Abstract:
Retinal lesion size has been a important parameter studied in clinical trials over the last few decades. Methods of converting measurements on film to 'size on the retina' were initially defined by optical principles and validated by empiric measurements. However, despite advances in computing technology and the advent of digital imaging, little has been done to further validate these numeric conclusions in recent years. How do we know that what we call a ‘retinal millimeter’ is actually a millimeter? For instance, refractive error has been known to affect retinal image magnification for decades but is not commonly factored into lesion size calculations. In addition, factors such as the optical design of the fundus camera and the field of view affect measurements from photographs or digital images. Although these variables are relatively invariant for data obtained from a single subject, they may introduce significant error into clinical trial data gleaned from large populations of patients.
280 color and red–free fundus photographs taken with different cameras and sensors were obtained prospectively in a non–randomized fashion from 8 eyes of 4 subjects wearing contact lenses of various powers. Common landmarks were identified in all images from the same subject eye in order to empirically calculate relative magnifications due to the induced refractive errors. In addition, the range of focus–dependent magnification alteration in one brand of camera was quantified and modeled. Finally, 3 eyes with intraocular retinal prostheses of known dimensions were measured and compared to empirical models.
Magnification changes due to refractive errors can be approximated with a simple linear relationship. A similar relationship exists for inadvertent magnification due to non–telecentric focusing. These relationships are validated with an intraocular object of known size.
Induced refractive errors and non–telecentric focusing of fundus cameras may alter the size of fundus features by more than 10%. These effects are predictable, can be modeled accurately, and can be neutralized retrospectively or in real–time.
Keywords: imaging/image analysis: clinical • clinical (human) or epidemiologic studies: systems/equipment/techniques • age-related macular degeneration