Abstract
Purpose :
A need exists for cost-effective standardized tools to objectively characterize morphology and volume of orbital and adnexal anatomy. We compared a custom Photogrammetry for Anatomical CarE (PHACE) system with other low and mid-cost 3D imaging technologies for quantitative facial modeling.
Methods :
We compared two low-cost and two mid-cost 3D scanning technologies. The PHACE system and the iPhone Scandy Pro application (Scandy, USA) are low-cost systems (<$500), whereas the Einscan Pro 2X (Shining 3D Technologies, China) and ARC7 facial scanner (Bellus 3D, USA) are two mid-priced structured light sensing scanners (~$5000). Imaging was performed on human subject faces with Fitzpatrick skin types 2, 4 and 6, with and without 3D printed phantom lesions affixed above the brow line. The PHACE system used two Google Pixel 3 smartphones on two rotating turntables to acquire 90 photographs, which were then rendered into 3D models using Metashape (Agisoft, Russia) - a photogrammetry software. The rendered hemisphere volumes using PHACE, Einscan, Scandy, and ARC7 devices were calculated with CloudCompare (open-source point cloud software). The performance of each facial scanning technology was evaluated by quantitatively comparing digitally measured volumes to the calculated volumes of the 3D printed phantom lesions.
Results :
The Einscan qualitatively (Fig. 1A) and quantitatively rendered 3D printed phantom lesions with greatest resolution and accuracy (mean difference of 2.2 ± 2.8% for 33.5 μL). The mean differences between the calculated volume of the small phantom lesion (124 μL) and the rendered volumes for the PHACE, Scandy, and ARC7 systems were 4.7 ± 3.7%, 9.1 ± 0.9%, and 22.0 ± 17.9%, respectively.
Conclusions :
The PHACE system outperforms the Scandy and more expensive ARC7 system to accurately render volumes as little as 124 μL. The Einscan can accurately render volumes down to 33.5 μL but it is 10x the cost. We demonstrated an optimized system to generate 3D models of the human orbit and adnexa while balancing cost and performance.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.