COLOR RETENTION AND BLEEDING POTENTIAL OF A LIPSTICK 401 to eat, and refrain from excessively licking/touching their lips during the day. Eight hours after application of the lip color, the participants returned to the laboratory to have their images taken for the fi nal 8-h (H08) time point. QUANTIF ICATION OF LIP PROPERTIES BY IMAGE ANALYSIS Image a nalysis was performed on the clinical images using ImageJ, a freeware created and distributed online by the National Institutes of Health. Algorithms and analysis programs were developed in our laboratory using JavaScript. Steps involved in the analysis included (i) color correcting the images, (ii) outlining the lip area, (iii) measuring color intensity, and (iv) calculating lip color properties, namely, color retention and bleeding potential. To carry out the aforementioned analysis, each clinical image was fi rst color-corrected against the embed- ded standard color chip using the least-square method described elsewhere (22). Lip area was detected using our in-house developed, automated algorithms to minimize inconsistent measurement results introduced by manual processing. Color intensity of the lips was mea- sured from the detected lip area using L*, a*, and b* in CIE-LAB color space. Color reten- tion was derived from the L*a*b* values based on the defi nition shown by equation (1): % % q100%, %E IMM BSL IMM IMM BSL Color retention, % T E E (1) w here % 2 2 2 * * * * * 1 2 1 2 1 2 E L L a a b b* , in which the subscripts 1 and 2 are any time points of interest. Subscript BSL = baseline, IMM = immediate after lipstick application, and T = any time point at which a lipstick is evaluated for color performance. In this study, T = H08, 8 h after wearing the liquid lip color. Some samp le images with lip areas outlined for color retention analysis are shown in Figure 1A and C. Color ble eding describes a phenomenon that occurs when the applied lip color spreads outside the boundaries of the lips, typically through wrinkles and lip lines. Examining the phenomenon, we defi ned a parameter of lip color bleeding as the ratio of line rough- ness of the detected lip border at time point T (which can be any time point after IMM, e.g., in this study, T = H08) over that at IMM, together with the infl uence of its corre- sponding color retention, as shown by equation (2). qColor IMM IMM IMM Line Roughness Bleeding Retention Line Roughness . T T T (2) Sam ple im ages of detected lip border before and after color bleeding are displayed in Figure 2A and C. PANEL PERC EPTION Consumer r elevance of the color retention and bleeding potential was obtained from two separate perception studies in Shanghai, China, as well as in the United States. Color

JOURNAL OF COSMETIC SCIENCE 402 retention acceptability was determined using seven pairs of lip images from the IMM and H08 time points with varying levels of retention. Eighty-four panelists were shown the images and instructed to select the pairs of images exhibiting “acceptable” color retention Figure 1. Illustration of outlined lip areas for color retention analysis. (A) Lips of a volunteer at the BSL time point, (B) lips at the IMM time point, and (C) lips at the H08 time point. Color parameters L*, a*, b*, and 'E were measured, and color retention were calculated from those outlined regions of interest.