Keratoconus screening based on data of the high resolution anterior segment OCT Casia 1000

Abstract

AbstractPurpose Anterior segment OCTs are more and more used in clinical practice for evaluation of corneal disorders. The goal of the study was to derive an automatic screening method for diagnosis of forme fruste and early keratoconus based on data extracted from the OCT CASIA‐1000 (Tomey Inc., Japan).Methods From the Casia‐1000 we exported the following raw data: anterior and posterior surface dioptric power data, anterior and posterior surface elevation data, pachymetric data (central, thinnest + X‐Y‐position). These data were decomposed into Zernike base of radial degree 6 and single indices and combinations (compound index 1 and 2) were tested for specificity and sensitivity (ROC analysis, area under curve). Ninety eyes of 45 patients were included in this study (1 eye each with clinical manifest signs of keratoconus the other without clinical signs of keratoconus each) and 48 eyes of 48 normals as reference.Results For the single indices the tilt and coma component of 1st and 2nd radial degree showed the best discriminance level between normal and suspects/keratoconus (area up to 0.74/0.86). Compound index 1 considering anterior and posterior surface coma of 1st and 2nd radial degree and pachymetry yielded a discriminance level of 0.821/0.953 and compound index 2 without considering pachymetry showed 0.799/0.935, respectively.Conclusion The single indexes do not seem do provide better discrimination between normals and suspects/keratoconus compared to the literature data for corneal topography (e.g. Klyce, Maeda, Smolek) or Tomography (Belin&Ambrósio). But in contrast, especially compound index 1 shows an excellent performance. These results have to be validated in a larger case series comparing Casia‐1000, TMS and Pentacam.