Abstract
Purpose :
Previous studies have shown that regression models incorporating axial eye length or refractive error significantly mitigate magnification bias in the assessment of nerve fiber layer (NFL) thickness using optical coherence tomography (OCT). This study investigated the potential of using the autofocus reading from OCT machine as a surrogate refractive parameter for magnification bias mitigation.
Methods :
Healthy eyes were scanned with 6x6mm disc cubic scan using Solix spectral-domain OCT (Visionix/Optovue, CA, USA). The autofocus values (range -100~100) used in optimizing the scans were exported from OCT. Axial length was measured with Zeiss IOL master 700. Scans with signal strength index (SSI) less than 50 and eyes with prior cataract and keratorefractive procedures were excluded. The association between autofocus value and axial length was analyzed by Pearson correlation. Utilizing multiple linear regression models, we determined the coefficients for adjusting peripapillary NFL thickness. We evaluated three models for reducing population variation of the overall average NFL thickness: Age only, age and axial length, and age and autofocus.
Results :
43 eyes of 43 participants were included in the study. The age of the participants was 58.1 ± 13.5 years (mean±SD), and their axial length was 21.28 - 27.95 mm (range). The autofocus value was -6.2 ± 38.5 (mean±SD). Our findings revealed a strong negative association (Pearson correlation, r=-0.853) between autofocus reading and axial length (Figure 1). Additionally, models incorporating axial length or autofocus significantly reduced population variation in peripapillary NFL thickness (CV=7.8%, p=0.006 and CV=8.6%, p=0.030, respectively; Table-1).
Conclusions :
The autofocus reading, measured directly by OCT, is highly correlated with axial length and can be used to mitigate magnification-related bias in NFL thickness and the tighten normative range. Since the autofocus reading is automatically available for each OCT scan without requiring additional measurement or manual operator data entry, it may be a convenient and practical way to improve glaucoma diagnostic accuracy and provide a solution to the problem of false positive disease classification in myopic patients.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.