Innovation: Surgery enters a new dimension at Guy's: Software creates 3D images to help guide the hands of doctors

NEUROSURGEONS at Guy's Hospital are now using moving three-dimensional computer images of the inside of patient's skulls to plan and perform operations.

The 3D images are created by linking together separately derived 2D images of the patient's bones, nerves and blood vessels, using software developed by computer scientists at the United Medical and Dental Schools of Guy's and St Thomas's Hospital.

These images enable surgeons to plan and monitor their operations, making them faster and less invasive. The computer workstation displays a colour- contrasted, 3D image of the patient's head, cut away to mimic the surgical procedures.

So far the system has been used in about 30 operations to remove benign brain tumours. In such operations, if the surgeon can cut a hole in the skull and successfully navigate a route past critical structures such as major cranial nerves and blood vessels, the patient can be cured.

The verdict of independent assessors - surgeons who have attended operations carried out using the system - is that the technique is improving treatment.

Discussions are now taking place with manufacturers of the scanners, including Siemens and GEC, on selling the software. It may be possible to adapt the system to drive mechanical devices that would perform parts of the surgery.

Surgeons already rely on 2D images from X-ray computed tomography (CT), which shows where bones are, and magnetic resonance imaging (MRI), which shows the location of nerves, blood vessels and tumours, to plan operations. But because the scans are taken by different machines on different days, and often at different hospitals, it is difficult to overlay the images for a perfect match.

Even if the images are registered properly under the cumbersome existing methods, the composite image is still 2D: it does not show the 3D relationship between anatomical features. Even highly skilled surgeons may find it difficult to relate what is seen inside the patient's skull to the 2D images of it.

The software enabling 3D matching was developed by David Hawkes and his team, working closely with clinical colleagues led by Anthony Strong, neurosurgeon at the Maudsley Hospital, and Michael Gleeson, ear, nose and throat surgeon at Guy's Hospital.

It ensures that images from different machines are matched correctly. It uses 16 internal structures, such as the cochlea in the inner ear, as reference points to register one image on top of another and build up a 3D image. The registered images are displayed at the rate of 25 images per second. 'This makes the images move, creating a much stronger impression of 3D,' Dr Hawkes said.

'This is the first time any group has combined CT images of bone structures with MRI images of soft tissue and blood vessels to define the structural relationship of brain tumours in the base of the skull, for surgical planning and guidance.'

The merging of the images is accurate to plus or minus one milimetre.

Work is now under way to relate videos recorded during the operation to the pre-operative scan, so that at all stages of the operation the surgeon will know precisely where he or she is in relation to the scan.

The techniques of Image Guided Therapy are also being used to detect malignant tumours, by superimposing positron emission tomography (PET) images, which show the distribution of metabolic functions, with structural MRI and CT images.

This enables doctors to detect sites of active metabolism associated with the recurrence of malignant tumours.

Brainwave: the picture at the top of the page shows a sample of the three-dimensional images that are made up by the software from three merged scans, with the bone shown in white, the tumour in green and blood vessels in red. The composite picture beneath it shows examples of the scans that were used to construct the 3D image. At top left is a computer tomography image (CT scan) that shows the bones of the skull. The magnetic resonance (MR) image at right marks out the tumour as a large white area, and the bottom image, once again from an MR scanner, shows the blood vessels.

(Photograph omitted)