Cross-Device Field-of-View Scale Differences in Clinical OCT Angiography: Phantom Calibration and Large-Scale Clinical Validation
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Purpose
Device-specific field-of-view (FOV) may affect quantitative optical coherence tomography angiography (OCTA) metrics. This study quantifies inter-device FOV scale differences across four commonly used clinical OCTA devices.
Methods
Four devices were evaluated under a nominal 6×6 mm (or 20°×20°) scanning protocol: Zeiss Cirrus, Heidelberg Spectralis, Topcon Triton, and Topcon Maestro2. Device-specific spatial scale was assessed by (1) phantom calibration using a model eye with a 1×1 mm checkerboard, computing scale factors relative to Zeiss Cirrus; and (2) in vivo validation, registering same-eye, same-visit retinal images from the three devices to those of Zeiss Cirrus and extracting scale factors from the transformation matrices. Correlation between DICOM-implied FOV and spherical equivalent (SE) was assessed, where DICOM-implied FOV equals pixel-spacing × image-dimension.
Results
Phantom measurements showed that Spectralis, Triton, and Maestro2 each captured a systematically smaller FOV than Cirrus, with geometric mean scale factors of 0.933, 0.954, and 0.967. The same pattern was observed across 801 eyes from 556 participants, giving geometric mean scale factors of 0.939, 0.943, and 0.960, agreeing with phantom estimates within 1.2%. DICOM-implied FOV was fixed at 6.000×6.000 mm for three devices but varied from 5.212×5.209 to 6.743×6.739 mm for Spectralis, correlating with SE (ρ = −0.635, P < 0.001).
Conclusion
Systematic FOV discrepancies exist across clinical OCTA devices despite nominally identical protocols, and DICOM pixel-spacing should be interpreted with caution as a spatial measure.
Translational Relevance
The reported device-specific scale factors could serve as magnification corrections for quantitative OCTA metrics, addressing the systematic spatial inter-device discrepancies.
