Rapid measurement and machine learning classification of colour vision deficiency

Abstract

Colour vision deficiencies (CVDs) indicate potential genetic variations and can be important biomarkers of acquired impairment in many neuro-ophthalmic diseases. However, CVDs are typically measured with tests which possess high sensitivity for detecting the presence of a CVD but do not quantify its type or severity. In this study, we introduce Foraging Interactive D-prime (FInD), a novel computer-based, generalisable, rapid, self-administered vision assessment tool and apply it to colour vision testing. This signal detection theory-based adaptive paradigm computed test stimulus intensity from d-prime analysis. Stimuli were chromatic Gaussian blobs in dynamic luminance noise, and participants clicked on cells that contained chromatic blobs (detection) or blob pairs of differing colours (discrimination). Sensitivity and repeatability of FInD colour tasks were compared against the Hardy-Rand-Rittler and the Farnsworth-Munsell 100 hue tests in 19 colour-normal and 18 inherited colour-atypical, age-matched observers. Rayleigh colour match was also completed. Detection and discrimination thresholds were higher for atypical than for typical observers, with selective threshold elevations corresponding to unique CVD types. Classifications of CVD type and severity via unsupervised machine learning confirmed functional subtypes. FInD tasks reliably detect inherited CVDs, and may serve as valuable tools in basic and clinical colour vision science.

Keywords: colour detection; colour discrimination; colour vision deficiency; cone-isolating directions; k-means clustering; unsupervised machine learning; vision diagnostics.

Figure 1.

Illustration of FInD stimuli and experimental procedures. (a) FInD detection (left) and discrimination (right) task interfaces. (b) The colour wheel shows a cross-section of the Hue, Saturation, Value (HSV) space for V=1, from which six hues (0° to 300° in 60° steps) and two saturation levels (0.5 and 1) were chosen and used in the discrimination task. If, for instance, yellow (60°) discrimination was tested, then two colours were symmetrically selected the same angular hue distance (θ/2) away from yellow with a fixed saturation level. (c) Illustration of signal detection theory. The noise distribution (blue) and signal distribution (grey) bell curves lie on the normalised Z-score abscissa. Detectability or discriminability (d’) and criterion (λ) are depicted. The areas under the curves correspond to “hit”, “miss”, “false alarm” and “correct rejection”, respectively, according to stimulus presentation and response. d’ can be calculated by z(false alarm)−z(hit). (d) FInD experimental procedures with cone isolating direction detection stimuli as an example. The dashed arrow represents the adaptive procedure that selects a range of stimulus intensities on the second chart based on analysis of the responses to stimuli on the first chart. (e) An example of a typical psychometric function: blue data show the probability that the observer reported the presence of a stimulus as a function of intensity; vertical lines show binomial standard deviation; the red curve shows the best fitting function for equation 1 and black dashed lines represent the upper and lower 95% confidence intervals. The leftmost data point, which is on the left of the break on the horizontal axis, indicates the false alarm rate (2.4% in this case).

Figure 2.

FM100 hue test results. Left: average error score pattern of 19 colour normal (CN). Hues of coloured caps are numbered from 1 to 85 with the corresponding mean error score (black line) and standard error range indicated along the radial coordinates for each hue. Upper and lower standard error ranges are depicted by the red and blue dashed lines, respectively. The outermost circle where the colour dots reside represents an error score of 3.5, and the centre error score is 2, indicating the lowest possible error score. Right: error score pattern (black line) of an example colour vision deficient (CVD) observer (CVD#5). Note that the largest radial scale is 16. The mid-point and total error score (TES) of this observer as well as diagnostic curves (colour arcs) are also shown. The mid-point of this observer fell in the protan range.

Figure 3.

FInD Color detection and discrimination thresholds. Thresholds of all colour normal (CN) participants are plotted as coloured circles in all panels as references, and thresholds of each colour vision deficient (CVD) observer are denoted by black squares (detection) or crosses (discrimination) in separate panels. (a) FInD Color detection thresholds (upper panels) are plotted as cone contrast vector length and (b) low-saturation discrimination (lower panels) thresholds are plotted in degrees of Hue, Saturation, Value (HSV) colour space angle.

Figure 4.

Equiluminant colour discrimination. The colour wheel on the left shows the equiluminance plane with four primary axes representing the red-green and blue-yellow postreceptoral mechanisms. For example, in the purple (S+, 90°) discrimination condition, two colours are symmetrically selected at the same angular distance (θ/2) away from purple. The panels on the right show results of 6 colour vision deficient (CVD) participants for the FInD Color discrimination task with equiluminant stimuli. Thresholds of all colour normal (CN) participants are plotted as coloured circles in all panels as references, and thresholds of each CVD observer are denoted by black crosses in separate panels. The anomaloscope diagnosis for each participant is indicated.

Figure 5.

Classification of colour normal (CN) and colour vision deficient (CVD) results. (a) Step one clustering results illustrated in LMS detection threshold space. All 37 individual data points are shown. CVD and CN individuals are represented by red and black asterisks, respectively. Four clusters, denoted in differently shaped and coloured symbols (green circles, blue diamonds, cyan squares and orange triangles) around the asterisks, are identified. (b)-(f) show step two clustering results. Only individual points surrounded by green circles (n=28) in (a) are plotted. These thresholds were clustered as four groups denoted by numbers. Red squares and green circles represent CVD and CN, respectively.