Haze-Dependent Breakdown Of Pachymetric Agreement Between Scheimpflug Imaging And As-Oct

Published 2026 - 44th Congress of the ESCRS

Reference: FP02.06 | Type: Free paper | DOI: 10.82333/fp79-2e30

Authors: Dr ASIM BURAK GUNDUZ* ¹ , Prof. Dr EMINE AKCAY ²

¹OPHTHALMOLOGY,ANKARA BILKENT CITY HOSPITAL,ANKARA,Türkiye, ²OPHTHALMOLOGY,ANKARA YILDIRIM BEYAZIT UNIVERSITY,ANKARA,Türkiye;OPHTHALMOLOGY,ANKARA BILKENT CITY HOSPITAL,ANKARA,Türkiye

Purpose

To investigate whether corneal haze severity affects the agreement between Scheimpflug-based pachymetry and anterior segment optical coherence tomography (AS-OCT), and to determine a quantitative haze threshold associated with clinically significant inter-device pachymetric discrepancy.

Setting

Ankara Bilkent City Hospital Department of Ophthalmology, tertiary referral center. Patients with previous corneal cross-linking (CXL) and healthy control eyes were included. Imaging was performed using Scheimpflug tomography (Sirius anterior segment analysis system (CSO, Florence, Italy)) and anterior segment OCT (3D OCT-1 Maestro (Topcon Corporation, Tokyo, Japan)).

Methods

Forty-two CXL eyes and 31 control eyes were evaluated. Thinnest corneal thickness was measured with Scheimpflug imaging and AS-OCT. Inter-device difference (OCT–Sirius) was calculated. Stromal haze was quantified on AS-OCT B-scans using ImageJ by measuring mean pixel intensity in predefined anterior and posterior stromal regions of interest. An anterior/posterior stromal reflectivity ratio (Ant/Post haze index) was computed. Clinical haze grading was recorded. Agreement was assessed with Bland–Altman analysis. Correlation, regression, and ROC analyses were performed. Clinically significant discrepancy was defined as ≥30 µm.

Results

The Ant/Post haze index was significantly higher in the CXL group compared to controls (1.34±0.17 vs 0.99±0.11, p<0.001). Clinical haze grading strongly correlated with the quantitative haze index (Spearman r=0.84, p<0.001). Across the entire cohort, haze index showed a strong positive correlation with OCT–Sirius pachymetry difference (r=0.69, p<0.001), which remained significant within the CXL subgroup (r=0.45, p=0.003). Bland–Altman analysis demonstrated a mean bias of 35 µm (limits of agreement −30 to +100 µm), with increasing heteroscedasticity at higher haze levels. ROC analysis revealed excellent discrimination for predicting clinically significant inter-device difference (AUC=0.91). The optimal cutoff value for haze index was 1.14.

Conclusions

Corneal haze severity-both quantitatively measured by anterior/posterior stromal reflectivity ratio and clinically graded at the slit lamp-significantly reduces agreement between Scheimpflug-derived and AS-OCT pachymetry measurements. Increasing clinical haze grade was strongly associated with widening inter-device discrepancy, indicating a haze-dependent measurement bias rather than random variability. While quantitative reflectivity analysis provides objective confirmation, the mere presence of clinically visible stromal haze should alert clinicians to potential pachymetric inaccuracy with Scheimpflug imaging. In such cases, anterior segment OCT may provide a more reliable assessment of corneal thickness.