An Open-Source, Excitation-Resolved UV-NIR Imaging Platform for Bioluminescence, Luminescence-Decay, and Chemical Discrimination Measurements

Abstract Multispectral imaging is a cornerstone of chemical imaging, with bioluminescence, fluorescence, and phosphorescence imaging underpining most of preclinical and analytical chemical measurements. However, conventional implementations are typically proprietary, costly, and resolve spectral content from the reflectance of a broadband source, without control over the excitation spectrum. This architecture cannot isolate excitation-dependent photophysical processes. Spectral discrimination through sequentially resolved narrowband excitation, detected on a single broadband-sensitive camera, instead enables excitation-dependent fluorescence, phosphorescence, and reflectance measurements not available to such broadband approaches. We describe AURORA-MSI, an open-source multispectral platform that inverts this arrangement with 14 narrowband light-emitting diodes (LEDs) spanning 367–940 nm, a 120-element annular illumination ring with eight independently addressable azimuthal sectors, and a thermoelectrically cooled monochrome complementary metal-oxide-semiconductor (CMOS) camera. Radiometric calibration, spatial uniformity mapping, and camera noise characterization establish the quantitative measurement foundation. In bioluminescence imaging, AURORA-MSI localized sources in a calibrated tissue-mimicking mouse phantom comparably to a commercial Revvity In Vivo Imaging System (IVIS) Spectrum, detected luciferase-expressing HSJD-GBM1-001 glioblastoma cells across a dilution series, with reduced replicate consistency at the lowest densities, and mapped substrate-free fungal-pathway emission in an intact bioluminescent Petunia hybrida, none requiring photon-counting instrumentation. Time-resolved phosphorescence decay imaging of six inorganic phosphors over 33 min found triexponential models adequate at early times, while distributed-lifetime models were preferred at extended time scales. Multispectral image-texture features extracted across all 14 excitation bands discriminated ten pharmaceutical and household powders spanning distinct chemical compositions and, in two cases, distinct formulations of the same compound. Generality beyond these regimes was established through dual-excitation fluorescence fingerprinting of 15 mineral specimens and wavelength-selective plant-tissue imaging exploiting UV and NIR penetration-depth differences. Excitation-side spectral encoding enables photophysical measurements that detection-side systems with uncontrolled broadband illumination cannot isolate, and the cooled-camera architecture supports weak-signal modalities without photon-counting instrumentation.