The field of computer vision focuses on the development and implementation of algorithms which allow computers to "understand" image and video data at various levels depending on the task at hand. Task-oriented image "understanding" may offer assistance to human perception, cognition and decision-making, such as in computer-aided diagnosis systems, or may enable more natural ways for human-computer interaction (HCI) in perceptual interfaces and in pervasive computing systems. Such vision-based technologies find promising applications in several areas of health care, including but not limited to image-based diagnosis and therapy planning, minimally invasive surgery, assistance and support for people with disabilities and elderly.

However, HCI design for medical applications is a difficult problem. Scientific models for the clinicians' acceptance of technology attempt to identify why physicians and other care providers are reluctant to introduce new HCI technology in their daily work routine.

It is therefore expected that a successful integration of computer vision algorithms in health-related HCI should consider both user-centered and task-based design paradigms. In return, these paradigms influence the basic assumptions as well as the algorithmic development of computer vision techniques.

This tutorial will cover design issues for computer vision-based HCIs in several medical key-fields:

• data visualization for image-guided diagnosis, therapy planning and surgery
• the operating room
• assistance to motor-impaired patients
• monitoring and support of elderly for independent living

The emphasis will on the contribution of the embedded computer vision techniques to the usability of interfaces for each specific domain.

Sponsored by IEEE Canada and the IEEE Newfoundland and Labrador Section
by the