Multimodal optical imaging of the oculofacial region using a solid tissue-simulating facial phantom

Spatial frequency domain imaging (SFDI) applies patterned near-infrared illumination to quantify the optical properties of subsurface tissue. The periocular region is unique due to its complex ocular adnexal anatomy. Although SFDI has been successfully applied to relatively flat in vivo tissues, regions that have significant height variations and curvature may result in optical property inaccuracies.

We characterize the geometric impact of the periocular region on SFDI imaging reliability.

SFDI was employed to measure the reduced scattering coefficient ( ${{\mu }_s}^{\text{'}}$ ) and absorption coefficient ( ${\mu }_a$ ) of the periocular region in a cast facial tissue-simulating phantom by capturing images along regions of interest (ROIs): inferior temporal quadrant (ITQ), inferior nasal quadrant (INQ), superior temporal quadrant (STQ), central eyelid margin (CEM), rostral lateral nasal bridge (RLNB), and forehead (FH). The phantom was placed on a chin rest and imaged nine times from an "en face" or "side profile" position, and the flat back of the phantom was measured 15 times.

The measured ${\mu }_a$ and ${{\mu }_s}^{\text{'}}$ of a cast facial phantom are accurate when comparing the ITQ, INQ, STQ, and FH to its flat posterior surface. Paired $t$ tests of ITQ, INQ, STQ, and FH ${\mu }_a$ and ${{\mu }_s}^{\text{'}}$ concluded that there is not enough evidence to suggest that imaging orientation impacted the measurement accuracy. Regions of extreme topographical variation, i.e., CEM and RLNB, did exhibit differences in measured optical properties.

We are the first to evaluate the geometric implications of wide-field imaging along the periocular region using a solid tissue-simulating facial phantom. Results suggest that the ITQ, INQ, STQ, and FH of a generalized face have minimal impact on the SFDI measurement accuracy. Areas with heightened topographic variation exhibit measurement variability. Device and facial positioning do not appear to bias measurements. These findings confirm the need to carefully select ROIs when measuring optical properties along the periocular region.