Newtone Technologies, 13 bis place Jules Ferry, Lyon, F-69006, France.
DSM Nutritional Products, Personal Care & Aroma, Wurmisweg 576, Kaiseraugst, CH-4303, Switzerland.
Int J Cosmet Sci. 2020 Feb;42(1):99-107. doi: 10.1111/ics.12593. Epub 2019 Dec 25.
There are methods to evaluate skin colour on defined areas over the face but no approach automatically and accurately evaluates skin colour variations on large facial areas, comparing subjects, treatments and/or time points. We propose such an image-based approach to visualize quickly the outcome of clinical studies on colour variations.
Among 54 Asian women, one group applied a vehicle twice daily, during 28 days, and the other group an anti-ageing emulsion, taking facial images at baseline and after treatment. Changes in Lab* values were studied on four pre-selected facial regions. We also reconstructed average facial images from which the Lab* parameters were extracted for every pixel, computing relevance (ΔE) and significance data. Using colour gradients, we mapped these results onto the average facial images.
After treatment, Lab* parameters show no statistically relevant colour changes in the vehicle group. In the 'active' group, skin was lighter at the upper cheek and, overall, redness decreased. Relevance and significance maps confirmed no visible colour changes in the vehicle group. In the 'active' group, the mapping approach revealed colour changes and their location. Skin became lighter below the eye, cheek and forehead. It was less red below the eyes, on the cheek, jawline and forehead, and generally more yellow.
Our image-based mapping approach proves to be powerful. It enables us to identify precise facial regions of relevant and statistically significant colour changes after a topical treatment, regions that would have otherwise been undetected.
有一些方法可以评估面部特定区域的肤色，但没有一种方法可以自动且准确地评估大面积面部区域的肤色变化，比较受试者、治疗方法和/或时间点。我们提出了一种基于图像的方法，以便快速可视化临床研究中关于颜色变化的结果。
在 54 名亚洲女性中，一组每天两次使用载体，持续 28 天，另一组使用抗衰老乳液，在基线和治疗后拍摄面部图像。研究了四个预先选择的面部区域的 Lab值变化。我们还从平均面部图像中重建了平均面部图像，从中提取了每个像素的 Lab参数，计算了相关性（ΔE）和显著性数据。使用颜色梯度，我们将这些结果映射到平均面部图像上。
治疗后，载体组的 Lab*参数没有显示出统计学上相关的颜色变化。在“活性”组中，脸颊上部的皮肤更亮，整体上红色度降低。相关性和显著性地图证实载体组没有可见的颜色变化。在“活性”组中，映射方法揭示了颜色变化及其位置。眼睛下方、脸颊、额头的皮肤变浅，眼睛下方、脸颊、下颌线和额头的皮肤红色度降低，整体上黄色度增加。
我们的基于图像的映射方法证明是强大的。它使我们能够识别出局部治疗后具有相关性和统计学意义的肤色变化的精确面部区域，这些区域否则可能无法检测到。
