Purchase this article with an account.
Dingcai Cao, Pablo Alejandro Barrionuevo; Estimating photopigment excitations from color sensor outputs of a personal light exposure device. Invest. Ophthalmol. Vis. Sci. 2014;55(13):2361.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The human circadian clock is regulated by light exposure. To fully understand how daily light exposure contributes to circadian rhythm, it will be helpful to know the relative photopigment (rod pigment, cone pigment and melanopsin) excitations of light exposures. The ActiWatch Spectrum (Phillips Respironics) is a commercially available device to record personal exposure to light. The purpose of the current study was to estimate photopigment excitation from the color-sensor outputs.
The spectral sensitivity functions of the red (R), green (G), blue (B) sensors were obtained from a previous published work (Price et al, Light Research and Technology, 44:17-26, 2012). The outputs of the R, G, B sensors were computed under 64 illuminants (27 daylight illuminants, 27 fluorescent lights, 5 high pressure lights, and 5 LED-based lights; correlated color temperature CCT: 2000K-10000K). The excitations of rod and cone (S-, M-, and L-) photoreceptors and melanopsin were computed for each illuminant using human corneal spectral sensitivity functions. The predictability of photopigment excitations and CCT from R, G, B outputs was assessed using a regression method.
For all of the illuminants, a linear combination of R, G, B outputs could predict the excitations well for S-cones (R2 = 97.8%), rods (R2 = 97.9%) and melanopsin (R2 = 96.5%). The predictability from R, G, B outputs became relatively poor for L-cones (R2 = 89.1%), M-cones (R2 = 84.5%) or luminance (R2 = 88.1%), particularly for three-band fluorescent lights due to low ActiWatch R and G sensor sensitivities between 570nm -590nm and spikes in the fluorescent light spectrum. A separate function for three-band fluorescent lights was needed to improve the fits of L- or M- cone excitations. Meanwhile, a second order polynomial function combining R, G, B outputs predicted correlated color temperature satisfactorily for all illuminant types (R2 = 96.8%). Finally, R, G, B outputs could classify illuminant types (fluorescent vs. daylight illuminants) satisfactorily (sensitivity = 85%, specificity = 84%).
The R, G, B outputs produce a reasonable estimate of photopigment (S-cone, M-cone, L-cone, rod and melanopsin) excitations, but separate weightings are required for disparate illuminant types. R, G, B outputs may also be useful for classifying illuminant types.
This PDF is available to Subscribers Only