Beyond the cutting edge
High-end medical application techniques are coming back into mainstream computing.
Surgery has been slow to adopt some of the automation and IT robotic technologies that have transformed other fields over the last generation. However, it is now making up for lost time.
Greatest interest has been created by new 'minimally invasive' robotic techniques that reduce (or even eliminate) external scarring and enable surgeons to operate remotely, and advanced technologies are branching out to benefit the whole spectrum of operation, from diagnosis to post-operative recovery. At the same time, IT developed for surgical applications is finding its way into commoditised applications.
Some of the challenges involved will be familiar to practitioners in other IT-related fields, such as the need to minimise delay and jitter in communicating with robots as they perform operations, and the importance of software that is resilient to failure. Training is also an issue, since surgeons may no longer be in immediate contact with their instruments, instead directing them from a computer console, without being able to rely on touch or direct hand-eye coordination.
In a few cases surgical instruments are being dispensed with altogether, as in 'ultrasound ablation', which can be used to remove some tumours by focusing high-frequency sound on the specific area within the body, destroying the target cells by heat. Microwaves or higher frequency radio waves can also be focused on target areas with the same effect, and, for neurosurgery, gamma rays can be used for eliminating tumours and treating other conditions, such as Parkinson's Disease.
These robotic and other minimally invasive techniques rely on new imaging capabilities to allow excisions and other tasks to be performed remotely, or, in the case of ultrasound or radiation ablation, to ensure that the beam is focused on the right spot. Equally important in the case of robotic surgery are algorithms that compensate for the loss of hands-on legerdemain on the part of surgeons; indeed, surgery has become one of the most challenging applications of robotics, stimulating research across the whole field.
Debugging the software is an issue, given that operating errors can literally be fatal. Although surgeons can practice robotic techniques on animals (especially pigs, because of their relative anatomical similarity to humans), there is limited scope here for expunging bugs that may not readily show up during a trial operation.
The first keyhole operation - on a dog, by Georg Kelling in Dresden in 1901 - involved a laparoscopic procedure, in which surgical instruments such as scissors, forceps and scalpels are inserted through small incisions typically around 1cm across, avoiding opening up the whole field of operation. The first laparoscopic operation on humans followed in 1910, conducted by Hans Christian Jacobaeus in Sweden, to remove scar tissue from the chest of a tubercular patient.
But keyhole surgery made little further progress for over half a century, remaining confined to a few relatively straightforward operations, often gynaecological. The simple reason was that it was hard for surgeons to see what they were doing once the instruments had been inserted through a small incision. But this all changed with the introduction in the 1970s of single-chip video cameras coupled with telescopic magnifying systems that, together, could be inserted alongside the surgical instruments through endoscopes or laparoscopes (respectively tubes inserted through the throat or small incisions in the chest or abdomen) to magnify the field of view. The image could then be projected onto a screen, with the additional advantage that the surgeon now had both hands free to manipulate the instruments.
There were important accompanying developments, notably improved anaesthetic procedures, which were significant because keyhole operations tend to last longer because of the extra technical difficulty involved. Taken together, such advances led to greatly increased use of minimally invasive techniques for an ever growing range of operations, particularly for abdominal and chest surgery.
A recent extension of keyhole surgery has come with the development of non-invasive techniques that avoid making an external incision at all, instead going in through orifices. Internal incisions may still be needed to reach the site of operation, but orifice surgery, dubbed NOTES (Natural Orifice Translumenal Endoscopic Surgery), has the obvious appeal of avoiding visible external scarring completely. Healing also tends to be faster and risk of infection is reduced. The technique is only just being introduced and Europe's first NOTES operation was performed in May 2007 to remove a gall bladder from a female patient in Strasbourg. The instruments were inserted through the vagina whose rear surface was then pierced in order to enter the abdominal cavity. Healing was very quick, and the patient returned home after two days with no external evidence that surgery had occurred.
Neither NOTES nor 'traditional' keyhole surgery depend absolutely on new technology, but in practice require advanced imaging. Furthermore, both are ideal vehicles for robotic devices. Eventually small robots will probably perform operations unaided, with the surgeon reassigned to being a program---mer rather than an operator.
Robotic surgery has various potential advantages, notably in avoiding the hand tremor that even the best human surgeons suffer from, and allowing operations to be performed remotely in telesurgery, which can be invaluable for dealing with emergencies. Telesurgery also allows less common types of operation to be performed at hospitals close to the patient without the surgeon having to travel.
Furthermore, telesurgery allows training and remote collabora-tion, enabling the required skills or consultancy to be provided from different places around the world. The initial impetus for telesurgery has come mostly from defence agencies, especially in the US, where the Telemedicine and Advanced Technology Research Center has focused its research particularly on battlefield applications.The aim is to develop portable surgeries in the field, equipped with robots that can perform emergency reconstructive surgery under control of a consultant residing safely in a hospital back home.
But robotic surgery is already being used quite widely for some 'civilian' applications, notably for prostate cancer removal. This a common yet tricky operation that plays to one of the strengths of robots, which is the ability to operate accurately within more confined spaces and at smaller dimensions than humans. The traditional prostate operation is often successful in removing the tumour without immediate recurrence, but quite commonly with the unfortunate side effects of leaving the patient impotent or incontinent or both. This is because the muscles controlling the bowel, and nerves triggering erections, are close to the prostate and easily damaged during the operation. Use of miniature, precisely targeted robots reduces that risk, and this has already led to a rapid increase in their use in the US. The Da Vinci robotic system, from Intuitive Surgical, has made the biggest impact on prostate cancer, with some 700 insalled systems being used now for a growing percentage of such operations in the US.
Although less prolific for other types of surgery, robots are starting to be used for more complex procedures on the heart and brain. There is footage on the Internet of one of the world's first robotic operations to repair a damaged mitral valve, which controls the flow of blood back into the left side of the heart after it has been oxygenated by the lungs. Without treatment, a damaged mitral valve forces the heart to pump harder, enlarging it, leading to symptoms such as shortness of breath and, ultimately, cardiac failure.
The operation was traditionally performed via open-heart surgery, but can now be done robotically with much less invasion and risk, along with faster recovery time. The transmission of a full operation of such complexity over the Internet in October 2007 was itself a milestone, but the real significance lay in the execution of a procedure requiring high skill by robots. The operation is not routine, with account having to be taken of individual anatomical differences between patients that only become clear as surgery proceeds.
Such operations have been made possible by improvements in the ability to control and manipulate robots and compensate for the loss of dexterity on the part of surgeons. One of the challenges lies in simulating the feel that a surgeon normally has in order to generate the right amount of force by robotic devices. This is not straightforward, because surgeons well trained on robotic systems become more adept at applying just the right touch using just visual cues on the display, while novices struggle. The traditional 'feel', however, can be simulated by a technique called haptic feedback, allowing surgeons to sense from the controls when the force applied by instruments changes.
A recent study at Johns Hopkins University in the US recruited surgeons with varying robotic skills to perform basic tasks with and without haptic feedback. "The conclusions were that visual force feedback resulted in reduced suture (stitching material) breakage, lower forces, and decreased force inconsistencies among novice robotic surgeons, although elapsed time and knot quality were unaffected," says Allison Okamura, one of the study's authors. However, experienced surgeons derived no benefit from haptic feedback. The broader conclusion is that robotic systems need to cater for different skill levels with a variety of optional feedback and imaging techniques. It is advances in imaging that have really enabled all these less invasive techniques to come forward; indeed, imaging techniques and robotics are inseparable and come under the heading of robotic surgical systems.
Improving surgery's image
The first video cameras used for keyhole surgery yielded rather flat images, but since then techniques such as computer tomography have allowed much more accurate and illuminating 3D images of cross sections to be reconstructed. These have been augmented by a litany of non-visual imaging techniques such as X-ray radiography, magnetic resonance imaging (MRI), fluoroscopy, and ultrasonography. Some of these techniques originated for diagnosis rather than imaging during surgery, but the line between the two is blurring.
A major objective of malignant cancer surgery is to remove all malignant cells to avoid a second operation, but at the same time take out as little healthy tissue as possible. It is impossible to avoid taking out a little healthy tissue, and this is referred to as the 'surgical margin'. In the case of a brain tumour keeping the surgical margin down can make the difference between retaining and losing vital functions or mobility.
Until recently, surgeons tended to estimate how much tissue to take out, relying on rather large surgical margins. The only way to do better was to send tissue away for biopsy during surgery, but this was often impractical because the analysis could take half an hour or more, with little scope for repeating it during the operation. But there are signs that the traditional biopsy could be replaced by real-time imaging techniques both in diagnosis and surgery, using minute cameras as small as 1.8mm in diameter, that can either be inserted into endoscopic/laparoscopic tubes or even sited on surgical instruments. The latter can give surgeons better visibility while conducting keyhole operations.
Apart from direct visualisation, these cameras can be used in conjunction with light sources to analyse tissue and identify tumour cells during the course of an operation, giving cancer surgeons real-time feedback, and enabling them to work with much smaller margins. Tumour cells scatter light differently from healthy cells, and this can be detected by the cameras - indeed it may be possible to separate cells into categories of malignancy. But the data can be somewhat subjective at present, leaving surgeons to judge whether they have cleared all the malignant cells and reached healthy tissue.
There is a need is for algorithms that automate this tissue identification process, and one place where these are now being developed is Carle Foundation Hospital in Illinois, US.
"The key here is that all types of cell, including the various categories of tumour cell, produce distinctive coherence patterns that can be readily detected," says Stephen Boppart, director of the Carle Foundation Hospital's Mills Breast Cancer Institute. "There is the potential for automating the process of cancer detection to make life easier for the surgeon and reduce the opportunity for mistakes. There is even the potential in future to hook these algorithms up to robots that perform the tumour excision automatically, perhaps leaving the surgeon just to verify that the task has been completed correctly."
For invasive surgery, technology is playing an increasing part in pain relief both during and post operation, by accurately locating critical nerves that can be stunned. In particular, the ability to locate nerves can be exploited to replace general anaesthesia, which can be risky for some patients, with local anaesthesia, which leads to faster recovery.
Anaesthetists at the Mayo Clinic in Minnesota, US, for example, have applied 3D ultrasound imaging to locate and block nerves in 150 varieties of surgery, including some where general anaesthesia was necessary before. In one such operation, a major tumour in the middle ear was removed under local anaesthetic, which would always have required a general until now.
As the Mayo Clinic Florida anaesthetist Steven Clendenen notes, the key to extending local anaesthesia to such operations lies in matching the 3D imaging with detailed anatomical knowledge of nerve networks pinpointing the nerves serving the target location of the operation. "We thought nerves could be very ultrasound friendly because they run in a linear plane and are wrapped in a sheath of fat," says Dr Clendenen, "so they easily reflect back ultrasound waves."
The 3D imaging also allows the spread of the anaesthetic to be monitored after administration, so that the surgeon can begin the operation just as soon as pain relief has taken effect, but no sooner.
This theme of applying technology to reduce trauma on the patient runs through many of the current developments, and applies equally to diagnosis. The same imaging techniques used in the operating theatre can also be deployed in the GP's surgery, to avoid the need for biopsies for example. There are common IT challenges, which essentially revolve around algorithms for navigation and identification of tissue, along with high speed networks exhibiting minimal latency and jitter, given that medical images generate large amounts of data that must be transported reliably, especially in real-time surgical applications as in the tumour removal, or the mitral heart valve repair. Best effort may be good enough for the surgeon, but not for the underlying computer network.