Using eye-tracking to assess visual attention in head-mounted devices (HMD) opens up many possibilities for virtual reality (VR)-based therapy. Existing therapy concepts where attention plays a major role can be transferred to VR. Furthermore, they can be expanded to a precise real-time attention assessment, which can serve as a foundation for new therapy approaches. Utilizing HMDs and eye-tracking in a clinical environment is challenging because of hygiene issues and requirements of patients with heterogeneous cognitive and motor impairments. In this paper, we provide an overview of those challenges, discuss possible solutions and present preliminary results of a study with patients.
Virtual Reality (VR) is a promising field for new forms of therapy. It is a flexible tool for therapists to provide an immersive and interactive environment for patients. VR has been used in the treatment of phobias , therapy of addictions  and motor rehabilitation . Implementing eye-tracking into head-mounted devices (HMD) introduces a precise real-time assessment of attention and opens up new possibilities for VR-based therapy, e.g. analyzing cognitive processes  or diagnosing medical conditions . Additionally to common VR-related issues, specific conditions persist when using HMDs in a clinical context. Strict hygiene requirements have to be met to protect patients, particularly considering the prevalence of multi-resistant organisms.
To achieve sufficient positional accuracy of the eye-tracking, a re-calibration is required every time the HMD has been moved . Hence, eye-tracking creates additional workload for patients and therapists through the calibration procedure. Adequate procedures have to be found to minimize this additional workload in order to be applicable for patients with cognitive deficits.
Setting up eye-tracking systems in head-mounted devices in a way that it (i) meets the hygiene requirements for use in clinical applications, (ii) is suitable for patients with heterogeneous motor and cognitive limitations and (iii) still yields valid and reliable attention data throughout a therapy session is a complex challenge we like to address in this paper.