Abstract
Early detection of mild cognitive impairment (MCI) is essential for preventing progression to dementia. Conventional neuropsychological tests, such as cube copying and clock drawing, are widely used to assess visuospatial and executive functions; however, these assessments rely largely on qualitative observation by clinicians, which may limit their objectivity. Therefore, a quantitative method for objectively evaluating visuospatial function is desirable. This study develops and validates a virtual reality (VR)-based assessment using eye-tracking that enables objective differentiation between healthy elderly adults and individuals suspected of having MCI. VR content presenting periodic depth stimuli was developed, and gaze behavior was recorded using a head-mounted display equipped with an eye-tracking system. Changes in interpupillary distance induced by depth perception were analyzed to derive quantitative indices, including the coefficient of determination (
R
2
) obtained using sinusoidal fitting and spectral power in the frequency domain. Experiments were conducted with elderly participants consisting of a healthy group and individuals suspected of having MCI, as determined based on the Japanese version of the Montreal Cognitive Assessment. Significant differences were observed between the groups (
p
=0.026), and linear discriminant analysis achieved accurate classification, with no misclassification in the healthy group and two misclassifications in the suspected MCI group. These findings indicate that the proposed method enables objective and quantitative evaluation of visuospatial function and may contribute to the early detection and screening of MCI.