---

5 Conclusions:

recommendations for future directions at Umeå

June 1998

Work here in Umeå should be chosen according to 3 criteria:

• 1) existing strengths on which research can capitalise,
• 2) a strong local need,
• 3) not an already over-researched topic from which products are already available

As strengths have often been developed to reflect local needs, the first two tend to go together.

It is also important to bring together strengths in both medical and information technology.

Some relevant local skill areas in medicine are:

• Surgery and Surgical Education
• Dentistry and Odontology
• Radiology
• Physical and Mental Rehabilitation (including Geriatrics and Clinical Psychology)

These skills are strongly represented in the departments of Surgery, Dentistry, Radiology, Applied Psychology, Geriatircs, amongst others.

And in information technology:

• Advanced Computer Graphics and Visualisation
• Human-Computer Interaction
• Distance Learning and Telemedicine

Located in the departments of Computer Science, Informatics, Pedagogy, Psychology, and so on.

There is a great deal of research and development going on around the world in the application of virtual reality to medicine. The hard problem in surgery, that has not been completely solved anywhere, is to provide fine detail and rapid interactivity. This is particularly difficult with soft tissue that moves - e.g. the abdominal region - where most surgical intervention is focused. The brain and spinal chord, although more complex and therefore requiring very fine detail in visualisation, do not move much when prodded. The problem there is in some ways more tractable.

Some areas - such as endoscopic simulators for training surgeons - are receiving enormous attention, largely because the mechanical interaction problems are tractable and companies see profitable products emerging already or very soon, We should not duplicate the mechanical side of these trainers, However, they suffer from poor realism in the visualisations, and in the simulation of behaviour when interacting with the visualised structures. Although there are many endoscopic simulators with realistic feel (since real endoscopes give only crude tactile feedback), there are none that look realistic, especially when the surgeon interacts with the visualised organs and other tissues. For example, so-called "wet-look" graphics only look wet until they are prodded and moved about; then they appear to be covered with a film of plastic. The big weakness is in simulating the behaviour and appearance of soft tissues of different densities during interaction.

Other areas - such as remote robotic surgery - are highly specialised and would not build on a particularly strong local skill base. It is probably also unwise to focus on such topics.

Excluded from consideration in this survey are applications in health education for the general public (largely covered by on-line or CDROM-based multimedia presentations and not addressed by VR), visualization of large-scale medical databases (i.e. medical records from large numbers of patients - although VR is being applied in this area), and the application of VR to the archetectural design of medical centres.

The VR-Lab provides excellent faciltites for work in Medical VR, particularly in the area of advanced 3D graphics and rapid interactivity.

Given these biases, I would tentatively recommend that research work at Umeå should focus on the following 3 broad areas:

• 1. Mental and physical diagnosis and skills rehabilitation
• 2. Complex medical data visualisation for diagnosis and surgery planning, including telemedical (remote diagnostics) applications.
• 3. Integrated simulators for medical training, especially in relation to 2.

All three share a need for realistic, precise and rapid interaction with complex, three-dimensional realisations of data. This focus plays to Umeå's main strengths. Less suitable are applications that depend on mechanical simulation. But such devices will soon be available in the market and should be amenable to integration in, for example, trainers where the local content is mostly comprised of the 3D realisations and interaction strategies. This also implies the use of high-end, and expensive, graphics computers, rather than the cheaper systems - such as PC versions of medical atlases - intended for mass distribution.

For mental and physical rehabilitation, the virtual worlds created must be believable and realistic, combining detail and fast response times. For visualising complex anatomical data, for diagnosis and surgery planning, as well as training in such operations, the needs are the same.

There should also be a strong focus on user involvement and trials, from the earliest stages of needs analysis and research project planning. No department can succeed alone in these areas, which are by their nature multidisciplinary.

Perhaps the two main weaknesses of existing applications of VR to Medicine are the relatively poor realism, and weak usability. Both of these issues could be addressed using local skills. The main research issue in Medical VR is to convey sufficiently finely-detailed 3D structures with very fast interactivity. The combination of these two factors, which tend to trade off each other, is the key to using VR in medicine for more than fairly crude simulations of relatively simple and routine activities.