1 Introduction

Accurately observing how cellular states evolve in response to molecular interactions is crucial for advancing our understanding of biological systems and the development of new, novel medicine [1, 2]. Real-time monitoring of these changes, particularly in live cells, requires multi-dimensional imaging techniques that can capture spatial, spectral, and temporal dynamics with high precision. Such capabilities are indispensable for revealing how perturbations at the molecular level influence the broader cellular environment [3].

Optical microscopy is vital for conducting minimally invasive measurements of dynamic cellular processes. Fluorescence microscopy has become increasingly valuable due to its high target specificity, which is also driven by advancements in fluorescent probe design [4,5,6]. This technique enables spatially resolved measurements of subcellular processes and, when coupled with dynamic or time-resolved imaging, provides unique insights into molecular mechanisms of action, potentially leading to new understandings of drug or biological activity. Although widefield intensity imaging with sequential spectral band measurements is common, it encounters limitations in complex environments such as thick and multispecies samples. The thickness of a sample can result in poor image quality and reduced specificity due to the influence of out-of-focus light, while spectral crosstalk complicates the identification of multiple species [7, 8].Time-resolved fluorescence microscopy offers a novel approach by providing high specificity for fluorescence signals [9]. Time-resolved measurements are challenging due to the intrinsic jitter of electronics and detector technology, but advances in optical and electronic techniques improve lifetime measurement accuracy [10, 11]. Confocal resolution further enhances this capability, particularly in crowded environments or when imaging thicker samples with dense spatial labeling and inevitable spectral overlap. This ensures that only species within the diffraction-limited focal volume affect the measurement [12,13,14]. Techniques such as Förster Resonance Energy Transfer (FRET), which facilitates the indirect observation of molecular interactions, benefit significantly from fluorescence lifetime imaging microscopy (FLIM) as it is more robust than traditional intensity-based measurements [15]. FRET is the non-radiative transfer of energy from an excited donor fluorophore to an acceptor fluorophore, and its process is reflected in a reduction in the donor’s lifetime due to additional non-radiative decay pathways. This process requires the donor and acceptor fluorophores to be within approximately 10 nm, providing quantitative measurements of protein binding and conformal change when the FRET fluorophores are attached to proteins of interest [16, 17].

Live cell imaging with confocal FLIM is inherently slow, primarily because image acquisition occurs one pixel at a time due to the absence of commercial multipoint confocal FLIM systems. This poses an obstacle to effectively using FLIM in pharmacological research because many biological signaling events and molecular interactions occur within the sub-second timescale. To avoid photobleaching and phototoxicity, the lowest possible excitation powers are used [18, 19]. This results in photon count rate limited pixel dwell time that determine the overall frame rate needed for accurate measurement of fluorescence decay [20]. A number of techniques have been developed to push the boundary of FLIM imaging to higher speeds due to the abundance of promising applications [21].

A promising alternative to single point confocal is multiplexed confocal imaging, which enables the parallel acquisition of multiple pixels, significantly increasing the photon count rate compared to single point confocal imaging. By allowing parallel spatial scanning, this method not only permits the use of laser powers that are compatible with live-cell imaging but also enhances contrast under challenging imaging conditions, all without compromising the frame rate [22]. Many multipoint techniques, however, use multiphoton imaging, which is limited by the total laser power a sample can sustain before photothermal damage occurs and typically has limited multiplexing factors due to the quadratic dependence of two-photon fluorescence on excitation laser intensities and limited optical efficiency of beam shaping optics [23, 24].

We have previously demonstrated a proof of concept multiplexed confocal FLIM system capable of acquiring time gated FLIM images in 3 s, proving its suitability for FLIM-FRET screening applications [25]. In this study, we present an improved optical system designed to enable high-speed imaging in a dynamic process that requires high frame rate measurements, such as the applications of caged compounds and triggered biological pathways, or in cyclic cardiovascular and neurological models, both of which require sub-second timescale measurements [26,27,28,29]. The system demonstrates improved optoelectronic synchronization and is capable of generating three time-gated images for a real-time FLIM capture frequency of 4 Hz at 960 × 960 pixel resolution, enabling dynamic FLIM measurements in live cells.

2 Methods

In the following sections, we demonstrate the optical system design, discuss mechanical scanning limitations, and synchronization of the window scanning to prevent imaging artifacts. We also present a method for calibrating the multiple temporal gates of a single photon avalanche detector (SPAD) array, ensuring accurate lifetime estimation. This includes a detailed measurement of fluorescent solutions to establish the photon count requirement for precise lifetime measurements. Finally, we demonstrate the system’s performance using fixed and fluorescently stained samples and live samples expressing a transient FRET construct.

2.1 Optical system design

The simplified optical system is shown in Fig. 1. The system uses a pulsed diode laser (LDH-440 M, Picoquant) that produces 440 nm laser pulses with pulse widths of ~ 1 ns at a pulse repetition rate of 50 MHz. The output beam is coupled to a multimode fiber to improve the uniformity of the output excitation beam. The divergent fiber output is collimated using a 300 mm lens (AC508-300-A, Thorlabs) into a 1.4 cm beam to provide a large and collimated illumination wavefront. The system uses a 300 µm pitch, 32 × 32 MLA (300 μm center-to-center pitch, square microlenses, SUSS MicroOptics), for generating 1024 points of excitation for multiplexing to enable fast acquisition time. The MLA excitation image is transmitted through a pair of achromatic doublets (AC508-200-A, Thorlabs), which form a telecentric relay and are reflected off a 442 nm long-pass dichroic (Di03-R442-t1-25 × 36, Semrock) onto a pinhole array (Walthy Precision Co) with a pitch designed to match the MLA excitation image footprint. The pinhole array has a square array of 25 µm pinholes at 300 µm pitch for optimal confocal imaging with a 60 × objective. A pair of galvanometer window scanners (GM-1000, Cannon) are responsible for raster scanning in X and Y dimensions [30]. The image of the excitation spot array passes through the camera port of a microscope (TI, Nikon) equipped with a 60 ×/1.40 NA objective (Plan Apo 60 ×/1.40 Oil, Nikon). The fluorescence emission is de-scanned by the window scanners and spatially filtered through the pinhole array. The fluorescence emission is then relayed through the emission telecentric relay (AC508-100-A, Thorlabs), the 442 nm long pass dichroic, and a 488 nm/40 nm band pass emission filter (FF01-483-25, Semrock). Finally, the signal image is collected by a SPAD array detector (SPC3, Micro Photon Devices). Software improvements were made to increase real-time imaging and image processing speed.

Fig. 1
figure 1

The optical system with a simplified optical path and custom pinhole array

Full size image

2.2 Window scanner mechanical limitations

Achieving a 4 Hz imaging rate resulted from improving the mechanical scanning control. Previous scanner control would step the scanner to each pixel position and stop, producing a short delay before movement to the next pixel position. The new scanner control instead moves in a continuous raster scan motion without stopping from the start of imaging to the end. A detailed calibration and characterization of the scanner motion was performed to ensure consistently high-quality image generation at high frame rates. The scanner calibration involved measuring the scanner waveforms from the galvanometer drivers to determine the total scan duration and confirm the ranges of motion. Figure 2 provides measurements of the total scan duration and maximum scanner velocity as a function of the pixel dwell time. Figure 2a shows that the updated scanning system achieves frame durations as low as 0.1 ms for a full frame. At 200 µs pixel dwell time, the expected image duration is 0.24 s, resulting in the system's highest mechanical frame rate of 4.1 fps. As indicated by Fig. 2b, however, the scanner velocity diverges from the expected behavior at a pixel dwell time of less than 200 µs. There is an additional 64 ms per frame that corresponds to the time required to move data from the camera to the computer at the end of the imaging period, reset its memory in preparation for the next acquisition, and reset the position of the galvo scanners to their start position. This reset period is an order of magnitude faster than previously demonstrated [25]. To compensate for the lost imaging time from the reset period, a decision was made to use a pixel dwell time of 250 µs for live cells, corresponding to a total scan duration of 0.28 s. This decision was made based on findings that the improvements to imaging rate are marginal when compared to the signal loss at shorter pixel dwell times. Thus, all dye standards and live cell images are imaged with a real-time frequency of 3.7 Hz.

Fig. 2
figure 2

a Frame rate and b scanner velocity measurements of galvanometer scanners for identification of the highest reliable mechanical scanning rate

Full size image

2.3 Synchronization and artefact calibration

At higher frame rates, significant artifacts became apparent in the fluorescence images. A difference in the scanner position during the bidirectional raster scan forms at scan rates faster than 0.7 frames per second. This is likely a result of the scanners’ acceleration limit. The scanner-derived artifact was calibrated by adjusting the scan start delay for each pixel dwell. This was accomplished by measuring the difference in the position of two adjoining rows and tracking the difference across a series of increasing scan start delays. These tests consisted of 100 measurements in 40 µs steps between − 2 ms and 2 ms to measure the appropriate trigger delay time for a specific pixel dwell setting. The tests were conducted at the lowest possible laser power to prevent photobleaching during the course of measurement, which was determined to be 0.15 µW per focus. Figure 3.a demonstrates the results as the change in optimal trigger delay as a function of the pixel dwell time. The results have a linear trend when the pixel dwell time is less than 1400 µs (1.3 s scan duration). This linear trend is stored within the software to allow adjustments to the scanner trigger delay dynamically as the user changes imaging parameters. Figure 3b demonstrates the effects on image quality before (top) and after (bottom) calibration, respectively, during 4 Hz imaging of a fixed convallaria specimen. The scale bar indicates a 2 µm dimension, and no filtering has been applied to the images. The red region highlights an area where the image quality is poor because of incorrect delay settings. Each highlighted region is accompanied by an illustration of the difference in edge alignment before and after the synchronization issue is corrected. The uncorrected edges are discontinuous and present a periodic structure, whereas the corrected image shows the same edge as smooth and continuous.

Fig. 3
figure 3

Bi-directional raster scan artifacts from incorrect camera trigger delay. a Delay correction as a function of pixel dwell time. b Pre (top) and post (bottom) correction results on image quality. Highlight ROI illustrates the difference in image quality for edge quality. Scale bars are 2 µm

Full size image

2.4 Calibration of the multiple temporal windows for lifetime measurement

Measurement of the temporal decay profile of a fluorophore requires capturing multiple temporal data points for fitting. The SPAD array used in this research is capable of measuring three distinct temporal gates. To verify the correct operation of the temporal gating, a 1 µM solution of Coumarin6 was used as a standardized sample. The positions of the temporal gates were each convolved across the laser-induced fluorescence decay by adjusting the delay between laser triggering and the start of gate 1. The delay is internally controlled by the sensor, which allows the placement of all three gates within a 20 ns period corresponding to the 50 MHz laser frequency. The gate widths are each set to 3 ns, with each gate starting only after the previous gate has finished, and due to internal limitations of gate placement, a 2 ns additional delay occurs between the end of the second gate and the beginning of the third. As a result, the total gate duration for all three gates is 11 ns, meaning a maximum temporal shift of 9 ns is possible within the 20 ns laser period. The delay positions of the gate convolved fluorescence decay were measured and compared to the programmed values. The natural logarithm of the gate convolved fluorescence decay functions for all three gates is shown in Fig. 4. Gates 1–3 are illustrated in red, blue, and green, respectively. The camera software limitation to gate placement is characterized by the vertical black unbroken line. Each gate shows a mono-exponential decay characterized by its linear profile on the natural logarithmic scale. The results of these measurements show that the appropriate temporal width between gate 1 and gate 2 is 2.6 ns for the software-selected gate width of 3 ns. This gate separation was used throughout all experiments to maintain lifetime consistency. The decay profiles also allow an optimum delay to be set such that each gate begins in the correct position of the fluorescence decay. That position is detailed by the vertical black broken line, which indicates where the beginning of the first gate position is set. The temporal zero point is relative to the signal transmission time through the wired connections.

Fig. 4
figure 4

Three gate windows response to convolution with 1 µM Coumarin6 fluorescence decay with 1000 ms pixel dwell time. Gates 1, 2, and 3 are plotted in red, blue, and green, respectively. The selected delay time to run experiments is indicated by the broken vertical black line. The internal software imposes a limit to the delay of the gates, and that is shown in the graph as an unbroken black vertical line

Full size image

The two gates that are selected for lifetime determination are gate 1 and gate 2. Gate 3 is used to collect the remaining fluorescence decay photons to allow higher photon counts in the intensity images. To calculate the lifetime of a sample, the method of rapid lifetime determination (RLD) is used [31].

$$\tau_{RLD} = - \frac{\Delta t}{{\ln \left( {\frac{{G_{2} }}{{G_{1} }}} \right)}}$$
(1)

where \({G}_{1}\) and \({G}_{2}\), are the photon counts recorded in gate 1 and gate 2 of the camera exposure, and \(\Delta t\) is the temporal separation of the two gates. RLD is a fast and efficient algorithm that is compatible with real-time visualization for lifetime features and is useful for our 3-gate camera. Two-gate RLD computes the average lifetime of the sample and does not provide the ability to separate multiexponential decays, but given the mono-exponential nature of the fluorophore decays we are interested in measuring, the technique is sufficient. RLD is known to be optimized for the measurement of a limited range of lifetimes given the error of having only two temporal samples, but error analysis of the technique suggests it performs well for lifetimes as low as a quarter of the gate width given sufficient photon counts [32, 33]. Novel schemes for reducing error in the estimation of shorter lifetimes are possible using overlapping gated schemes, as well as separating multiexponential decays given three gates, but those are not explored in this work.

2.5 Dilution series measurements for lifetime precision and accuracy

To measure improvements in lifetime estimation at high frame rates, a dilution series was performed. This allows us to measure changes to the accuracy and precision of lifetime estimation. It also allows us to set appropriate threshold values for noise rejection in lifetime images. Care was taken to reduce the influence of noise signals without adversely affecting the large field of view and to establish a limit of measurement for all samples.

2.6 Sample preparation

The Convallaria Rhizome (Boston Electronics) was used to demonstrate image quality on a spatially complex lifetime sample. This convallaria is stained with two dyes: Safranin and Fast-Green. The excitation power used to measure the sample was 0.5 mW, measured before the objective lens, corresponding to 0.5 µW of power per each focal point. Spatial filtering with a 5 × 5 average filter was used to improve image quality by removing the effect of shot noise in the lifetime calculation. The intensity image however, had no spatial filter applied to it and was used to create an alpha layer in the final processing step of the composite image. This filtering and processing step is applied to all composite images produced.

A 5 mM stock solution of Fluorescein (MW = 332.31 g/mol) in phosphate buffered saline (PBS) was created and diluted to 1 mM, from which a series of 10 × dilutions were performed. 200 µL of 1 mM fluorescein in PBS was pipetted into 1.8 mL of PBS to form 2 mL of 100 µM fluorescein. This 10 × dilution step was repeated three times to yield 10 µM, 1 µM, and 100 nM solutions. A 2 mM stock solution of Coumarin-6 (MW = 350.43 g/mol) in ethanol was created, then diluted 2 × to form a 1 mM solution. A series of 10 × dilutions were performed by pipetting 0.1 mL of Coumarin-6 into 0.9 mL of ethanol. 100 µL of each of the Fluorescein and Coumarin-6 dilutions were pipetted into separate wells on a 384-well plate, and the remaining solutions were stored at 4 °C and protected from light.

HEK293 T cells (ATCC CRL-3216, Sigma-Aldrich, St. Louis, MO, USA) were cultured in accordance with previously established protocols, using Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS) and incubated at a temperature of 37 °C and 5% CO2. Samples were prepared using Phenol Red Free DMEM-F12 with 10% FBS to reduce autofluorescent signals and transfected using protocols adapted from previously published methods [25].

3 Results and discussion

3.1 Convallaria

Figure 5 demonstrates the confocal FLIM of the Convallaria at 4 Hz. The confocal intensity and confocal FLIM are demonstrated separately in Fig. 5a–b, and the composite where the intensity is used as an alpha mask on the confocal FLIM image is demonstrated in Fig. 5c. The convallaria demonstrates an excellent sample to investigate lifetime differences, as lifetime changes are associated with the differences in the environment of the stain. Although the two stains, Safranin and Fast green, have emission wavelength peaks at 590 nm and 660 nm, we are still able to measure the fluorescence lifetime of the bright stains [34]. The convallaria is only meant to demonstrate the image quality of the confocal lifetime imaging system. The measured confocal lifetime displayed variations around either 2.4 ns or 2.9 ns, based on the location of the stain binding, which agrees with previous measurements for the stains in this sample [35].

Fig. 5
figure 5

Confocal FLIM image of the Convallaria Majali’s at 4 frames per second. a confocal intensity, b confocal FLIM, c composite image. Scale bars are 20 µm

Full size image

3.2 Coumarin-6 dilution series for lifetime error, precision, intensity, and threshold selection

The measurements from the dilution series and the associated accuracy and error analysis, as shown in Fig. 6a, highlight the importance of maintaining a threshold of 25 photons to provide accurate lifetime estimation using RLD. This photon count allows for imaging 10 µM concentrations of Coumarin-6 with a pixel dwell time of 250 µs. At this threshold, 99% of the field of view is preserved, with only minimal losses occurring in the corners where excitation power is lowest. Figure 6b illustrates the system’s sensitivity as a function of the Coumarin-6 concentration and demonstrates the results of the improved optical system by maintaining sufficient signal to measure accurate lifetimes at concentrations as low as 1 µM, with pixel dwell time of 500 µs and 1000 µs, corresponding to 2 Hz and 1 Hz imaging, respectively. Below the cutoff threshold, a notable increase in both the estimation and the error of the lifetime occurs. As photon counts decrease, so does the signal-to-noise ratio, limiting the accuracy of the rapid lifetime estimation. Currently, the signal threshold of 25 photon counts and careful background subtraction results in a noise floor of 0.6 photon counts. This means that the SNR is just below 5 at the lifetime measurement limit. This is a result of the low noise of the SPAD technology, which enables precise measurement of the mono-exponential decay.

Fig. 6
figure 6

Fluorescence measurement limits. a Fluorescence lifetime measured through changes in mean pixel photon count. b Fluorescence photon count against coumarin 6 dilution series for pixel dwell time of 250 µs, 500 µs, and 1000 µs

Full size image

Measurements of the full field of view for Coumarin-6 and Fluorescein are presented in Fig. 7. Figure 7a demonstrates that at 250 µs dwell time, the system accurately measures 10 µM concentration solutions while maintaining an excellent field of view, allowing the full multiplexing factor during confocal FLIM imaging. In Fig. 7b, the histograms of Coumarin-6 and Fluorescein are presented, which were measured as 2.5 ± 0.2 ns and 4.1 ± 0.7 ns, respectively. Additionally, Fig. 6a reveals the presence of hot pixels, which appear as square mosaic regions with higher-than-average lifetimes or darker-than-average regions due to threshold removal. These regions are due to sensor fabrication artifacts, not illumination issues. However, the corners of the Fluorescein image in Fig. 7a (bottom) show increased error rates due to non-uniform illumination from the multimode fiber output. Together, these figures demonstrate the system's high accuracy and precision in measuring fluorescence lifetimes, particularly for Coumarin-6 and Fluorescein, which are regarded as gold standard samples due to their exceptionally stable lifetimes.

Fig. 7
figure 7

a Full field of view (960 × 960 pixel) confocal lifetime images for Coumarin-6 (top) and Fluorescein (bottom) b Histograms for Coumarin-6 and Fluorescein

Full size image

3.3 Live cell accuracy and precision analysis at high frame rates

Live cell FLIM measurements are a better demonstration of accuracy and precision at high frame rates in live cells. In Fig. 8a, we demonstrate the confocal lifetime images of two cells with different lifetimes. The two subcellular regions of interest in Fig. 8 are selected from cells transfected with both mCerulean3 and Venus and show different ratios of expression of the two fluorescent proteins. Figure 8b demonstrates the histograms of the two 10 × 10 ROI labeled A and B from each field of view in Fig. 8a. Lifetimes for each field of view for regions A and B are 3.7 ± 0.2 ns and 3.3 ± 0.2 ns for a pixel dwell time of 1000 µs and are 3.8 ± 0.3 ns and 3.5 ± 0.3 ns for a pixel dwell time of 250 µs. The broader distribution of the estimated lifetime (shown in Fig. 8b) is the result of higher lifetime estimation error in low photon count regions due to shorter exposure time (250 µs). The lower average lifetime in longer exposure dataset (1000 µs) indicated limited level of photobleaching even though the excitation power is relatively low (0.5 μW/foci) in our experiments.

Fig. 8
figure 8

a Live cell measurements of cytosolic expression of the fluorescent proteins mCerulean3 and Venus, co-transfected in HEK293 T cells, imaged with pixel dwell time of 250 µs and 1000 µs. Scalebars of 10 µm. b Histograms of the fluorescent lifetimes for two regions labeled A and B for each pixel dwell time

Full size image

3.4 Live cell measurements for lifetime changes over FRET constructs

A live cell FLIM-FRET experiment was conducted to demonstrate the potential of our system to measure FRET in an assay format. Figure 9 demonstrates the results of the transfection of HEK293 T with mCerulean3, Venus, mCerulean3, Venus co-transfected, and a construct where mCerulean3 and Venus are linked and undergo FRET. Figure 9a demonstrates the full field of view measurements of the transfected cells and demonstrates the differences between the single protein transfected cells and the co-transfected mCerulean3 + Venus cells. Figure 9b demonstrates the normalized fluorescence lifetimes of the same fields of view. The histograms alone show a small change in the average lifetime as a result of the cotransfection. With the confocal FLIM images, we can see the specific cells that are affecting the lifetime distribution. Our assay demonstrates lifetime values of 3.8 ns ± 0.5 ns, 3.5 ± 0.6 ns, 2.9 ± 0.4 ns, and 2.3 ± 0.3 ns for mCerulean3, Mcerulean3 + Venus, Venus, and the mCerulean3-Venus FRET construct and is in good agreement with previously published lifetime values [36, 37].

Fig. 9
figure 9

a Live cell FLIM-FRET experiment demonstrates full field of view images of HEK293 T cells, transiently transfected with mCerulean3, Venus, mCerulean3 and Venus, and an mCerulean3-Venus linker protein that undergoes FRET. Scalebars of 20 µm. b The normalized histograms of the four fields of view

Full size image

4 Conclusions

We developed a fast multiplexed multipoint confocal FLIM with substantially improved sensitivity and speed for imaging live cell dynamic events at subcellular resolutions. This system obtains 960 × 960 pixel fluorescence lifetime images at a real-time imaging frequency of 4 Hz. This system features a fully redesigned optical path that maintains the full field of view while increasing the image and optical quality of resulting confocal FLIM images. Improvements to software structure permit simultaneous image processing and acquisition.

Pinhole array based confocal imaging is not a new technology, as demonstrated in technologies such as the CSU series spinning disk confocal (Yokogawa) and the VT-Infinity4 confocal system (VisiTech International). These technologies will use methods such as pinhole rotation or pinhole sweeping to achieve confocal imaging with the sparse array. In each method, a pinhole is responsible for imaging one smooth arc or line through the sample [13]. This technology is unique in its use of the window scanner as a means of uniformly displacing large images to achieve scanning. The window scanner, as demonstrated in [30], achieves image displacement by utilizing the refractive properties of the window to impart uniform displacement. This is a perfect match for large field-of-view scanning when compared to mirror-based scanners, which suffer from distortion during large image scanning and, as a result, are incompatible with pinhole array-based confocal. The window scanners are responsible for performing a bidirectional raster scan between the pinhole spacings, covering a two-dimensional subarea.

Measurements of the fluorescent lifetime standards, Coumarin6 and Fluorescein, demonstrate lifetimes in agreement with literature reported values [38, 39]. The lifetime precision of Fluorescein is significantly lower compared to Coumarin6, which can be attributed to the mismatch in excitation wavelengths. Fluorescein, with an excitation peak at 498 nm, is suboptimal for excitation using our 442 nm laser wavelength. Improvements should be made to the signal processing for extracting multiexponential decays. Changes to the signal processing could be explored by replacing the gated SPAD array with a TCSPC detector such as the PF32 (PF32, Photon Force), as this would provide higher temporal resolution multi-exponential fitting.

The system’s frame rate is now limited by the fastest mechanical scanning time of the window scanners rather than the optical design. This is due to the improved optical throughput from the reduced number of surfaces in the emission optical path when using the pinhole array for confocal imaging. Traditional TCSPC systems are limited to a maximum of 0.1 photons per excitation pulse [40]. With a 50 MHz laser, this limits the photon count rate to 5 Mcps. Under identical imaging conditions and the same photon count rate limitations, the multipoint confocal samples 32 × 32 points simultaneously, meaning a maximum theoretical count rate of 1024 Mcps. Improvements to the system’s maximum imaging frequency will depend on new optomechanical designs, such as resonant galvanometric scanners, which may allow the technology to reach the tens of Hz [41]. This is a substantial improvement as currently fast confocal FLIM microscopes typically limit the field of view to a 256 × 256 pixel region to achieve faster than 1 Hz imaging rates. We can conduct dynamic imaging experiments without compromising on the field of view.

This system serves as one step forward toward the design of a video rate multipoint confocal FLIM microscope. This technology has the potential to impact various fields of biological imaging and drug discovery by providing dynamic multidimensional imaging that can identify spatial and lifetime changes of labels in live cells.