Abstract
Purpose: :
To demonstrate the value of using a logarithmic transformation of Ocular Coherence Tomography (OCT) data in clinical research.
Methods: :
Retrospective analysis of data comparing OCT retinal thickness measurements with LogOCT.
Results: :
The log transformation of OCT thickness data helps normalize the typically skewed distribution in persons with macular thickening from diabetic macular edema (see figure). In addition, the log transformation helps address the issue of the same micron change in OCT being qualitatively different for a retina that is very thick at baseline compared with a retina with only a small amount of thickening, just as the use of the logarithm of the minimal angle of resolution (LogMAR) helps address a similar problem with visual acuity, where thresholds for change are larger at lower levels of visual acuity. The proposed transformation is made in a way to be clinically useful by dividing the OCT measurement by 200 prior to taking the log, so that LogOCT of 200 microns (near normal retinal thickness) is equal to zero, making it comparable to the LogMAR scale where LogMAR of zero is 20/20 or near normal visual acuity (see table). In addition, one log step change in the LogOCT scale is more than twice the error of measurement for any degree of retinal thickening, making it useful as a possible outcome variable for clinical research.
Conclusions: :
Use of the LogOCT scale may facilitate data analyses using OCT data in a similar fashion as use of LogMAR data has facilitated data analyses using visual acuity data.Supported by NEI and NIDDK, NIH, DHHS; EY14231, EY14269, EY14229
Keywords: diabetic retinopathy • imaging/image analysis: clinical