High-speed hyperspectral single-pixel microscopy via line-scan detection with data fusion-based enhanced resolution

The development of hyperspectral microscopy techniques is recognized as an active field of research. Although commercial alternatives are available, these systems remain slow, expensive, and limited in flexibility, indicating clear opportunities for improvement. Coded‑imaging strategies, including single‑pixel imaging (SPI), offer promising alternatives by enabling computational reconstructions with spectral resolution through parallel detection. When combined with conventional cameras via data fusion (DF), these reconstructions can achieve high spatial resolution. However, traditional implementations often rely on spectrometers, which suffer from limited bandwidth and low-speeds acquisition, restricting SPI’s versatility. To overcome these challenges, in this work is developed a hyperspectral single‑pixel microscopy platform (HySPM) assisted by DF. The system employs a customized optical setup that disperses light into spectral components, captured by a high‑speed line‑scan camera in parallel detection. Full Hadamard scans of 64×64 pixels are completed in 0.82 s at 10 kHz. Up to 73 spectral bands are acquired in parallel with 4 nm resolution, without compressive algorithms. Performance is validated through reflected intensity and fluorescence imaging. DF further integrates HySPM with high‑resolution monochrome images, expanding reconstructions from 64×64 to 605×605 pixels while enhancing spatial detail and preserving spectral signatures. HySPM thus provides a scalable, cost‑effective solution that bridges temporal limitations in current hyperspectral implementations. Zapata-Valencia and colleagues report a high-speed hyperspectral microscopy system that combines single-pixel detection with data fusion to enhance resolution. This approach enables rapid, detailed imaging for scientific and biomedical applications.