¿ìè¶ÌÊÓƵ

The scene is repeated in thousands of hospitals, all around the world,
every week of the year: a neurosurgeon is preparing to remove a walnut-size
tumour from deep within a patient’s brain. To plan an approach, the surgeon
examines the latest magnetic resonance imaging (MRI) scans taken of the
patient’s head. Each image represents a ‘slice’ of the head, 3 to 5 millimetres
thick, cut as if the patient’s head were a loaf of bread. The first slice
shows nothing but scalp and bone, but with each successive image the brain’s
structure emerges. MRI, which distinguishes tissue types by their proton
density, shows a different shade of grey for bone, grey matter, white matter,
cerebrospinal fluid, blood vessels-and the tumour itself. In the several
slices that cut across the tumour, the malignant tissue glows a bright white,
easy to distinguish from the darker healthy tissue around it, and from the
black dots and streaks that represent the veins and arteries entangling
it.

But this information presents surgeons with a formidable challenge:
they must take these two-dimensional, black-and-white images and create,
in their own brains, a three-dimensional map of the tumour, the healthy
brain tissue around it and the blood vessels on every side of it. With this
mental map, the surgeons must plan an approach that damages as little healthy
tissue as possible, does not cut any blood vessels and yet enables them
to remove every cell of the tumour.

The task becomes even harder when they start to operate. Unlike thoracic
surgeons, who can make a long incision, pull the ribcage open and have clear
access to the heart, neuro-surgeons must drill a hole in the skull and gently
move delicate brain tissue aside to reach the tumour. And the tumour, at
the end of this wet tunnel, looks no different to the surgeon’s eye from
the healthy brain tissue around it. ‘In such situations,’ says James Zinreich,
a neuroradiologist at Johns Hopkins Medical Institutions in Baltimore, ‘surgeons
have two choices: either remove a great deal of healthy tissue to make sure
that they get all the tumour, or be conservative and risk leaving part of
the tumour. Since the ‘healthy tissue’ is the patient’s brain, they almost
always choose the latter.’

Zinreich, along with other radiologists, surgeons and computer programmers,
is working to change this picture, to literally bring it into another dimension.
Armed with graphics workstations similar to those used by engineers and
designers, the team is creating three-dimensional images from the two-dimensional
data provided by MRI scans and computerised tomography (CT), which distinguishes
tissue types by the distribution of radionuclides.

For surgeons, this technology means the ability to plan, and even rehearse,
operations. For patients, it means shorter, less invasive procedures. Within
a few years, it may even lead to operations in which brain tumours are destroyed
using a needle, with little tissue damage and virtually no recovery time.

Zinreich and his fellow radiologists and radiological technicians work
closely with surgeons to plan cranial and spinal surgery. They begin with
what is normally the end product: CT and MRI scans. Workstations display
two-dimensional maps of slices of a patient’s body on large, high-resolution
monitors. The radiologists then use false colours to label the features
in each slice: skin, bone, air, cerebrospinal fluid, grey matter, white
matter, tumour. When the computer finds a feature in the image of a new
slice that has the same colour, or grey-scale value, and approximately the
same location as the area labelled ‘bone’ for instance in the previous one,
it labels the feature ‘bone’ as well. This segmentation of images is similar
to the process used to add colour to old black-and-white movies. The colouriser
does not need to tint John Wayne’s bandanna red in each of the 180 000 frames
of a black-and-white film; it is done once, and the computer remembers to
make that feature red for the rest of the movie.

At Johns Hopkins a superminicomputer, using eight processors working
in parallel, converts the two-dimensional, black-and-white images into three-dimensional
colour ones.

For a simple problem, such as a herniated disc in the spine, this may
be the final step. A single side view in three dimensions, showing the disc
in relation to the vertebrae and spinal cord, may be all the information
a surgeon needs. A technician takes a 35-millimetre colour slide of the
screen image, which depicts the scene in three dimensions, and hands it
to the surgeon.

For more complicated problems, such as brain tumours, the computer allows
surgeons to rotate the image of a patient’s head on screen to examine the
growth from any angle. The surgeons can peel away tissues, layer by layer.
They can remove all skin, and in a few seconds the pink flesh will be peeled
away, replaced by bare white skull. When they remove the bone, the cerebral
cortex-the grey matter with its gyri and sulci-is exposed, and this too
can be rotated to find the least invasive route to the tumour. Surgeons
can also subtract every type of tissue until only the tumour is displayed,
then work backwards by adding the blood vessels, white matter and grey matter.
They can also make any layer or layers transparent, so that a tumour, for
example, is visible through the bone, brain and blood vessels overlying
it.

David Altobelli, a cranio-facial surgeon at Boston’s Childrens Hospital,
uses the technology to rehearse operations. One of his most recent patients
was a 14-year-old girl with Pfeiffer’s syndrome, a congenital defect that
distorts the major bones in the face. The girl’s eyes were too far apart
and protruded so much that the orbital bones, which normally surround and
enclose the eyeballs, did not protect them. Her lower jaw protruded beyond
her upper jaw, which was unsightly and made eating difficult.

‘What we needed to do was move her forehead and orbits forward, move
her eyes closer together, bring her upper jaw forward, move the lower jaw
back and get them to mesh,’ says Altobelli. The operation would be complicated
and each step would depend on the results of the previous step. In the past,
the patient might have had to undergo two or three separate operations,
each followed by weeks or months of recovery.

Instead, Altobelli, working with radiologists Ferenc Jolesz and Ron
Kikinis, rehearsed the operation on screen. Using imaging software that
Kikinis and Jolesz have developed with scientists at General Electric, the
trio produced three-dimensional images of the patient’s head. Altobelli
‘peeled back’ the skin from the girl’s face to expose the bones beneath.
He ‘cut’ bone from the orbits and ‘moved’ the patient’s eyes closer together.
He ‘repositioned’ the jaws and simulated chewing on screen to ensure that
both halves meshed smoothly. After each screen osteotomy, or bone excision,
the computer readjusted the patient’s skin to the new bone structure, so
that Altobelli could estimate how to remould the child’s face.

Practice makes perfect

Altobelli also used the computer data to program a milling machine to
produce a plastic model of the skull, which enabled him to practise his
surgery. ‘I found that there was not enough room to cut complete circles
of bone away around her eyes, so instead we removed C-shaped sections from
each orbit, then brought her forehead forward,’ he says.

After hours of rehearsal, Altobelli was able to restructure his patient’s
facial bones, and the soft tissues covering them, in a single operation.
Extending the technique to even more elaborate operations, for which there
is no textbook procedure, promises to make rare surgery less traumatic-and
less expensive-for patients, says Kikinis. ‘It even makes some operations,
rejected in the past as too complex and risky, possible now,’ he adds.

Surgeons in hospitals across North America, Europe and Japan now use
three-dimensional imaging equipment before operating on knees and hips as
well as on heads and spines. At Johns Hopkins, Zinreich has gone one step
further: he has connected the imaging system to a robotic arm to which is
attached a blunt pointer, or sensor, that guides the surgeon’s actions as
the operation progresses.

To remove a tumour deep within the brain, for instance, the operation
begins with a monitor displaying an image of the patient’s head and brain,
with the growth highlighted in false colour to distinguish it from surrounding
healthy tissue. The medical team must then show the sensor how the image
on the monitor relates to the patient on the operating table. To do this,
radiologists mark five prominent features on the image of the patient’s
head, such as the bridge of the nose, the cleft in the chin, the space between
top front teeth and the centres of each ear lobe. With the sensor, surgeons
touch each corresponding location on the real patient, whose head is fixed
in a support frame. An image of the sensor then appears on the screen, and
every movement the real sensor makes is shown in relation to the image of
the patient.

With the monitor showing the three-dimensional image of the patient’s
brain, the surgeons move the sensor over the patient’s head until the monitor
shows the best spot to make the first incision. They then put the sensor
to one side and take up a scapel. After the surgeons have cut away a flap
of skin to expose the skull, they take up the sensor again to guide them
for the second incision. On the screen, the patient’s ‘skin’ is peeled back
in exactly the same way, exposing imaginary white bone underneath.

Once through the bone, the surgeons use the sensor to guide them through
the narrowest and shortest passageway to the tumour. When they reach the
tumour, they can not distinguish it as they peer into the incision, because
it looks no different from the healthy tissue around it. But the monitor
clearly shows the tumour; on MRI scans, the growth glows bright white against
the darker healthy tissue. Using the monitor to guide them, the surgeons
insert a scalpel or laser into the incision and carefully remove or destroy
the malignant tissue.

Finally, they insert the sensor into the void they have cut in the brain.
When the sensor’s image on the monitor reaches the boundaries of every bit
of the former tumour, the surgeons can be confident they have removed all
of it, even though they could hardly see the scalpel or laser remove the
diseased tissue. The technique enables surgery to be more precise and less
invasive. ‘The size of the cranial flaps that have to be opened is much
smaller,’ says Zinreich. ‘Reducing the size of the incision reduces the
danger of infection and speeds recovery.’

Radiologists want to develop the technique further so that neurologists
can go well beyond conventional surgery, which needs large incisions to
lay open and get tools into the brain. Along with three-dimensional images
for studying the operation in advance, they would like to be able to present
the images as they are needed-during surgery in the operating theatre. Jolesz
is already performing animal experiments with just this goal: surgery under
MRI control.

Jolesz forecasts that powerful computers, using many processors working
in parallel and connected directly to the MRI scanner, would provide images
on a monitor in the operating theatre. Surgeons would drill a tiny hole
in the patient’s skull, then insert a hollow needle into the brain. They
would guide the needle towards the tumour, watching its progress on the
screen to avoid blood vessels and vital areas of the brain. When the needle
reached the tumour, the surgeons would turn on an infrared laser, delivering
its power down an optical fibre threaded through the needle. They would
‘cook’ the malignant tissue until it was destroyed. With MRI scanners able
to provide a temperature map on the screen, accurate to 1 degree C, says
Jolesz, surgeons ‘could tell when all the tumour had been heated enough
to kill it’.

Complex imaging devices and computers, along with high-technology surgical
tools, are stretching the limits of conventional surgery, says Jolesz. ‘This
will result in less human suffering, shortened hospital stays, faster recovery-but
most important, improved outcomes.’

Jonathan Beard is a freelance journalist based in New York.