# Ishihara Color Vision Test

Free online color vision deficiency (CVD) screening based on the pseudoisochromatic plate method originally published by Shinobu Ishihara in 1917. The online version at ishiharatest.com runs entirely in the browser, produces a numeric score and a plain-language interpretation, and is available in 13 languages.

## What the test does

The user is shown a sequence of 15 pseudoisochromatic plates, each containing a number or pattern formed by colored dots that is visible to observers with normal color vision but hard or impossible to read for observers with the relevant form of color vision deficiency. After answering all plates the user receives:

- A score (correct answers / total) with a simple verdict (normal, mild anomaly, likely CVD).
- A short explanation of the likely CVD type (red-green, blue-yellow or monochromacy) in the user's language.
- An optional downloadable PDF report with the score and explanation.

The test is not a medical diagnosis. Limitations include uncalibrated consumer displays (sRGB vs wide gamut), ambient lighting, screen size, device color profile, and the small plate count (15) compared with the full Ishihara booklet (38). The test is intended for preliminary self-screening and education.

## How the plates are generated

Plates are generated by an algorithm rather than scanned from copyrighted books. Key parameters:

- Background and foreground dots share luminance but lie along a confusion line in CIELAB space aligned with the relevant CVD axis (protan, deutan or tritan).
- Dot radius varies in a bounded range to break edge cues that could leak shape information to CVD observers.
- The target shape (digit or path) is rendered into a raster mask; dots inside the mask take foreground color, dots outside take background.
- Placement uses Poisson-disk-like sampling to avoid grid artifacts.

Source of color models: standard CIE data and published pseudoisochromatic plate literature (Ishihara S., 1917; Birch J., 2006; Dain S., 2004). The algorithm is tuned so a trichromat observer can read each plate with high confidence while a simulated protan/deutan observer (Brettel-Vienot-Mollon 1997 model) fails to distinguish target from background.

## Validation

Validation against a laboratory HRR-style screening on a panel of 178 adults yielded overall agreement of 92% with the clinical result. Per-observer-class accuracy:

- Trichromat: 96% (correctly classified as normal).
- Deuteranope: 89% (correctly flagged).
- Protanope: 91% (correctly flagged).
- Tritan defects: lower sensitivity, noted in the research page.

Disclaimer: a consumer display validation is not equivalent to clinical practice. Users with a positive or uncertain result are referred to a qualified optometrist or ophthalmologist for anomaloscope or Farnsworth-Munsell testing.

## Languages

Thirteen language variants with properly localized UI, result text, PDF output and meta tags: en-US, en-GB, de, fr, es, it, pl, cs, sk, pt-BR, hi, th, tr. Each language has its own canonical URL under /LANGUAGE/, with reciprocal hreflang alternates and matching structured data.

## Embedding

The test can be embedded on third-party sites via an iframe widget (https://ishiharatest.com/en-us/share/). The widget supports selecting which test mode is available (casual, screening, or both) via a URL parameter and is used by schools, clinics and HR teams.

## Privacy

No account required, no login, no email collected. An anonymous session identifier is used to deduplicate impressions and to retrieve the result for PDF export within the same browser session. Aggregated statistics (question-level difficulty, language distribution) are used to improve plate selection. No personal identifiers are stored.

## Organization

- Operator: Websy s.r.o., Slovakia.
- Contact: tompokryvka@gmail.com.
- Source of truth for plate generation and validation: https://ishiharatest.com/en-us/research/.
- Repository: private (internal).

## FAQ highlights

Q: Is this a replacement for a professional color vision exam?
A: No. It is a screening tool. A positive result should be confirmed by an eye-care professional using calibrated hardware.

Q: Why only 15 plates?
A: Balance between user completion rate and diagnostic confidence. The plate selection covers the main CVD axes (protan, deutan, tritan) with redundancy.

Q: Why pseudoisochromatic plates at all, given modern alternatives?
A: They are fast, familiar to clinicians, and well-studied. Our algorithm avoids reproducing copyrighted plates while keeping the diagnostic principle.