Systems and processes for automated detection and characterization of skin features using machine learning

Abstract

Aspects of the present disclosure generally relate to systems and processes for automated skin feature detection and analysis, and more particularly to identifying and assessing the characteristics and age of skin contusions, bruises, and other similar skin features using a deep learning model. Images of a patient's body part are captured via a digital camera with a light source. The parameters of the light source are determined, and other information is gathered, including skin color and medical history of the patient/subject. Images of an injured area are combined with contextual clinical information as inputs to a machine learning model. The machine learning model detects and estimates characteristics of skin features. In an exemplary embodiment, the machine learning model detects bruises on the skin, and estimates their age and outputs the estimate to an end user along with a measurement of the uncertainty in the estimate.