prototyping
A significant challenge was to automate the selection of the body
parts to be transformed into a model. Dedicated software was developed
so that if the surgeons selected just one data point - representing
a fragment of the lower jaw, for example - the workstation could
identify all the data points representing the entire jaw. In this
way the surgeons can select whatever they want modelled in a few
minutes, rather than laboriously specifying hundreds of data points
Another problem is that many body parts are shaped so that they
cannot stand upright as they are being created within the Rapid
Prototyping machine. The partners therefore wrote software to help
the users add special support structures into the model.
Extra dimensions of data
Zeneca, naturally, contributed the resins. As for Siemens, the
new application set them some challenging requirements - the resins
had to be entirely non-toxic, for example, as they would be used
in surgical theatres.
More importantly, almost all of the surgeons surveyed by the KUL
agreed that simple 3D models faced one outstanding problem: they
could not show the difference between tissue types. For a surgeon
planning to remove bone tumour from the surrounding bone, for example,
this distinction is literally a matter of life and death.
The partners took their cue from the medical scanners, which display
bone, cartilage and other tissue types using different colours on
their screens. But how to add colour to the models? Zeneca solved
this problem by studying resins that can be coloured, as well as
cured, by laser beams.
The resulting non-toxic resin becomes a transparent solid when cured
at normal laser beam intensity, and goes darker and darker red as
the beam intensity increases. In this way different organs, or tumours
can be shown in red throughout the transparent model. Zeneca has
applied for at least two patents on this new product.
KUL validated the technology throughout the project, providing
more than 20 surgeons with experimental models for almost 50 different
patients. The surgeons rated the models as Not Useful (6% of the
time), Useful (33%), Very Useful (40%) and Essential (21%).
Materialise and its partners have therefore developed, tested and
patented a new technology which looks set to radically improve complex
surgery. For Materialise, which has already sold around 20 licenses,
opened a sales office in the USA and increased its size from five
to 40 employees, this is just the first step. The future, they believe,
will see larger scale applications in customising implants and prostheses.
Today, for example, inserting a knee implant involves opening up
the knee and selecting the 'best fit' implant from one of five standard
sizes and shapes. PHIDIAS technology will allow surgeons to design
each implant for each patient, ensuring a perfect fit and improving
the implant's lifetime and effectiveness enormously.