Combination of Imaging Modalities for Highly Asymmetric Keratoconus
By Lynda Seminara
Selected By: Stephen D. McLeod, MD
Journal Highlights
Ophthalmology, December 2018
Download PDF
Hwang et al. assessed whether variable sfrom Scheimpflug imaging and/or spectral-domain optical coherence tomography (SD-OCT) could help distinguish clinically unaffected eyes of patients with asymmetric keratoconus (AKC) from normal control eyes. They found that a combination of metrics from the 2 modalities was useful for this purpose and was superior to the metrics of either modality alone.
The authors reviewed medical records of 30 patients with AKC. In these patients, 1 eye had clinical evidence of keratoconus (based on slit-lamp, retinoscopic, and topographic findings) and corrected distance acuity worse than 20/20. The fellow eye was clinically unaffected, with corrected distance acuity of 20/20 or better. The control group consisted of 60 normal eyes of 60 patients who had uneventful LASIK and at least 2 years of follow-up. Scheimpflug imaging and SD-OCT were obtained for all study eyes, and receiver operating characteristic curves were generated to determine area under the curve (AUC), sensitivity, and specificity for each machine-derived variable and each combination of variables. The main outcome was the ability to distinguish clinically unaffected AKC eyes from controls.
According to the analyses, no individual machine-derived metric from Scheimpflug or SD-OCT technology was able to produce an AUC >0.75. With Scheimpflug imaging, the best results were achieved by combining 5 metrics: index height decentration, index vertical asymmetry, pachymetry apex, inferior-superior value, and Ambrosio’s relational thickness maximum. Together, they produced AUC of 0.86 and sensitivity and specificity of 83%. For SD-OCT, an aggregate of 11 thickness-related parameters achieved the greatest accuracy, yielding AUC of 0.96 and sensitivity and specificity of 89%. However, the best results were obtained with a mix of 13 metrics from the 2 modalities, which produced AUC of 1.0 and sensitivity and specificity of 100%. The most influential variables in combination models were epithelial thickness and total focal corneal thickness (from SD-OCT) and anterior curvature and topometric indices (from Scheimpflug). No posterior corneal metrics were helpful.
Identifying corneal ectasia at its earliest stages is a challenge and will likely remain so until it’s possible to directly measure corneal biomechanics rather than corneal morphology alone. At present, a combination of metrics from the Scheimpflug and SD-OCT modalities appears to have excellent discriminative utility. (Also see related commentary by Stephen D. Klyce, PhD, in the same issue.)
The original article can be found here.