Unifying the spectrum: a framework for the meta-analysis of human skin tone scales

Quantification of human skin tone is fundamental to dermatology, medical imaging, computer vision, and social studies. However, the field currently lacks a standardized framework, relying on fragmented methodologies, from the Fitzpatrick skin phototype (FST) to the Monk skin tone (MST) scale. Even nowadays, this creates barriers to scientific progress and introduces systemic errors and bias in medical devices and AI fairness.

The aim of this study is to establish a meta-analytic framework, termed a "scale of scales," to critically assess and quantitatively compare the interoperability and accuracy of prevailing skin tone quantification systems.

We utilized the International Skin Spectra Archive (ISSA), composed of 15,256 spectral reflectance measurements from a globally diverse population, as a biological baseline. We overlaid the categories of major verbal, physical, and digital scales onto this dataset to evaluate their coverage and fidelity.

Our analysis reveals that widely used scales such as FST and MST fail to capture the continuous, multidimensional nature of human pigmentation. Furthermore, physical targets such as the Pantone skintone guide exhibit significant metameric failure, rendering them unreliable under varying illumination.

Current subjective and categorical proxies are insufficient for robust scientific application. The field must transition toward a unified, data-driven framework rooted in spectral reflectance data to ensure interoperability and inclusivity in medical and technical applications.