King's College London

New eye tracking technology has been developed which enables gaze based human computer interaction to operate immediately and robustly without any explicit set-up tasks. Integrating this technology as a control interface for the MR-compatible VR system allows young children to instantly engage in an immersive and interactive VR experience during MRI scans.

The novel eye tracking controlled VR system has been successfully tested on 23 children aged 2 to 13, allowing them to play games and watch films during MRI using only their eyes.

In human-computer interaction (HCI), gaze estimation is the process of determining where a person is looking. Gaze-based HCI systems mostly require calibration or complex setup at the start of each session, hindering instant control. This limitation affects usability, making immediate interaction challenging, especially for young children who benefit from easy and fast technology use.

A research team from the School of Biomedical Engineering & Imaging Sciences has developed an innovative eye tracking technology that allows for instant user control. This technology forms the core interaction interface of their groundbreaking MR-compatible VR system, designed to immerse children in an interactive virtual world during MRI scans. 

MRI scans are noisy and stressful, often causing discomfort and movement especially in children, which can lead to scan failure. Instant interaction is crucial as it quickly engages children with VR, reducing anxiety and minimising movement.

To make the gaze-controlled VR experience engaging, game and video content has been developed, allowing easy customisation for each child's preferences. Interaction is simple: the child holds their gaze on items on the screen, triggering actions like playing a game, watching a video, or interacting with their favourite cartoon character. Maintaining a continuous sense of control is crucial for immersion in the VR system. Therefore, the innovative eye tracking system updates itself based on user interaction. The more the child interacts, the more accurate the gaze estimation becomes.

The VR experience also aims to reduce a child's tendency to move their head during the scan as much as possible, but not all head movement can be eliminated. To remedy this, the team have also used the DISORDER method previously developed at King’s College London for baby MRI scans, which retrospectively performs motion correction on scanned images.

Combining these two innovations, enables the creation of a new system that can acquire high quality brain MR images from awake young children.

Read the full study: Instant interaction driven adaptive gaze control interface.