T. M. O'Brien, P. Murray & J. O' Byrne
Department of Orthopaedic Surgery, Royal College of Surgeons in Ireland Cappagh Orthopaedic Hospital, Dublin

G. Chamberlin, K. Cameron & P. Cochrane
BT Research Centre, Martlesham Heath, Ipswich

Summary
We present our experience in developing a system for transmitting 3D imaging of surgery from an operating theatre to a distant location. The surgeon wears head mounted cameras which transmit the 3D image through ISDN lines to a distant location where the 3D image is reformed by the use of infra-red controlled special glasses. We tested the system in an experimental situation and then in the operating theatre. The system offers the potential for improved remote training and monitoring of surgery and will have significant implications for the future training and practice of surgery.

Introduction
Intra-operative videotaping and 2-Dimensional conferencing of surgical techniques have become popular as teaching tools [1, 2]. However the flat-natured presentation of what are often complex three-dimensional procedures makes them a poor substitute for hands-on training. Experience suggests that while these techniques provide a good record of the operative procedure they do not give the feeling of depth required to develop a good surgical technique [3]. This lack of the third dimension is also well appreciated by those who carry out minimally invasive surgery where the lack of depth of the image seems to cause increased morbidity and mortality (BBC Panorama). Recent advances in endoscopic technology can now provide a three-dimensional image in such procedures which will hopefully overcome these difficulties [4]. In an attempt to provide a feeling of depth in video-conferencing of open surgical procedures we set out to develop a system for the transfer of a three-dimensional image from an operating theatre to a remote site using an ISDN telephone connection. The object was to provide real-time three-dimensional images from the operating theatre to another location where the image could be seen by a student or instructor. In devising this system many decisions had to be made regarding the placement of cameras, the control of cameras and an assessment of the quality of the three dimensional picture provided live from the operating area.

Methods and Materials
The 3D imaging system consisted of a pair of video cameras connected to a 3D display. The display shows alternate left and right views at a frame rate of 100Hz. The observer wears a pair of glasses containing liquid crystal shutters. When the left view is shown on the display, the left shutter is opened and the right closed, conversely for right eye view. An infrared clock signal is used to maintain synchronicity between the display and the glasses. The high frame rate and high contrast ratio of the shutters ensures that the observer sees a clear, flicker free three dimensional image.

To enable comfortable viewing of a three dimensional image it is necessary that the recording and replay geometry be accurately matched to the human visual system. The system used in this experiment was built around the desire to replay the image on a 17" monitor, with a viewing distance of 0.75 metres. Accordingly a pair of 1/3 inch diameter video cameras fitted with 6 mm focal length lenses were mounted on a base with a spacing of 40 mm. An adjustment system enabled them to be aligned optimally to view an object positioned at arms length. The complete camera assembly was itself mounted on a headband system, positioned in the centre of the wearer's forehead.

To provide the data link, a pair of ISDN (Integrated Services Digital Network) codecs were used in loop-back mode with one codec dedicated to each of the cameras. This allowed various data rates to be evaluated during the course of the trial. Audio communication was provided using a loudspeaker and microphone at each end. Careful positioning removed the need for more sophisticated echo-cancellation techniques.

Initially a simulated state was designed to replicate an operative procedure. A freshly sacrificed animal model was used and the operative procedure was carried out in one room while being directed from an adjacent room with no direct visual or verbal contact except through the ISDN line. A junior hospital doctor carried out an operative procedure under instruction from his senior colleague in an adjacent room.

At a later date we tested the system in an operating theatre with two patients undergoing anterior cruciate ligament reconstruction. Each patient gave their consent to the operative procedure and the conferencing. A Senior Surgical Registrar (operating surgeon) carried out the surgical procedures under direction by the Consultant Surgeon (surrogate surgeon) who was located in a room adjacent to the operating theatre. The Senior Registrar was familiar with the surgical technique and had frequently carried out the procedure both under direction and independently. The purpose of this trial was to assess the view of the surgical procedure being provided in the distant location and not its value as a training tool in this first instance. However it was also hoped that the contact with the senior colleague who was more experienced with the operative technique might help the surgical technique of his junior partner.

Results
The system was tested at various data rates between ISDN 2 (2 X 64 kbit/s) and ISDN 30 (30 X 64 kbit/s). All images were transmitted at CIF resolution (352 by 288 pixels), the highest video resolution supported currently by the telecommunication standards. During the course of the trials, the transmission rates were changed without the knowledge of the surgeons participating. When using tripod mounted cameras, rates as low as ISDN 6 per channel were considered acceptable. On moving to the head mounted cameras, the increased amount of image motion rendered this rate unacceptable. ISDN 12 was felt to be very acceptable and no significant benefits were experienced on moving to the higher data rates of ISDN 30.

In the simulated surgical situation the initial use of tripod mounted cameras was found to be much inferior to the head mounted camera. Among the problems encountered were frequent obstruction of the field of view by the operating surgeon's hands and problems with shadowing across the field of view. Despite the benefit of a 3D view in the distant location the tripod mounted set-up failed to create the unity of vision subsequently achieved by cameras mounted on a headband worn by the operating surgeon. Head mounted cameras on the operating surgeon conveyed a much better feeling of depth to the surrogate surgeon. They also ensured that both surgeons were focused on the same field of view. Initially the head mounted cameras were fixed on a headband and could not be controlled from the adjacent room. It was found that some degree of individual control by the surrogate surgeon was required for situations where the operating surgeon approaches and moves back from the operative site. The effect of movement of the operating surgeon's head besides was not regarded as a significant problem.

In the true surgical setting a complex approach to the knee joint was carried out under full view of the surrogate surgeon in the adjacent room. During the operative procedure the surrogate surgeon was able to direct the precise movement of surgical instruments by the operating surgeon and could advise on the best surgical technique throughout. Although this surgical procedure was carried out through a rather restricted incision the view of the operative field transmitted to the surrogate surgeon was most satisfactory. The surrogate surgeon found the 3D view realistic and much superior to the view available to an observer in theatre. The only difficulty again encountered was the tendency for the centre of attention of the remote picture to change when the operating surgeon moved his head closer or further away from the operative field. Both surgeons felt very comfortable with the relationship throughout the operative procedure, in particular the junior surgeon felt that he had more operative freedom than would be provided by the actual presence of the surrogate surgeon within the operating area. The surrogate surgeon in the peripheral location also felt that he was not interfering unnecessarily with the operative procedure being carried out by his junior colleague yet he had a high degree of control over the procedure. Overall the operative procedure using the surrogate three dimensional video-imaging was regarded as a most useful technique in both observing and supervising the operation.

Discussion
It is difficult to assess the benefits of a new technique when it is as subjective as outlined above. However we compiled this report to publicise the feasibility of three-dimensional conferencing of surgical procedures as we felt this development had significant implications for the future. The improved monitoring of surgery with this method combined with the capacity to learn three-dimensional procedures without the need to be present in a particular operating theatre has the potential to revolutionise the development of surgical training and audit. While we have demonstrated that it is possible to transfer a true 3D image from a theatre to a remote location using the ISDN network the implications of this development have not been fully analysed. However the technique does offer the possibility of remote training of surgical procedures without the need to travel to view the technique being performed in person. Furthermore surgery can be monitored from a remote location in a meaningful educational manner. Finally audit of surgery may be possible without depending on morbidity and mortality figures which at best are only crude indicators of surgical skill and technique. There are many other implications of this development to be addressed, none more important than the medico-legal responsibilities of the surgical partners. However we have demonstrated that the marriage of modern Telecomms and 3D imaging is likely to enhance the training and monitoring of surgical skills to the benefit of future educators and patients alike. References

1. Coosemans W, Leroot T E, Van Raemoonck D E. Thoracic surgery: the Belgian experience. Annals of Thoracic Surgery. 1993, 56 (3): 721-30.
2. May M, Korzec K R, Mester SJ. Video telescopic sinus surgery techniques for teaching. Transactions - Pennsylvania Academy of Ophthalmology & Otolaryngology. 1990, 42: 1037-9.
3. Smeak D D, Beck M L, Shaffer C A, Gregg CG. Evaluation of video tape and a simulator for instruction of basic surgical skills. Veterinary Surgery. 1991, 20(1): 30-6.
4. Wenzl R, Lehner R, Vry U, Pateisky N, Sevelda P, Husslein P. Three-dimensional video-endoscopy: clinical use in gynaecological laparoscopy. The Lancet. 1994, 244: December 10; 1621-2