Science Behind the Project

The innovation behind EAS-ID isn’t one technology. It’s three separate innovations combined together to create an entirely new way to assess injuries. It combines Alternate Light Source (ALS) imaging, advanced health informatics, and deep learning into a new forensic examination technique.

Alternate Light Source (ALS)

When an observer looks at an object, the color perceived is dependent on the wavelengths of the reflected light from the object’s surface. The skin is often examined using white light, which includes wavelengths across the visible spectrum (e.g., overhead or focused lighting, sunlight). The detection and characteristics of the bruise are then defined by the light reflected from the skin’s exterior.

The solution is to control the nature of the light source through ALS imaging. How human skin interacts with light depends on both the wavelength (color) of the light and properties of skin itself. ALS works by leveraging how blue and violent wavelengths of light interact with the hemoglobin that comprises a bruise to make injuries visible despite melanin concentrations.

However, some of the light’s wavelengths can penetrate the skin’s surface to deeper layers where it is absorbed by certain molecules associated with skin pigmentation and trauma, including blood and its breakdown products. Unfortunately, the reflected white light can overpower the observer’s ability to see light absorption from bruising, which appears dark compared to the surrounding skin. This is particularly evident when the skin pigmentation is dark.  

By limiting the light emitted during the examination of bruises to certain wavelengths (e.g., colors) known to be absorbed by blood components, we can lessen the amount of reflected light viewed by the observer. Such technology, called alternate light, is commonly used in forensic science to identify physical evidence that is difficult to see. Recent scientific advancements have further supported the use of this technology to enhance the visualization of subtle injuries, specifically bruises.


Multiple data sources, multimedia data, and several applicable data standards make data integration and processing a complex task. Health informatics is about gathering, processing, and analyzing health information, and provides us with a set of tools critical to the success of this work.

Our data platform is designed to integrate injury imaging with data from sources like Electronic Health Records and injury reports, and self-reported conditions. It also allows for the processing of precise injury measurements collected in a controlled laboratory environment, with retrospective data from healthcare providers, data collected from enrolled patients, and data coming from mobile data collection tools.

Behind the informatics solution is a common data structure, modeled after the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), heavily relying on an ontology of concepts and standardized terminologies (SNOMED, ICD, CPT).

The data platform is created with security at its core. Data separation and multiple levels of access control help us protect participant data from unauthorized access, even though no personal identifiers are included.

Computer Vision and Deep Learning (DL)

When people talk about artificial intelligence (AI) today, they are typically referring to the field of deep machine learning. This class of tools are based on building massive artificial neural networks – collections of interconnected artificial neurons designed to mimic the human brain. Such networks are very good at performing detection and analysis tasks across a vast range of applications. Computer vision in this context refers to the design of deep neural networks that can process images as neural network inputs. The EAS-ID system uses a deep network to take in images of bruises (including ALS images), combine them with critical health informatics information, and ultimately provide assessment of a victim’s injuries.