Acquisition-parameter effects on image utility in near-infrared dorsal hand imaging: linking image signatures to repeatability and separability

Abstract

Near-infrared dorsal hand imaging is often evaluated through downstream recognition, while acquisition settings are treated as fixed experimental details. This study examined whether controllable acquisition parameters alter objective image-derived signatures and protocol-level utility before recognition modelling. Thirty-four participants were imaged using a repeated-acquisition matrix with two illumination conditions, two contrast/current set-points, four automatic white-balance gain configurations, two hand sides, and four repeats. Standardized dorsal hand regions of interest were summarized at the participant-protocol level. Automatic white-balance gain showed the strongest retained effects, particularly for dynamic range (partial eta-squared 0.633), verification-oriented evidence (0.603), and the Protocol Image Utility Score (0.425). The top-ranked setting was dark illumination, c60 contrast/current, and an automatic white-balance gain configuration with red-channel gain 2.2 and blue-channel gain 1.6, yielding a PIUS of 0.721. Sensitivity analyses showed that this ranking was not driven by a single weighting choice or utility component.