Análise do fotótipo cutâneo através de sensoriamento óptico e aprendizado de máquina

Abstract

Visual methods are often used to subjectively classify human skin photo type. However, with advances in artificial intelligence technology, methods are emerging to improve medical diagnoses. The use of artificial intelligence to improve diagnostic medical care is a rapidly growing area of research, and this work presents a new perspective for classifying phototype using a simple color sensor and neural network. Melanin, a critical protein for protection against ultraviolet radiation, is the main determinant in defining skin phototype. Several methods can classify melanin concentration, such as clinical methodologies, visual comparisons and regional common sense. However, the Fitzpatrick Scale is widely used and classifies melanin concentration levels. The objective of this study is to develop a phototype classifier approach that can assist several medical areas, including cosmetics, dermatology, photobiomodulation and tattoo removal. The process used in this study used RGB data obtained from the color sensor reading, which was sent to a neural network built in KNIME. By analyzing the RGB color channels, it was revealed that the green and blue regions of the spectrum are key to skin color identification, resulting in an overall classification accuracy of 91%. The integration of the color sensor with artificial intelligence proved to be a tool, allowing independent readings of ambient lighting and insights into the patient's health. The research also overcame recruitment challenges and demonstrated the relevance of color sensors over traditional cameras, highlighting possible applications in the medical and cosmetic areas and the potential to enrich medical practice with artificial intelligence technologies.