How to forge criteria for assessment of visual depth perception in Virtual Environments

San Diego 31 January 1998 On the last conference day of Medicine Meets Virtual Reality:6, which took place in San Diego at the end of January, Dr. Helene Hoffman, professor at the University of California (UCSD) School of Medicine, presented a well-defined concept for a behaviourally based method to obtain precise information on the impact of Virtual Environment (VE) design choices on perception and performance of perceptual-motor tasks. This experiment fits into a series of studies to elucidate the relative merits of VR-based teaching systems such as the Anatomic VisualiseR © prototype and the role of interface design on educational outcomes. Before the VisualiseR © will be implemented within UCSD's preclinical anatomy curriculum, the exact cost-benefit relation of the application should be determined.

Advertisement

On the last conference day of Medicine Meets Virtual Reality:6, which took place in San Diego at the end of January, Dr. Helene Hoffman, professor at the University of California (UCSD) School of Medicine, presented a well-defined concept for a behaviourally based method to obtain precise information on the impact of Virtual Environment (VE) design choices on perception and performance of perceptual-motor tasks. This experiment fits into a series of studies to elucidate the relative merits of VR-based teaching systems such as the Anatomic VisualiseR © prototype and the role of interface design on educational outcomes. Before the VisualiseR © will be implemented within UCSD's preclinical anatomy curriculum, the exact cost-benefit relation of the application should be determined.

In the first proposed experiment, the team of Professor Hoffman intends to assess the effect of both the level of immersion or the viewing condition and the twofold aspect of layout/design of the VE on the performance of an engaging, game-like task. The test comprises four equipment configurations, representing three types of visual display: no immersion with regular colour monitor; partial immersion with shutterglasses and full immersion with HMD#1 and HMD#2. These four levels are combined with two layout scenarios: with or without depth cues such as shadow addition, perspective grid, size changes and so on. In this way, eight different experimental conditions are created and tried out ten times each by every testing subject.

The selected perceptual-motor task has to correspond to a fixed set of criteria. It should be simple without any demands on cognition such as problem solving, retrieval of previous learning or other analytic activity in order to avoid influence on the interpretation of the results. Yet, it has to be intrinsically motivating to make the testing subjects do their best. The Hoffman team therefore chose a twofold action: localisation of a virtual object in relation to other virtual objects and location of a virtual 'hand' or cursor responding to movements of the subject's own hand with relation to various virtual objects. Variations in the virtual objects' size and appearance speed progressively complicate the actions.

The subjects are recruited from the UCSD student population and will be submitted to preliminary clinical tests of depth perception enabling the researchers to take in account the individual differences in performance due to physiological and psychological factors which would otherwise negatively influence the testing results. The following parameters are integrated in the actual test: duration and accuracy of the performance; cognitive and physical workload as indexed by the NASA Task Load Index measurement instrument; fatigue as indexed by the Yoshitake Fatigue Scale and cybersickness indexed by the Kennedy Simulator Sickness Questionnaire.

Other variable factors constitute the binocular convergence and the focus of attention. These are measured by means of Video-Oculography (VOG), a micro-electronic technique using small CCD video cameras to capture a front image of each eye and track the direction of gaze. In this way, the subject's horizontal and vertical depth attending and convergence angle aligning can be determined. All these performance measures are crucial for the fast acquisition of spatially-based cognitive and perceptual-motor skills which are needed in the learning and application of functional human anatomy. Real-time interaction as well as visual, haptic and auditory rendering of objects in 3-D space help students to get used to Virtual Environments.

Yet, a lot of questions on which factors and variations affect the perception of depth in VE systems, still remain unanswered, according to Professor Hoffman. Another issue constitutes the absolute necessity of full immersion for the learning environment. Until final analysis has proven the specific effects of the different interface options, practical and philosophical considerations will determine the choice. The described methodology has to be refined and validated in order to define objective user-centred criteria for the design and use of VE systems such as the Anatomic Visualiser ©. If you like to know more about the educational projects of the Hoffman research team, please, consult the UCSD School of Medicine web site.


Leslie Versweyveld

[Medical IT News][Calendar][Virtual Medical Worlds Community][News on Advanced IT]