Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024;17(3):1-35.
doi: 10.1145/3694790.

International Guidelines for Photosensitive Epilepsy: Gap Analysis and Recommendations

Affiliations

International Guidelines for Photosensitive Epilepsy: Gap Analysis and Recommendations

J Bern Jordan et al. ACM Trans Access Comput. 2024.

Abstract

People with photosensitive epilepsy may have seizures caused by flashing lights, patterns, and video sequences. Because of this, there is increasing interest among researchers, developers, and companies toward making content safer. There are five major guidelines (from the U.K., Japan, ISO, ITU, and W3C) to limit risk in different technology domains that have been created since the mid-1990s. All address similar risk factors, but they are not fully harmonized and can be confusing. Furthermore, there have been significant changes in technology since the guidelines were introduced. This article compares and clarifies the guidelines, describes risk factors that have changed (such as the reduction of risk due to display technology), gaps in our knowledge, the coverage of new technology, and new characteristics (such as the duration and synchronicity of individual flash transitions) that may need to be accounted for. The authors suggest working values for new thresholds and propose updated guidelines.

Keywords: flash; flicker; games; seizure; standards; television; video.

PubMed Disclaimer

Figures

Fig. A1.
Fig. A1.
The relationship between visual angle (2θ), peripheral angle (θ), viewing distance (d) and object size (x).
Fig. 1.
Fig. 1.
In most guidelines, luminance changes of 20 cd/m2 or more are potentially hazardous (corresponding to the grayscale blocks on the axis). Also shown are two frames of a flash of lightning that differ more than 100 cd/m2 on average. Stills from video by The Element [Pexels reuse license], via Pexels (https://www.pexels.com/video/thunder-and-flash-of-lightning-2657691/ ).
Fig. 2.
Fig. 2.
The CIE 1976 UCS chromaticity diagram showing the sRGB gamut (large triangle) and values of select red colors (marked with plus “+”) that differ enough in WCAG 2.0 critical values from pure red (#FF0000, in the upper right, red corner of the gamut triangle) to be considered a potentially hazardous transition. The harmonized ISO and WCAG 2.2 critical difference is represented by the dotted arc. Picture adapted from original by Adoniscik [Public domain], via Wikimedia Commons (https://commons.wikimedia.org/wiki/File:CIE_1976_UCS.png ).
Fig. 3.
Fig. 3.
The percentage of people (n = 170) with PSE responding with PPRs at different flash rates. Only 3% had PPRs at 3 Hz. The response rate quickly goes up to nearly 90% responding at peak of 16 Hz followed by a decrease in people responding as the frequency increases beyond the peak. At 65 Hz, only 4% of people responded. Data from [42].
Fig. 4.
Fig. 4.
An illustration of the area threshold (shaded red) on televisions and computer displays, which are viewed at closer distances. The threshold area is 0.006 steradians, which is a quarter of the central 10° field of view (dashed circles). This is approximately 25% of a television screen at typical viewing distances. Note that flashing areas do not need to be contiguous, or the specific shape shown here. Device illustrations from Twitter Twemoji [CC-BY-4.0 license] (https://github.com/twitter/twemoji/ ).
Fig. 5.
Fig. 5.
Spectra of the three primary colors (blue, green, and red) on three screens with different display technologies (CRT, LCD, and OLED). Most notable are the two red peaks of the CRT at 625 and 704 nm. Neither of the other screens output light with peak wavelengths that long. Figure adapted from [71] (used with permission).
Fig. 6.
Fig. 6.
A target-like pattern that would just fail WCAG if it were to flash to white more than 3 times/s, and a person was focused on the center or more peripherally toward the annulus. This figure has a 5°-diameter circle in the middle (with a solid angle area of 0.006 steradians) and an annulus with an inner diameter of 15° and outer diameter of 16.6° (0.012 steradians). Overlaid in a dotted line is a representative, off-center 10° field of view. Laptop illustration from Twitter Twemoji [CC-BY-4.0 license] (https://github.com/twitter/twemoji/ ).
Fig. 7.
Fig. 7.
Illustration of balanced and unbalanced flashing patterns. The balanced pattern here is a black-and-white checkerboard that reverses black and white blocks with each frame. Balanced flashing patterns (and white noise) with small enough blocks pass WCAG because of an exception. The unbalanced pattern here shows a first frame made up of a gray checkerboard with brighter blocks are nearly 40 cd/m2 brighter than the darker blocks. The second frame is solid gray, which is the same gray as the light gray color of the first frame’s checkerboard. The dark squares are therefore flashing on and off and cover 50% of the screen with a luminance difference exceeding the 20 cd/m2 threshold. When viewed at enough distance, the difference in the average luminance is less than the threshold. However, this unbalanced pattern would not pass WCAG.
Fig. 8.
Fig. 8.
Two waveforms at 60 fps and 30 fps showing transitions that evolve over one or more frames. The 60 ms time windows are represented by the gray lines below each peak. At 60 fps, peaks labeled (A)–(D) take less than 60 ms to exceed the luminance threshold over 1–4 frames, respectively. In (E), it takes longer than 60 ms to exceed the luminance threshold (5 frames at 60 fps), thus there is no up transition. In (F), the waveform is not monotonically increasing, but the luminance threshold is exceeded in less than 60 ms, thus it counts as an up transition over the 4 frames. At 30 fps, a potentially hazardous transition can take 1 or 2 frames (which takes less than 60 ms) to exceed the threshold.
Fig. 9.
Fig. 9.
Luminance waveforms at 60 fps for two areas of the screen that exceed the area threshold in sum, but not separately. The synchronicity time window of 20 ms is represented by the gray lines below the waveforms. In (A) and (B), the leading edges of the two areas are within 20 ms, thus the areas are summed. In (C), the leading edges are more than 20 ms apart, thus are not summed. (D) shows Area #2 flashing twice within a single flash of Area #1, which counts as two transitions where the areas transition in the same direction simultaneously. (E) shows Area #1 undergoing a two-frame luminance ramp that exceeds the luminance difference threshold of 20 cd/m2 within 20 ms of the leading edge of Area #2, thus the areas are summed.

References

    1. Ackerman Evan. 2020. Bosch gets smartglasses right with tiny eyeball lasers - IEEE spectrum. IEEE Spectrum. Retrieved July 17, 2024 from https://spectrum.ieee.org/bosch-ar-smartglasses-tiny-eyeball-lasers
    1. Aker Jackie. 2019. Epilepsy Foundation Files Criminal Complaint and Requests Investigation in Response to Attacks on Twitter Feed. Retrieved July 17, 2024 from https://www.epilepsy.com/stories/epilepsy-foundation-files-criminal-comp...
    1. Alzubaidi Mohammad A., Otoom Mwaffaq, and Al-Tamimi Abdel-Karim. 2016. Parallel scheme for real-time detection of photosensitive seizures. Comput. Biol. Med. 70, (March 2016), 139–147. DOI: 10.1016/j.compbiomed.2016.01.008 - DOI - PubMed
    1. Ando Soichi, Kida Noriyuki, and Oda Shingo. 2002. Practice effects on reaction time for peripheral and central visual fields. Percept. Mot. Skills 95, 3 (December 2002), 747–751. DOI: 10.2466/pms.2002.95.3.747 - DOI - PubMed
    1. Anyanwu E. 1997. Mental imagery of photic stimulation provokes paroxysmal EEG activity in a photosensitive patient who self-induces seizures. Ital. J. Neurol. Sci. 18, 2 (April 1997), 93–100. DOI: 10.1007/BF01999569 - DOI - PubMed

LinkOut - more resources