Optimizing the detection and characterization of bruises using multispectral imaging

Abstract

The detection and visualization of sub-dermal hematoma (bruises) plays a key role in suspected physical abuse cases, as it aids in the evaluation of both victim and suspect statements. Current methods rely on visual inspection, frequently aided by alternate light sources (ALS). Ideally, ALS increase visual contrast by exploiting differences in light absorption (due to the formation and clearance of chromophores within the bruise). However, in practice the achievable contrast is often limited by light-scattering: the short-wavelength region of the spectrum (comprising most of the chromophore-specific absorption peaks), is also strongly scattered by the dermal tissue. This, in turn, limits achievable penetration depths, effectively obscuring deep-lying bruises. ALS-based contrast enhancement is further complicated by bruise healing; diffusion and enzymatic activity alter the chromophore concentrations as well as their 3D-distribution within the tissue. To overcome these critical limitations, we employ a multi-spectral camera (8 wavelengths simultaneously) in conjunction with both observer-based scoring and a contrast-quantification algorithm to determine the optimal wavelength for the detection and characterization of bruises over time. We show that (i) bruise contrast significantly increases at 480 nm, 620 nm and 850 nm and (ii) the wavelength achieving optimal contrast gradually changes from 850 nm to 578 nm–480 nm as the bruise heals.