Magnetic resonance imaging (MRI) is a widely used, expensive procedure to obtain detailed images of the human body for diagnosis of many medical conditions. The quality of the images is significantly affected by movement of the patient, with some images even being rendered unsuitable for use. This is important because of the substantial cost involved in use of the scanner and clinical support of the patient. To minimize the risk of anxious movement causing a failed scan, some patients are sedated using medicine, however it may be that this is sometimes unnecessary. Only those who undergo an MRI without sedation first, will know if sedation was required, incurring significant cost.

This master thesis project investigates a new low-cost technology option, the use of a fully immersive virtual reality (VR) simulation of the medical procedure itself as a means to estimate patient response. A prototype is developed and tested in a user study, first in the lab with volunteers, and then with MRI hospital staff on site. Once refined, the virtual reality simulation is offered to patients immediately prior to their scheduled MRI scan. Patient anxiety levels are recorded throughout the VR and MRI procedure to gain a clearer understanding of stress profiles of individual patients. The main question is whether VR is useful in predicting anxiety of patients during MRI.

Results showed that there was a strong correlation between patient anxiety in VR and in MRI but also, if VR were removed a strong correlation existed between before MRI and during MRI. Both significantly predicted average anxiety during MRI, with VR accounting for 71.4% of anxiety during MRI of which 58.8% could be predicted by using anxiety before any scan. The main effect of exposure showed that there was no statistically significant difference in anxiety level between those who had VR and those who did not. The main effect of stage showed there was no statistically significant difference between anxiety level at the different time points of the scan. Overall anxiety level data showed no statistically significant interaction.