IMAGE GUIDANCE – a personal history
Image guidance, frameless stereotaxy or neuronavigation has been an essential part of my neurosurgical consultant practice. Indeed the introduction of the technology to neurosurgery and its development will probably be my main legacy to the profession. Like my predecessor Brian Cummins who was one of the first neurosurgeons to apply CT scanning to general neurosurgery giving him an experience for which he became world renowned, I was as fortunate to be in the right place at the right time both to recognize the potential of image guidance and to be able to develop its use without the restrictions imposed by the US regulatory body – the FDA. Ever since I have been at the forefront of developments in the technology, even referred to as an ‘opinion former’ in the field at one stage.
This essay is a personal history of image guidance and its application mainly in neuro-oncology and epilepsy surgery, my fields of expertise. It is not comprehensive as much of it is based on memory – I regret not keeping a diary or contributing to some of the early publications written on the subject – Bob Macuinas being an editor of early books on the subject in the ‘90’s. Apart from general interest, my main purpose is to illustrate the limitations of the technology with particular reference to epilepsy surgery but also to contribute to the discussion about the way forward in neurosurgery. The ‘fashion’ is for major capital expenditure on real time imaging – intra-operative MRI. At an international meeting I attended recently this technology was being out forward as the standard of care for the treatment of gliomas. (albeit an opinion expressed by an American neuroradiologist with a specific intra-operative MRI system to sell) I have real reservations about this, which I will discuss here.
For historical reasons epilepsy surgery is still not recognized as being ‘main stream’ neurosurgery. Traditionally epilepsy surgeons were surgeons with a greater interest in neurology than surgery and often practiced surgery separately from their colleagues. Aneurysm and skull base surgery were seen as the ‘difficult’ areas of surgery. Since the advent of endovascular treatment of aneuryms, intrinsic tumour surgery has taken a greater prominence as the discipline that defines neurosurgical expertise. But the peculiar difficulty of operating within brain tissue that looks macroscopically normal with little margin for error when resecting an epilepsy focus, was never fully appreciated by the profession in general and in many respects still isn’t. Yet as the application of image guidance has evolved it has been within epilepsy surgery that the limitations of the technology have been fully exposed. A full understanding of this is required if one is to appreciate how the future will unfold and how best to invest limited resources to take the profession forward.
Setting the scene.
I was appointed a locum consultant in my alma mater Hope Hospital Salford in December 1990, part of the Manchester senior registrar rotation. At that time ‘image guidance’ was limited to frame based stereotaxy, in our case using the Leksell frame. We treated Parkinson’s patients by lesioning the thalamus based on Schaltenbrand atlas calculation of coordinates from the AC-PC line defined on a CT scan and in a few cases we used the stereotactic frame to carry out stereotactic biopsies of deep seated lesions. Using the basic principle of stereotaxy, namely that any point in stereotactic space can be defined in terms of its x,y,z coordinates from frame centre, a single point for biopsy could be defined.
This activity represented only 3% of intracranial practice. In general the brain was approached via standard exposures using anatomical cues for orientation. Much emphasis was given to ‘experience’ – senior surgeons have the distinct advantage over their junior colleagues by the simple fact that they had ‘been there’ more often. Where anatomical cues were not forthcoming such as when trying to locate an intrinsic tumour within the brain, surgery became almost ritualistic. A standard approach was made exposing a large amount of brain. I remember the application of multiple curved artery forceps tied together with elastic bands and small wet towels clipped to the skin edges to prevent the diathermy igniting the anaesthetic gases! Much pontification would go on about which of the exposed gyri looked the most swollen. Multiple biopsies would then be taken for frozen section till the brain surface looked like a pin cushion. With luck, eventually a positive diagnosis would be obtained, there would be a collective sigh of relief and the registrar – me, would be left to close up the large would taking care to appropriate number of dural hitch sutures to prevent extradural haematoma. All in all this was a 4-5 hour procedure that achieved precious little! It was no wonder that the Liverpool policy of treating all malignant gliomas with burr hole biopsy, steroids and radiotherapy had as good results as resection at the time
On April fools day 1991, I started work as consultant neurosurgeon at the Frenchay Hospital, Bristol, taking over from Michael Torrens and joining the illustrious team of Huw Griffith, Brian Cummins, Hugh Coakham and Richard Nelson. I was appointed as the department’s ‘stereotactic surgeon’ despite that fact that at the time I little time for stereotactic surgery! “Why spend all morning doing a burrhole when one could be clipping aneurysms or taking out meningiomas”, I thought.
Luckily for me my protestations about the limitations of frame based stereotaxy did not fall on deaf ears as Huw Griffith had the same reservations as I. Between us we defined the requirement for a universal image guidance system. The ideal system would be frameless – allowing access to the skull base and posterior fossa; it should allow registration to the whole image dataset not just a single point; it should be possible to use pre-operative diagnostic images rather than scans done at the time of surgery (our radiology colleagues did not give up time on the CT scanner willingly) and it should be possible to use a patient’s own images for reference rather than refer to the compromise of an atlas.
With these ideas in mind we looked around at what technology was available. Watanabe in Japan had published on the use of a mechanical arm to replace the stereotactic frame but his system still registered to a single point in stereotactic space. In 1991 Huw and I took a day trip to Milan – unheard of in those days, to see a system that did register to the whole dataset of a scan but which was still fixed to a frame. We read about but did not follow up a German ENT system that, in retrospect did all the things we were looking for except for potential inaccuracies of intracranial application. There were other prototypes around as well, for example an ultrasound registration system for the Electa ‘Surgiscope’ a complex and very expensive ceiling mounted microscope, but none, which met all the requirements that Huw and I had defined.
Things changed when Huw came back from the AANS meeting in New Orleans in 1991, saying that he had seen a prototype of exactly what we were looking for – the ISG Viewing Wand. ISG were a Canadian company producing radiology review stations using new parallel processing computer technology. At the time sophisticate image processing was extremely limited – CT scan software could be programmed to create a single static 3D reconstruction from a dataset but this would take all night and could not be interrogated in any meaningful way. The ISG platform, on the other hand could produce a 3D reconstruction in seconds and as important allow real time reformat of images to allow their immediate detailed interrogation. Someone within the company had had the idea to attach the radiology platform to a six degrees of freedom mechanical arm called the Viewing Wand. I have always given Bruce Leggate the credit for writing the original controller software for this on his BBC micro computer. This ensured that at any one time the computer ‘knew’ where the tip of the wand was in space. All that then was required was to register the actual position of the patient’s head to the virtual image of the 3D reconstruction such that wherever the wand was moved to around and inside the head, its position relative to the preoperative image could be seen on the screen.
My introduction to the system came in February 1992, when I broke a trip to the western seaboard of Canada to go skiing to visit Jim Drake in Toronto. I watched him do an image guided 3rd ventriculostomy on a boy with congenital hydrocephalus, distorted ventricular anatomy and a thickened 3rd ventricle floor such that there were no anatomical cues as to where the basilar artery was. Without the technology I would not have even attempted the procedure. Jim, on the other hand could visualise a safe entry point for the balloon dilatation of the ventricular floor with ease. It was a ’eureka’ moment for me! The only problem appeared to be the company’s reluctance to develop the technology in Europe. They were then not allowed to market the technology in the USA and Canada as they could only work with beta development sites there. They told me they had only one technician in Europe. On enquiring where he was based I was told – “Bath, England”. It was clear that no one till then appreciated that the European ISG technician lived closer to Frenchay than the Hamilton, Ontario based technicians did to the Toronto children’s hospital! The deal to send a system to the UK was struck there and then. No market price for the technology had been defined as we were the first unit world wide that had offered to buy one! We agreed £100,000 – 10% down and rest paid after one year. A Leksell frame was £40,000 at the time and I estimated that we would get at least three times the use out of the Viewing Wand. I like to think that that decision helped to define the price of the new technology. Top of the range microscopes were selling for £0.5m at the time. There might otherwise have been an attempt to charge that sort of price for the technology which would have severely limited its uptake. Even today – a quarter of a century later £250k buys you the latest technology.
In June 1992 we took delivery of the first system in the world to be used in full clinical practice. The ‘magic wand appeal’ was set up to raise the funds to pay for the system, so named by my five year old daughter after I had tried to explain what the technology did. The first case we performed was a image guided temporal trephine and resection of a medial temporal glioblastoma, which led to its immediate uptake in 30% of all intracranial procedures. The economics were simple – bed stay for intrinsic tumour patients was halved from a mean of eight days to four and operation times significantly reduced also. As one senior neuro-anasethetist put in. “For the first time my neurosurgical colleagues seem to know where they are in the brain rather than pretending they do! It is the only new technology that I have seen them use that I would raise money for myself” (Peter Simpson)
In 1994 we published a series of 350 cases in Acta Neurochirurgica. (Sadly Huw Griffith died before the publication of this paper. It was the measure of the man that although instrumental in defining the need for image guidance, he elected to let the young man in his team run with the technology – a lesson in team building that at times we need to re-learn.) This paper led to the FDA licencing the technology in the USA and its subsequent rapid expansion. At the EANS meeting in Berlin in 1995, I counted a total of 15 different systems vying for market share. Radionics / brainlab / surgicscope
My recollection of the early days of neuronavigation are as follows, although I accept that this may not be entirely accurate. At the same time that the ISG wand was being developed a separate system had been devised independently primarily for spinal application. This was eventually marketed by Surgical Navigation Technologies – SNT. Their system relied on infrared localisation technology, which was to replace the mechanical arm systems. The ISG system was bought out by Electa, who had recognized the threat to their dominant position in the stereotactic frame market. However they did not have the technical expertise to manage the new technology. Indeed many of the key individuals at ISG went to work for SNT. The survivors in this highly competitive market were SNT, eventually bought out by Medtronic and Brainlab, a highly versatile company that, in the early days did not innovate themselves but had the technical infrastructure to copy other people’s ideas very effectively. Medtronic suffered the same problem as Electa buying a product to control its use but without the expertise within their organisation to develop it. Within a few years these two companies became the major players in the field with their competing commercial interests dictating both the positive and negative developments in the field.
Registration – the key to successful neuronavigation
The genious of image guidance was its simplicity. The early systems did no more than relate the surgeon’s position during a procedure to the pre-operative images. Current systems do no more than that either. At the time however the ability to plan a procedure without reference to anatomical landmarks was a revolution allowing for a simple minimally invasive approach to intracranial neurosurgery for the first time. Laser pointers attached to stereotactic frames were being marketed at the same time as image guidance took off but these were soon superceded. The key to successful image guidance was the registration process.
The gold standard for registration was the accuracy produced by a stereotactic frame – rigid fixation of a solid frame to the skull by pins embedded in the skull itself. Initially bony fiducials were applied to the skull before the acquisition of a preoperative CT Scan. The obvious disadvantage of this was the need for a dedicated scan for the surgery acquired at the time of surgery. It was immediately apparent to us that the accuracy of the 3D reconstruction of the 2D dataset was such that anatomical landmarks could easily be identified that could act as fixed fiducials. We used structures like the inner and outer canthi of the eyes, the bridge of the nose and the junction of the pinna with the scalp. The early digitisers then allowed for the application of 30 - 40 points on the scalp to be added to the registration. The software used a direct transformation algorithm to analyse the expected position of the point with the actual position thereby producing a ‘hat on head’ fit of the surface of the scalp. The software would only allow the procedure to continue of a sufficient number of points were applied over the complete circumference of the scalp. The surgeon was presented with a quantitative measurement of the accuracy of registration – a root mean square (RMS) value, usually +/- 10mm for anatomical landmarks alone and +/- 2-3mm once the random surface algorithm had been completed. The earliest version of the ISG software allowed the surgeon to choose the degree of accuracy that was required for a given procedure. Later versions would only allow the system to be used if a RMS value of <2mm was achieved. We developed a process still used today that involved checking each registration against recognisable anatomical landmarks and a quadrant check of registration to the skin in all four quadrants of the scalp making sure that there was no asymmetrical error brought about errors in scan acquisition. Once done the surgeon could be sure that having registered to the whole circumference of the head any error that occurred from working within the volume so defined would be less than then errors detected on the surface – at least until the head was opened!
The concepts of mechanical and application accuracy had already been defined by frame based stereotaxy. The mechanical accuracy of a system was the degree of precision that could be achieved in a laboratory setting – often referred to at the time as ‘submillimetre accuracy’ The application accuracy allowed for all the variables that could be introduced during the image acquisition and registration process. A target application accuracy of +/- 2mm was readily achievable with both frame based and frameless systems. What image guidance did was to introduce a third concept with respect to registration accuracy. We called this operational accuracy – the accuracy or to be more precise the inaccuracy introduced by soft tissue movement on opening the head – brain shift. It is this that has always limited the accuracy that can be achieved with image guidance and has dictated its usefulness over the years.
System development over the years
Since the turn of the century there have only been two main players in the world of image guidance – Medtronic and Brainlab. Initially both competed for market share by trying to improve the user interface of their systems. Improvement on digitiser technology allowed for more rapid analysis of more random surface registration points. Both companies developed laser pointers that scanned facial features, automatically inputting registration points. The Brain lab system persisted. The Medtronic system was replaced with the same surface matching algorithm that had been developed by SNT for spinal registration (I take the credit for this development). Yet in order to streamline the registration process, both companies marketed systems that did not insist on circumferential registration. The consequence was that routine registration only took place to part of the volume dataset. As SNT had established with spinal registration any errors in the registration process were accentuated when working outside the registration volume. The consequence of this is that rather than increasing in accuracy with the development of new software the registration process has actually become less accurate over the years, an important fact to recognize as the applications for image guidance have become more sophisticated. There was also no way to check the level of accuracy. It is relatively common experience that with both systems registrations can be found to be innaccurate for no apparent reason - this is the likely cause.
The new companies took development in different directions. Medtronic introduced an electromagnetic system of communication between the image guidance device and the computer. The main advantage of this was the ability to do away with the 3 pin headrest making the system more used friendly in children
Brainlab took a different approach. They recognized the increasing requirement for sophisticated image processing and the difficulty that every unit has in bringing all of a patient’s imaging onto a single platform. Cleverly they have made sure that their image guidance platform not only could bring all of a patient’s imaging together from the hospital PACS system but it will also allow specific sophisticated image processing within the Brainlab software such as tractography. Multimodality image fusion is an integral part of the platform. They have also worked hard on developing the interface with intraoperative MRI and their radiosurgical systems
In this sense they have become the lead provider of current image guidance systems. The only equivalent Medtronic initiative has been an attempt to develop a comprehensive platform for epilepsy surgery bringing together all the imaging and investigation modalities required for building the appropriate hypothesis for epilepsy surgery. It is too early to say how effective this will be although, even if successful, it cannot have the more universal impact that the Brainlab imaging infrastructure is having.
Applications of image guidance
The application of image guidance in neurosurgery has to be seen in the light of the development of other treatment modalities. The first immediate application was as a ‘flap planner’. Minimally invasive approaches to lesions that avoided big skin and bone flaps became the norm for tumour surgery. Whereas before a large craniotomy had be opened to allow orientation using anatomical landmarks, now a small trephine sufficed, the image guidance being used to plan the approach to a tumour. Following the right trajectory through the brain, the tumour could be identified, biopsied and resected as much as macroscopic differentiation between tumour and normal brain allowed. Standard approaches were still used for skull base tumours but image guidance was useful in identifying important structures in and surrounding the tumour and for assessing the degree of resection as one went along.
The system had applications in other areas of intracranial surgery too. Planning trajectories to small ventricles in shunt surgery, third ventriculostomy and planning trans-sphenoidal surgery were examples. Aneurysm surgery rarely required image guidance partly because endovascular treatment became the treatment of choice. However while there was still a need for surgical treatment of aneurysms image guidance was one way of planning infrequent surgical approaches such as approaches to the basilar artery.
The first spinal operation we performed used intracranial registration techniques to plan an approach to the odontoid peg in a resection of rheumatoid arthritis. Ciaran Bolger’s appointment to the department heralded the major development in spinal image guidance, using the system to plan pedicle screw fixation particularly and, in conjunction with real time imaging planning procedures requiring complex orientation such as the insertion of C1/C2 screws.
I started the epilepsy surgery program at Frenchay at the same time as we started using the Viewing Wand. Indeed our first case was a trans-gyral resection of a temporal glioma. This was so successful that there seemed to be no reason not to use the same approach for medial temporal lobe resection in epilepsy. Approaching the temporal lobe via a 2cm diameter trephine, I used image guidance to identify the inferior horn of the lateral ventricle. Once the Pes Hippocampus was identified, its en bloc resection for histology could be achieved without difficulty. Like Olivier in Montreal this was all the resection I did for temporal lobe epilepsy in the early days with apparently as good early results as more extensive resection. However I had two cases of delayed post operative SUDEP, one after a period of 5 years fit free. These persuaded me to return to a more extensive temporal lobe resection using a modified Spencer transcortical technique. Image guidance was used to plan the trajectory to the inferior horn of the lateral ventricle and to determine the posterior limit of the resection of the hippocampal tail.
Brain shift – the limiting factor in image guidance
The factor that limited the usefulness of image guidance in the early days was and still is brain shift. Once the head has been opened and CSF fluid drained the soft tissues in the head move. The degree of movement is variable and is dependant on many factors. Some can be compensated for like the effect of gravity – positioning the patient so that the gravitational effect of brain shift occurs in only one plane for example. Efforts can be made to mark the limits of a tumour prior to opening the dura by placing markers in the brain through small dural punctures – so called ‘fence posting’. Much academic effort has gone into trying to model brain compliance and predict brain shift but in my experience this is futile. The compliance of the brain and its susceptibility to shift is very variable not only between patients but in the same patient with time. For example one can do a wide resection of a glioma that one would expect to cause shift only to find that the turgor within the brain surrounding the tumour is resistant to shift. In another similar case registration is not maintained at all. It is true that the deeper one resects the less brain shift is an issue but at depth the margins of error are less.
Some have advocated using ultrasound, fusing the peroperative ultrasound images with the preoperative images reformatted in the same plane. I never fully understood why this was not adopted more universally as the fused images demonstrated by the system’s advocates always looked very good but in our hands our trial of a system did not convey enough of an advantage to warrant the additional cost.
Nevertheless a solution to the problem of brain shift is thought to be real time imaging – if not ultrasound then MRI or CT. The ‘O’ arm intraoperative CT marketed by Medtronic is one solution. Potential MRI solutions have been considered for many years starting with the GE doughnut in the 1990’s. Small bespoke systems such as the Medtronic Polestar may have had specific applications such a suprasellar pituitary tumours but specific applications like this alone cannot justify the $1m price tag. What we now have access to is a 1.5T MRI adjacent to a theatre where we can move a patients from theatre through to the MRI scan. This process is complex and not without risk and is not actually ‘real time’ imaging. Nevertheless the system has its advocates particularly for the surgery of low grade tumours. The latest intraoperative MRI systems move the scanner into the operating room rather than move the patient out which might be an improvement in logistical terms but I still ask the question as to what is the benefit of intraoperative MRI in the first place?
Take this example. A patient in her 60’s presenting with rapid mental deterioration over a month with a CT and MRI showing an extensive dominant temporal intrinsic tumour. The treatment options offered to her by different surgeons were – do nothing, do a biopsy, do a partial resection or do a radical resection. I offered the latter and achieved a subtotal resection as defined by early postop MRI. The patient had no immediate postoperative deficit. The histology was glioblastoma. I had an in depth discussion with the patient and her family about further treatment. We decided to complete the resection – the procedure taking place 5 days later. Compromise to the blood supply to the residual tumour at the first operation made identification and resection of the remnant quite straight forward easier than it might have been had the resection been attempted after an intra-operative MRI. Early radiotherapy starting within a month of the second operation followed. She is now a five year survivor. I argue strongly that intra-operative MRI would not have conveyed any advantage on her case and indeed may have been disadvantageous.
The solution for intrinsic tumour surgery is to recognise the limitations of image guidance – realistically without macroscopic pathological cues to guide one the systems cannot be trusted to an accuracy of greater than =/- 10mm and make use of other techniques to maintain safety during a tumour resection – awake craniotomy in dominant speech areas and peroperative neurophysiology to monitor somatosensory and motor activity. As my current senior registrar, Angelo Pichierri has shown me, careful image processing can generate useful tractography images that are really helpful in surgical planning
Image guidance – the future is robotic
The only other way to avoid the operational inaccuracy of brain shift is not to open the head in the first place! Therefore for all applications of intracranial neurosurgery that require genuine ‘millimetre’ accuracy ways have to found to operate through a twist drill or not open the head at all. This is the approach that is taken in our department in functional neurosurgery and the one we take in epilepsy surgery. It is to fulfil this technical requirement that robotics now has a place in neurosurgery.
There used to be an old adage that I first heard applied to laser technology – “if you don’t need technology, don’t use it!” Robots have been around a long time on the fringes of neurosurgical practice – at least as long as image guidance. I remember the Lausanne neurosurgeons carrying out a robotic burr hole biopsy in the early 1990’s. At that time it was technology looking for an application. It has only been in recent years that that the need to hit targets within the brain with an accuracy of +/- 1mm that the need for the precision of robotics has been identified. The two drivers for robotic technology in our department have been functional neurosurgery and epilepsy surgery. In functional neurosurgery the driver has been the need to place electrodes immediately adjacent to the subthalamic nucleus. Here the technical challenge has been to define the target, initially by painstaking manual definition of the margins of the nucleus but more recently by an automatic reconstruction algorithm using an atlas based reconstruction as a template (What goes around comes around!) A trajectory to the target is then defined, allowing a surgical approach via a twist drill, the stimulator electrode being after loaded down a guide catheter. This is work pioneered by Professor Steven Gill.
In epilepsy surgery the need for robotic precision has been driven by the clinical need for a better solution to intraoperative recording than subdural grids, which carried a significant and unacceptable risk and which only provided a two dimensional solution to a three dimensional problem. The first use of robotics in epilepsy surgery was therefore to place a predefined array deep brain electrodes to carry out stereo electrocorticography (SEEG). This has been an essential part of the investigation of complex temporal and extra temporal epilepsy. Based on the semiological hypothesis suggested by the epileptologists the recordings obtained using the technique allows for the identification of the origin and spread of focal seizures with a precision that was never possible before. Most importantly it is helping to define the concept of network epilepsy, in those patients where there is no focus.
The second use for robotics in epilepsy surgery has been to define a marker for resection of cortex where the electrophysiological onset of a seizure originates but which is anatomically normal. Without pathology that is macroscopically different in appearance to the normal brain there is no way to compensate for subtle degrees of brain shift brought on by opening the head. The reproducibility of robotics allows for catheters to be placed along exactly the same tract as the recording electrodes. Accurate resection of the brain surrounding the marker catheter is then straightforward.
Recently we have extended this principle to the placement of an electrode to carry out interstitial hyperthermia in the posterior insula gyrus of a man with intractable seizures originating from this area. In this case we after loaded the electrode down a catheter using the same approach as for functional neurosurgery. My thanks to my neuromodulation colleagues Neil Barua and Nik Patel for providing the expertise to achieve this.
It is interesting to me that a science fiction writer – Azimov should define the laws of robotic / human interaction, the first being that a human should not come within the sphere of action of a robot. In neurosurgery to date that is how we have used the robot – as a tool to define the trajectory to a target accurately. It should be straightforward to use robotic technology to do more – drill the twist drill, countersink the guided catheter etc but so far we have not taken that step, perhaps because of the cost of developing the necessary haptic feedback mechanisms to make active robotics possible. Other surgical disciplines notably urology have used robotics as an intelligent micromanipulator. While there has been some investment on the development of similar technology in neurosurgery, the need for it is limited in my view. A particular ‘red herring’ has been the enormously expensive development of robotic technology that can function within the magnetic field of an MRI scanner. The theory is that a surgeon can operate a non magnetic manipulator remotely while the patient is within the scanner – real time imaging? To what end I ask myself and to what degree of accuracy?
This leads to the question of which robot one should choose. It is easy to exclude some on the basis of cost – the MRI compatible one mentioned above for example. The two main players on the UK market are the Renishaw robot and the Rosa Robot, the latter having been developed by the team that developed the original Neuromate robot that became the Renishaw machine. In hardware terms many of the features of the Neuromate robot have been improved on in the Rosa machine – it moves quicker and can be used in passive mode for example. However for us the decision to choose the Renishaw system was absolutely straightforward, applying the lessons we learned in the very early days of image guidance, namely that with emerging technology it is essential to be able to work directly with the developers of the technology, just as we did with ISG all those years ago. It is only by direct communication with the technical team that clinical issues relating to the use of the technology can be addressed promptly. We have that direct working relationship with Renishaw which is resulting in the rapid evolution of the NeuroInspire software with a rapid expansion in the use of robotics in our department.
The element of robotics that we are only just beginning to appreciate is the power of ‘big data’ computing that backs up the technology. Take the example of subthalamic stimulation for Parkinsons’ disease. Not only does the technology provide the platform for planning and placement of the electrodes. It also stores all the data on previous cases enabling a calculation of optimal stimulation parameters of the nucleus for a given symptom complex. The potential of this kind of analysis in epilepsy surgery is huge. The ability to compare and contrast epilepsy onset and spread in different cases of SEEG recording will advance our understanding of the disease. To make the most of this we have to standardise date recording and share data. This requires creating an open platform for collaboration with units internationally. Overcoming commercial and intellectual property issues to allow this to happen is the major hurdle. I use the expression 'what goes around comes around' often. I credit the first expression of this concept to Jean Taliaraich and his idea of the Mophogramme, developed in the 1960's!
Convexion Enhanced Drug Delivery – CEDD
When I lecture on the future of neurosurgery, I make the rash statement that in 5 – 10 years time the commonest operation in neurosurgery will be the robotic implantation of micro catheters for CEDD. By controlling infusion rates through multiple microcatheters it is now possible to obtain a uniform distribution of large molecular weight substances within the blood brain barrier with minimal concentrations of infusion outside the brain. As substances suitable for neuromodulation are developed I see the real prospect of some of the major neurological disease currently not amenable to surgery becoming ‘surgical conditions’. As soon as trials to treat neurodegenerative conditions like Huntingdon’s have shown to be successful the door will be open for other conditions such as Alzeimer’s. If the infusion of growth factors can reverse stroke then another major group of neurological patients will come under the care of neurosurgeons. Current methods of treating intrinsic malignant tumours are ineffective yet early work suggests that CEDD of conventional chemotherapy agents into a tumour bed or an inoperable tumour is useful palliation. As soon as products are developed that can block mitosis, (viral agents?) infusion of these into a tumour bed may have the real prospect for effective palliation for gliomas. Maybe infusion of substances like Botox that will prove to be a better treatment for epilepsy than resection or oblation. What is sure is that the development of the discipline in the next quarter of a century is likely to be as dynamic as the 25 years that I have been part of. I'm proud to say that some the future great names in this field - Nik Patel, Neil Barua, Angelo Pichierri to name three all spent some time training with me.