The image of the ideal Landolt C (left),
imaged through the optics of our eyes
onto the retinal photreceptor mosaic.
One of the most frequent tasks for the ophthalmologist and optometrist is measuring visual acuity. It is also a primary measure in basic research on the physiology and pathophysiology of the visual system. Initially, measuring acuity seems very simple: find the smallest shape that is correctly detected. But what means ‘detection’, what is an appropriate test shape (‘optotype’), can we use letters, are all letters created equal? Analysis of these questions leads into deep waters of sensory physiology and signal analysis, but the problem can be adequately solved. A preferred optotype is the ‘Landolt C’, an annulus with a gap in one of 8 major directions. Signal detection theory has provided rapid and accurate procedures to estimate the sensory threshold, one of these being the “best PEST” (best Parameter Estimation by Sequential Testing, Lieberman & Pentland 1982). The entire procedure lends itself well for computer implementation, where graphic presentation of annuli with gaps derserves special attention.
 |  |  |
| Ideal shape on the pixel raster | Pixel-rastering without anti-aliasing | Pixel-rastering with anti-aliasing |
There is an unexpected problem, though: Our visual acuity is so high that even at 5 m distance the spatial resolution of computer screens is insufficient to adequately assess acuity (Bach 1996). One could increase observer distance, but that would require huge screens under pathologic conditions. LCD displays currently are worse than cathode ray tubes, and the latter are limited by shadow masks electron optics, so-called high-resolution screen are just bigger, the pixel size is not smaller. But advanced computer graphics provides a solution, the so-called ‘anti-aliasing’. This term has been popularized in flyers for CD players and is increasingly used in computer graphics. Spatial resolution is improved with the help of intensity resolution. Optics of the human eye represent a low-pass filter for spatial frequencies, and --after passage through the optics-- an anti-aliased image leads to the same retinal image as an image with veridical higher resolution. The principle is depicted in the adjacent figure (Bach et al. 1996).
The advantages of the Macintosh were quite apparent during the implementation of anti-aliasing: Anti-aliasing is a built-in feature of the Macintosh operating system! [For insiders: The “CopyBits” call with a destination Rect 4x smaller than the source Rect in “ditherMode” does it all.] Even on what now appears a very ‘lame’ computer like an LC (with an 68020-CPU), the test works fine. Very advantageous is also the Apple Desktop Bus: The screen should be observer in 5 m distance. So a small input box is necessary for the subject, and an external key pad for PowerBooks proves to be an ideal solution. Unused keys are covered, and the others have an appropriate sticker (see figure above). For the necessary extension cable an unexpected industry standard helps: The ADB plug is also used for Sony SVideo, so it can easily be obtained. [Note added in 2000: For more recent Macs with USB connectors appropriate key pads are readily available.]
With these ideas we implemented the “Freiburger Visual Acuity Test”. In the ophthalmic department of Freiburg University it is permanently installed at several sites. Children use it like a video game. Visual scientists use it for rapid screening of visual function. Many options in the program allow customization for various situations. Last but not least: The current version of the “Freiburger Visual Acuity Test” is available free (but without health guarantee) at: