Hints on Color Vision Testing


Among all the topics of neuro-ophthalmology, color vision may be the most forbidding one. Color vision is a difficult subject, with its own terminology, its own intricate test devices, and its own voluminous literature. Further, on the clinical side, the subject is burden by many misconceptions. For example, the ubiquitous Ishihara pseudo-isochromatic charts are often held to be useful for detecting and quantifying acquired color vision defects, in spite of the facts that they were expressly devised for congenital defects, and that they lack quantitative features. Surely, it should be possible to make better selections. The following text aims to give a Gestalt for color vision deficiences, and primarily acquired deficiences, to aid the selection of meaningful tests. The text will not be burdened by references. Readers looking for a detailed treatment may enjoy Congenital and Acquired Color Vision Defects, by Pokorny, Smith, Verriest, and Pinckers (Grune & Stratton, NY 1979). Philippe Lanthony's studies are also highly germane, particularly the paper, La zone neutre des dyschromatopsies acquises, published in the Annales d'Oculistique 210: 937 - 946, 1977.

The term color vision usually refers to central vision and this will be the case in the following presentation. Colors can be seen in peripheral vision too, but in a poor-sibling kind of way. In the distant past, color was a leading actor in visual field examinations. Color perimetry is now mainly of historical interest and will be left aside here.

Modelling color

A major obstacle when looking for useful tests for acquired color vision defects is the lack of a theoretical foundation. Instead, the problem has to be approached in an empirical way, by analysis of key differences between normals and congenital and acquired dyschromats. It may be best to begin by defining some basic color terms.

Color properties can be named and arranged in many different ways. The CIE diagram and color circles (right panels) are perhaps the most common but when trying for a Gestalt of normal and abnormal color vision, the top left panel has some advantages. It will be called a color square in the following. Readers familiar with the Microsoft Paint software will recognize this panel, with hues represented horizontally and saturation represented vertically. The narrow scale to the right represents the third dimension, the lightness. Hue and saturation go a long way to illuminate basic characteristics of normal and abnormal color vision. Additional characteristics are needed to define the pigment-based colors that are used in clinical settings, as well as the sources of illumination. These additional properties will be largely left aside in this brief overview.

The color square approach used here is not directly comparable with the standards used for defining pigment-based colors. The figures and diagrams shown below aim to illustrate general principles and cannot accurately mimic exacting test conditions.


Congenital color vision defects

Generally, congenital color vision defects are much better understood than their acquired counterparts. An important exception concerns the way people with congenital defects may perceive their deficits: in contrast to people with acquired defects, they have no normal, trichromatic standard to compare with. Therefore, descriptions of what the congenitally color deficient world looks like are hard to come by. It is not all gray and dreary, however. Actually, congenital dyschromats may perform better than normal subjects in some types of color tasks. A good example comes from the military world, where a specific type of congenital red-green deficiency confers a distinct advantage in recognizing the changes in foliage coloration that occur where troups recently have passed through.

Congenital dyschromatopsia is divided into two major groups, namely, red-green and blue-yellow categories. The latter is extremely rarely encountered and will be left aside here. Red-green variants occur in about 8% of males and 0.4% of females and affect vision within narrow red and green zones of the color spectrum. The deficit can be mimicked by sliding narrow vertical strips in the red and green areas of the color square upwards (lower panel), creating a washed-out appearance of colors in these areas. Expressed in other words, saturation thresholds are raised locally. The distribution of saturation defects has been carefully studied by Lanthony (op. cit.). The following diagram has been redrawn from Lanthony's data. The stippled areas illustrate the range and severity of saturation deficits in congenital red-green dyschromats.

Obviously, subjects with congenital red-green dyschromatopsia should tend to confuse color samples taken from regions with poor apparent saturation. The confusion effect is particularly striking when changing to a color circle model and asking the subject to place color samples in a nearest-neighbor fashion according to perceived hue. The samples should differ in hue only, saturation and lightness being kept constant. This is the format of the Farnsworth D-15 sorting test (right panel). Normal subjects will produce the expected color circle whereas congenital dyschromats will show confusions along well-defined diameters (solid black lines). As can be seen from the D-15 chart, the "circle" is not really a circle but an ellipsoid. The shape goes back to loci with equal saturation in the CIE chart shown above.

Th D-15 test was devised for vocational purposes and not for critical, diagnostic work. This can be gleaned from the presence of a gap to the left in the chart. This gap severely limits the utility of the D-15 in diagnosing acquired color vision defects (see further).

The D-15 is an easy test that intentionally allows low-degree dyschromats to pass. Lanthony has produced a more difficult, desaturated variant of the D-15. Unfortunately, this latter test is a bit too difficult: many normal subjects make errors in this test. The famous Farnsworth 100-hue test also builds on hue sorting. Being time-consuming and tedious to score, the 100-hue test is mostly used in research settings. Peculiarly, it contains 85 rather than 100 different hues.

Confusion also forms the the basis for various types of pseudo-isochromatic test charts, for example, the well-known Ishihara set. The pseudo-isochromatic chart shown to the right is more rarely encountered: it is the Farnsworth Tritan Plate, devised for diagnosing blue-yellow deficits. The pseudo-isochromatic term refers to the fact that the charts look more or less uniformly colored to congenital dyschromats. Most charts use one and the same saturation level and do not allow quantification of the deficit. A notable exception is the Hardy-Rand-Rittler test. The original version has been very popular but color renditions have been found faulty in at least one later edition.

There are many other types of color vision tests that primarily aim for congenital dyschromatopsia. Examples include anomaloscopes and lantern tests. Those who look for the convenience of screen-based tests may have to be patient. Numerous attempts have been made to transplant classical, pigment-based tests to computer displays, not the least by developers of smartphone applications, but exact matching is exceedingly difficult and validation is very demanding. Presently, developers of new tests should find it safest to take recourse to standardized pigment-based color swatches. Such swatches are available from the Munsell and NCS organizations (no affiliation).


Acquired color vision defects

Acquired color vision defects differ from the congenital counterparts in several aspects. Acquired defects are commonly To subjects affected by acquired color vision defects, the most striking aspect seems to be a saturation deficit that gives colors unaccustomed, washed-out appearances. There may also occur apparent changes in hue. For example, an object known to be red may appear to have changed its color to orange or to pale yellow, depending on the severity of the deficit. Such subjective changes can often be brought out quite clearly with simple test targets, particularly when presenting the target alternatingly to the left and right eye, as exemplified in the image to the right. Incidentally, subjects who know that they have congenital dyschromatopsia often state that they cannot do these types of tests because they are "color blind". The reasoning is false: acquired defects can very well be superposed on congenital defects and can also produce an asymmetry of color appearance between the eyes.

Red objects like some eyedrop bottle caps have a strong tradition in clinical color comparisons but red colors are not necessarily the best. Lanthony's diagram (above) indicates that acquired saturation deficits favor the bluish-green to purple regions of the color spectrum, and that optic neuropathies are somewhat more selective for the bluish-green region than are retinopathies. Another reason for staying away from red is that this color is invisible in the protanope variant of congenital dyschromatopsia. The Color Comparison Display presented elsewhere on this site allows selection between bluish-green and red test colors and offers better control over test conditions than informal confrontation tests like the one depicted above.

Subjective color comparisons between the eyes are clinically useful, but being qualitative, they provide little indication as to the degree of any asymmetry and they suffer from the disadvantage that it is never known how good an observer any given patient is. They also have to work against preconceived expectations of symmetry of perception. Therefore, there is much to speak for a quantitative threshold approach as offered by some formal pigment-based tests (see further). A browser-based quick-test like the Color Saturation Threshold Test, CSTT, may prove to be a useful alternative. The CSTT does not suffer from the same limitations as the screen-based replicas of standard pigment tests mentioned above because the CSTT does not need to target the very specific colors required for testing of congenital dyschromatopsia. The test results may still vary somewhat between different types of displays. Notably, pigment-based tests may also show variations between different set-ups unless illumination is carefully controlled. These latter tests are also susceptible to smudging and to wear and tear.

Graphic illustration of vision with acquired dyschromatopsia is difficult. The following color square panels must be taken as no more than highly schematic representations of differences between a normal subject (left panel) and two subjects with different degrees of optic neuropathy. The same type of representation of congenital dyschromatopsia is available above.

For comprehensive illuminations of acquired color vision deficits there are no alternatives to formal testing. For best results, formal tests for acquired dyschromatopsia should either broadly cover the full color spectrum or concentrate on the bluish-green to purple region. The highly selective tests for congenital dyschromatopsia are obviously disqualified on this basis. The popular D-15 test largely excludes the clinically most important part of the color circle. Expanded versions of the D-15, e g, Roth's 28-hue test, may be more useful but documentation is sparse. Like the D-15, the 100-hue test does not illuminate saturation deficits and may be too time-consuming for routine clinical work. From the above reasoning, I would argue that one of the best formal tests is Ph. Lanthony's New Color Test. Its sole disadvantage may be a somewhat long testing time. However, there is a short-cut: the Sahlgren's Saturation Test. The SST uses the same test principle but concentrates on the bluish-green to purple color region. Combination with Ishihara's pseudo-isochromatic charts is a good way to illuminate any concurrent congenital dyschromatopsia.

When trying to make an optimum selection among all the available color vision tests it is appropriate to consider the general utility of testing for acquired color vision defects. After all, acquired dyschromatopsia nearly always goes hand in hand with subnormal visual acuity. Maybe the best application of color tests is in cases with unexplained acuity loss. If color vision is normal, the cause is likely to be optical. There is a direct approach to assess optical quality, namely, by means of a double-pass refractometer.

Disclosure: The SST is produced and distributed by Visumetrics of which I am part owner.