Segmentation-guided photon pooling enables robust single-cell analysis and fast fluorescence lifetime imaging microscopy.

Significance: Fluorescence lifetime imaging microscopy (FLIM) can probe the metabolic environment of living cells in a label-free and noninvasive manner. However, endogenous fluorophores have low absorption and quantum yields, requiring long integration times to acquire the high photon counts needed for accurate pixel-wise multi-exponential decay fitting. Aim: A computationally light "region-of-interest" photon pooling technique was used to expedite label-free, single-cell FLIM acquisition and analysis, and its accuracy was compared with standard fitting techniques. Approach: We first characterized the accuracy and precision of "region-of-interest" photon pooling using known fluorescence standards and tested its ability to recover fluorescence lifetimes of single cells and large regions of interest with low photon budgets. Results: Single-cell metabolic information was accurately extracted from scanning periods as low as 1 s, and large FLIM mosaics were acquired 15 times faster than was possible with conventional pixel-level analysis. Lifetimes extracted using photon pooling were comparable to standard measurements requiring much longer integration times. The technique was also applied to measure fluorescence lifetimes in highly dynamic live samples. Conclusions: "Region-of-interest" (ROI) photon pooling extracts fluorescence lifetimes from live, dynamic samples with low photon budgets, expediting image acquisition while preserving cell-level or ROI-level lifetime information while sacrificing intra-ROI spatial resolution. The technique is computationally light, does not require machine learning algorithms, and can be integrated with commonly used analysis software and file types.