Full spectrum fluorescence lifetime imaging with 0.5 nm spectral and 50 ps temporal resolution

Optical microscopy techniques are increasingly used for rapid analysis and diagnostics across diverse samples. The demand for highly detailed spatial, spectral, and temporal optical signatures has spurred the development of new microscopy technologies. This study addresses the limitations of existing full-spectral fluorescence lifetime imaging microscopy (FS-FLIM) systems, which typically have a small number of parallel spectral channels, restricting the complexity of discernible signals and hindering the detection of subtle changes. The researchers aim to overcome this limitation by developing a high-speed, high-resolution FS-FLIM system that captures full emission spectra and lifetime data simultaneously for each pixel in a single acquisition. This approach is expected to significantly enhance the ability to unmix signals from complex fluorescent samples, where multiple fluorophores interact, leading to subtle shifts in emission spectra and lifetimes due to factors like resonant energy transfer, pH, viscosity, temperature, and quenching pathways. The development of such a system is crucial for monitoring subtle spectral or lifetime changes in complex environments like biological tissues, where these variations can reflect important biological processes such as those found in cancerous tissues. The researchers aim to demonstrate the capabilities of their new system through experiments on various samples, including freshly resected human lung tissue, showcasing its versatility and potential in diverse applications.

High-resolution time-resolved emission spectroscopy (TRES) enhances the unmixing of signals from complex fluorescent samples. Simultaneous acquisition of full emission spectra and lifetime datasets enables complete exploitation of fluorescence signals, revealing small changes in emission profiles. However, existing FS-FLIM systems have been limited by data acquisition speed and volume, restricting the number of parallel spectral channels. This limitation has hindered the ability to distinguish complex signals and monitor subtle changes resulting from environmental factors or interactions between multiple fluorophores. Previous systems often used line arrays of SPADs, but lacked the on-chip histogramming capabilities and integrated electronics of the system presented in this paper. Commercial systems like the Leica Stellaris achieve higher spectral resolution through sequential image captures, but this approach slows acquisition, introduces photobleaching and motion artifacts. The current work addresses these limitations by aiming to acquire the entire wavelength-lifetime spectrum for each pixel in a single pass, enabling high-speed data acquisition with significantly increased spectral and temporal resolution.

The researchers developed an achromatic, confocal laser scanning FS-FLIM system using a custom 512-channel time-correlated single-photon counting (TCSPC) sensor with on-chip histogramming. The sensor comprises 16 single-photon avalanche diode (SPAD) detectors per channel, enabling simultaneous lifetime measurements in each 0.5 nm spectral band. The system uses a fully reflective optical path (except for the final objective) to minimize chromatic aberration, achieving diffraction-limited performance from 400-900 nm. A supercontinuum white light laser, filtered by a tunable acousto-optic filter, is used for excitation. Two galvanometer mirrors scan the beam, and the fluorescence is de-scanned and directed through a spectrometer with a volume phase holographic grating onto the sensor. All photon timing electronics are integrated on-chip, simplifying the system and enhancing robustness. A Field Programmable Gate Array (FPGA) on the sensor PCB controls the sensor, acquires data, and directly controls the optical scanning system. The system was controlled using custom Matlab scripts for image assembly, processing, and display. Least squares fitting was used to process lifetime calculations due to its computational efficiency. A threshold (10 times background noise) was applied to select fluorescence events for lifetime calculations. For samples with low photon counts at the spectral edges, a moving spectral averaging filter (8 pixels) was applied to improve the signal-to-noise ratio. The samples analyzed included *Convallaria majalis* labeled with Safranin and Fast Green, a honeybee wing, and freshly resected human lung tissue (both stained and unstained slices). The system uses a variety of image processing and analysis techniques, such as color mapping and intensity-weighted transparency adjustments, to visualize the complex spectral and lifetime information within the resultant data cubes.

The developed FS-FLIM system achieved unprecedented speed and resolution, capturing 256x256 pixel images with 512 wavelength channels and 32 time bins in 6 seconds, with exposure times of 85 µs/pixel. The system demonstrated diffraction-limited performance, 50 ps time resolution, and 0.5 nm spectral resolution. Imaging of *Convallaria majalis* labeled with two fluorophores clearly distinguished the different spectral and temporal responses of the fluorophores. Analysis of a honeybee wing showed subtle autofluorescence spectral variations and significant spatial lifetime variation with strong spectral dependence, highlighting the ability to detect small changes in emission which would have been missed by systems with fewer spectral channels. The system was applied to ex vivo human lung tissue samples, both stained with H&E and unstained, directly revealing spectral and lifetime differences between cancerous and healthy regions. The stained samples showed a clear reduction in Eosin lifetime in cancerous regions, likely due to factors such as self-quenching and variations in pH and viscosity. Unstained tissue showed pronounced lifetime differences between cancerous (amorphous) and healthy (alveolar) regions, with longer lifetimes observed in the healthy tissue. In both stained and unstained slices, the cancerous regions exhibited a red-shifted emission. Imaging of fresh, unprocessed lung tissue further demonstrated the system's ability to distinguish cancerous and healthy regions based on spectral and lifetime variations. The results demonstrate the potential of FS-FLIM to provide optical fingerprints for distinguishing tissue types in fresh, unprocessed samples.

The results demonstrate the high sensitivity, versatility, and robustness of the developed FS-FLIM system for rapid data acquisition from diverse samples. The ability to capture the entire fluorescence lifetime spectral data cube opens opportunities for advanced applications, including full-spectral lifetime Förster resonance energy transfer (FRET) imaging, simultaneous fluorescence and Raman imaging, and multi-fluorophore analysis. The observed differences in spectral and lifetime characteristics between cancerous and healthy lung tissue regions suggest the potential for using FS-FLIM as a powerful tool in histopathology, potentially streamlining the pathology pathway by reducing reliance on traditional tissue staining and improving the speed and accuracy of cancer diagnosis. While further research is needed to fully validate the clinical potential of this technique, this study showcases its remarkable ability to reveal subtle, clinically relevant differences in the optical signatures of various tissues. The observed differences in the optical signatures between cancerous and healthy tissues highlight the potential for using FS-FLIM as a tool for in vivo real-time pathology.

This study successfully demonstrated a highly sensitive and versatile full-spectral fluorescence lifetime imaging microscopy (FS-FLIM) system capable of rapid data acquisition with high spectral and temporal resolution. The system's unique capabilities, including the acquisition of complete spectral and lifetime data cubes, opens avenues for advanced applications in various fields, particularly in the life sciences. The observed distinctions between cancerous and healthy tissue highlight the potential of FS-FLIM for rapid, label-free histopathological analysis, promising significant advancements in diagnostics. Future work will focus on optimizing the data transfer to achieve real-time imaging and exploring the application of the system to high frame rate imaging using dimensional binning and advanced analysis techniques.

While this study showcases the significant potential of the developed FS-FLIM system, some limitations should be noted. The current data transfer bottleneck via USB 3 connection limits the sequential imaging speed. The least-squares fitting method used for lifetime calculations assumes a single exponential decay, which may not be entirely accurate in regions with multiple emitting species. Additionally, the detailed analysis of the biological processes underlying the observed optical signatures remains a focus for future research. The small number of samples limits generalizability and requires validation in larger clinical studies before translation to clinical practice.