Fluorescence lifetime measurement method and system based on low-frame-rate video imaging

Abstract

The invention belongs to the technical field of fluorescence lifetime detection, and particularly relates to a fluorescence lifetime measurement method and system based on low-frame-rate video imaging. The method comprises the following steps: irradiating a to-be-detected sample by adopting an excitation light source, simultaneously collecting a video of a periodic luminescence attenuation process of the to-be-detected sample, and separating the video into a plurality of images with frame number marks; the excitation frequency of the excitation light source and the frame rate of video acquisition meet the following condition: FLED = Fcam + n, and n is a positive integer smaller than Fcam; all the images are converted into gray level images, and the average gray level value of light-emitting areas of the gray level images serves as a fluorescence signal intensity value; storing the fluorescence signal intensity value according to a frame number index, and screening out a signal attenuation sequence from the fluorescence signal intensity value; and fitting the signal attenuation sequence by using a least square method to obtain a fitting curve of the signal intensity value along with the time change, and obtaining the fluorescence lifetime of the sample to be detected from the fitting curve. According to the invention, the detection complexity and cost can be reduced, and the fluorescence lifetime of millisecond or even microsecond level can be measured.

Description

Technical Field

[0001] The present invention belongs to the technical field of fluorescence lifetime detection, and in particular relates to a fluorescence lifetime measurement method and system based on low frame rate video imaging. Background Art

[0002] Time-gated technology utilizes detectors combined with electronic delay devices to detect photons or signals within a specific time window. It is typically used to improve the signal-to-noise ratio (SNR) of a detection system. The core concept is to control the detector to receive signals only within a precise "time window," thereby eliminating unwanted background interference or scattered signals. Time-gated technology has numerous applications in ultrafast photophysical processes, the temporal behavior of chemical systems, and long-lived fluorescence detection. For example, in time-resolved fluorescence (TRF) detection, time-gated technology can be used to collect only the light signal within a specific delay time window after pulsed light excitation, effectively eliminating interference from the excitation light and short-lived background fluorescence, significantly improving the SNR. Due to its high SNR and sensitivity, time-resolved fluorescence detection technology has been increasingly widely used in medical testing fields such as cell and tissue imaging and in vivo / intraoperative imaging.

[0003] The most fundamental and core technical challenge in time-gated fluorescence detection is how to accurately measure the fluorescence lifetime τ. Fluorescence lifetime τ is the time it takes for the fluorescence intensity of a fluorescent material to drop to 1 / e of its maximum intensity during excitation, I0. It is an inherent property of fluorescent materials, and its changes can very sensitively reflect changes in the fluorescent material's microenvironment.

[0004] Currently, most time-gated imaging systems utilize high-precision specialized equipment, such as ultrashort pulse light sources (such as femtosecond lasers), high-speed photodetectors (such as single-photon avalanche diodes (SPADs) or PMTs), and high-frame-rate high-speed cameras (such as streak cameras or intensified CCDs). These specialized devices offer precise gating time control, but the hardware cost is very high, and they also require a high-precision synchronization control system. Some researchers have used low-cost mechanical choppers in conjunction with conventional low-frame-rate cameras for time-gated imaging and fluorescence lifetime measurement. However, these devices are mechanically complex, require high environmental stability, and are difficult to miniaturize. Other researchers have attempted to directly acquire time-gated images using smartphones or conventional cameras, then measure fluorescence lifetimes by exponentially fitting the pixel intensities within the time-varying time-gated frames. However, this approach is primarily suitable for measuring ultralong-lifetime phosphorescence or long-lasting luminescence, and is difficult to directly measure fluorescence lifetimes below the millisecond level.

[0005] Without the need for complex mechanical structures such as precision synchronization control systems and choppers, if time-gated imaging can be directly achieved on ordinary consumer-grade low-frame-rate cameras and the fluorescence lifetime τ can be measured in milliseconds or even microseconds through innovation in detection technology and algorithms, the application cost of time-gated technology will be greatly reduced, and the application of time-gated fluorescence imaging and time-resolved fluorescence detection technology in POCT, home self-testing and other fields will be promoted. Summary of the Invention

[0006] The purpose of the present invention is to provide a fluorescence lifetime measurement method and system based on low-frame-rate video imaging, which directly obtains fluorescence decay video through a video frequency difference sampling method, and then extracts time-gated images and fluorescence intensity change data in the video through image processing algorithms, thereby directly measuring fluorescence lifetime at the millisecond or even microsecond level.

[0007] To achieve the above object, the present invention provides a fluorescence lifetime measurement method based on low frame rate video imaging, comprising the following steps:

[0008] S1, using an excitation light source to illuminate the sample to be tested, while simultaneously capturing a video of the periodic luminescence decay process of the sample to be tested, and separating the video into a number of images with frame number marks;

[0009] Wherein, the excitation frequency F of the excitation light source is LED and the frame rate of video capture F cam Satisfied: F LED =F cam +n, n is less than F cam A positive integer;

[0010] S2. Convert all images into grayscale images, and use the average grayscale value of the luminescent area of ​​the grayscale image as the fluorescence signal intensity value;

[0011] S3. Save the fluorescence signal intensity values ​​according to the frame number index, and screen out a signal decay sequence therefrom; fit the signal decay sequence using the least squares method to obtain a fitting curve of the signal intensity value changing with time, and obtain the fluorescence lifetime of the sample to be tested from the fitting curve.

[0012] Furthermore, in step S2, the grayscale image is first subjected to noise reduction processing, and then image segmentation is performed to preliminarily extract the ROI area. Then, a region growing algorithm is used to use the detected center point of the ROI area as the initial seed point, and based on the pixel intensity difference threshold, the area is expanded outward from the initial seed point to generate a segmentation result of the ROI luminous area. Finally, the segmented area is filled with holes by the flooding algorithm, and the edges are filled by morphological dilation to finally obtain the luminous area.

[0013] Furthermore, step S3 obtains the best fitting result by multi-cycle data fusion, specifically including: extracting the signal attenuation sequence of each cycle, fitting the signal attenuation sequence of each cycle using the least squares method, and selecting the cycle with the best fitting effect as the reference cycle;

[0014] The theoretical starting time is deduced from the decay curve based on the best fitting effect, and the starting point t of the theoretical peak of the signal intensity is calculated. start , adjust the time axis of the reference period to align with the theoretical attenuation curve; then insert the signal attenuation sequences of other periods in sequence, and re-fit, evaluate the fitting results, if they meet the preset fitting effect, complete the multi-period data fusion and obtain the fitting curve; if not, the signal attenuation sequences of other periods are time-shifted and then inserted, and the time offset is continuously adjusted until the fitting result meets the preset fitting effect to obtain the final fitting curve.

[0015] Furthermore, the adjustment range of the time offset is within 100 μs before and after the reference period, and each time offset is 0.1-1 μs.

[0016] Furthermore, the fitting effect of the signal decay sequence of each cycle is obtained by R 2 The value and / or residual standard deviation are evaluated, R 2 ≥0.8, preferably R 2 ≥0.9, the range of residual standard deviation is -2 to 2.

[0017] Furthermore, the sample to be tested is a fluorescent material or a light emitting diode driven by an RC circuit;

[0018] When the sample to be tested is a fluorescent material, the excitation light source is a laser, a tungsten lamp or an LED lamp;

[0019] When the sample to be tested is a light emitting diode driven by an RC circuit, the excitation light source is the light emitting diode, that is, the video of the light emitting diode's light decay process is directly collected.

[0020] Furthermore, the duty cycle D and excitation frequency F of the excitation light source are LED and the fluorescence lifetime τ satisfy the following relationship:

[0021] (1-D) / F LED ≥5τ.

[0022] and / or, the F cam ≥30FPS, n≤10FPS; preferably, 30FPS≤F cam ≤240FPS. LED The unit is Hz, only F LED and F camThe dimensionless numerical equality relationship must be satisfied.

[0023] The present invention also provides a fluorescence lifetime measurement system based on low frame rate video imaging, comprising:

[0024] The excitation module is used to excite the sample to be tested to produce periodic luminescence decay;

[0025] The video acquisition module is used to collect the video of the periodic luminescence decay process of the sample to be tested and separate the video into several images with frame number marks; the excitation frequency F of the excitation module LED and the frame rate F of the video acquisition module cam Satisfied: F LED =F cam +n, n is less than F cam A positive integer;

[0026] An image processing module is used to convert all images into grayscale images, and use the average grayscale value of the luminescent area of ​​the grayscale image as the fluorescence signal intensity value;

[0027] The lifetime calculation module is used to save the fluorescence signal intensity values ​​according to the frame number index and filter out the signal decay sequence; use the least squares method to fit the signal decay sequence to obtain a fitting curve of the change of the signal intensity value over time, and obtain the fluorescence lifetime of the sample to be tested from the fitting curve.

[0028] Furthermore, the image processing module further includes:

[0029] a noise reduction processing unit, configured to perform noise reduction processing on the grayscale image;

[0030] An image segmentation unit is used to perform image segmentation on the grayscale image after noise reduction to extract the ROI area;

[0031] The region growing segmentation unit is used to use the region growing algorithm to take the center point of the detected ROI region as the initial seed point, and based on the pixel intensity difference threshold, expand the region outward from the initial seed point to generate the segmentation result of the ROI luminous area;

[0032] The segmented area filling unit is used to fill the holes in the segmented area by using a flooding algorithm and to fill the edges by using morphological dilation, so as to finally obtain the luminous area.

[0033] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the fluorescence lifetime measurement method based on low frame rate video imaging as described above are implemented.

[0034] In general, the above technical solutions conceived by the present invention have the following technical advantages compared with the existing technology:

[0035] 1. The fluorescence lifetime measurement method provided by the present invention is based on the excitation frequency F LED and the frame rate of video capture F cam The method uses a frequency difference to sample the video. This results in a fixed phase difference in the sampled data. By continuously increasing the delay time, the data sampling points can traverse the entire excitation cycle, forming a time-gated sequence. This effectively extracts the fluorescence decay period data from the time-gated frame sequence, and the fluorescence lifetime is obtained through curve fitting. This method uses a simple detection device and can measure fluorescence lifetime in the millisecond or even microsecond range.

[0036] 2. The present invention can realize fluorescence lifetime measurement based on the time gating technology of ordinary consumer-grade low frame rate cameras (30-240FPS), which can solve the problem that the existing time gating technology based on mobile phones or ordinary cameras requires additional mechanical or electronic delay devices.

[0037] 3. The present invention can significantly reduce background interference, improve signal accuracy, and thus improve the accuracy of fluorescence lifetime prediction through grayscale image denoising, segmentation, region growing, segmented region filling and other processing processes.

[0038] 4. The present invention continuously optimizes the fitting results through multi-cycle data fusion and iterative fitting based on time offset, and finally obtains an optimized result that integrates information from multiple cycles, thereby improving the accuracy of fluorescence lifetime measurement, especially the accuracy of short-lifetime (less than 100μs) fluorescence detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 F cam 30FPS, F LED Schematic diagram of 30Hz sampling simulation.

[0040] Figure 2 F cam 30FPS, F LED Schematic diagram of 31Hz sampling simulation.

[0041] Figure 3 F cam =30FPS, F LED = Schematic diagram of sampling results when 31Hz is used.

[0042] Figure 4 for Figure 3 Schematic diagram of the fitting of the decay sequence

[0043] Figure 5Schematic diagram of the processing flow of the actual fluorescence attenuation video obtained by video frequency difference sampling.

[0044] Figure 6 Schematic diagram of the process of multi-cycle data fusion.

[0045] Figure 7 This is a sequence of LED brightness change images indexed by frame number, where the images from the upper left to the right are frame_59.jpg to frame_72.jpg.

[0046] Figure 8 This is the result of multi-cycle data fusion (the figure shows the case of a 2μF capacitor).

[0047] Figure 9 Schematic diagram of the structure of the fluorescence detection chamber.

[0048] Figure 10 This is a sequence of fluorescence decay images, where the first image from the upper left is frame_117.jpg to frame_131.jpg in order from the right.

[0049] Figure 11 This is the result of a multi-cycle fusion fitting of Eu(TTA)3 solution.

[0050] Figure 12 These are TCSPC measurement results.

[0051] Figure 13 Comparison of fluorescence attenuation of MB-S, MB-L solutions and mixed solution.

[0052] Figure 14 is the time decay curve of the fluorescence intensity of the three wells.

[0053] Figure 15 The experimental materials and instruments used in Example 4 are as follows: (A) two sets of fluorescent immunochromatographic test strips (quality control cards) for testing; (B) BH1000 dry-type fluorometer; and (C) BH101 handheld fluorometer.

[0054] Figure 16 This is a representative sequence of fluorescence decay images of the T line of the fluorescent test strip.

[0055] Figure 17 Fluorescence lifetime test results of two test paper cards: (A) Fluorescence lifetime fitting results of MB-L test paper card; (B) Fluorescence lifetime fitting results of MB-S test paper card.

[0056] Figure 18 MB-L time-resolved measurement results: (A) is the linear relationship with the BH1000 result; (B) is the linear relationship with the BH101 result.

[0057] Figure 19 MB-S time-resolved measurement results: (A) is the linear relationship with the BH1000 result; (B) is the linear relationship with the BH101 result. DETAILED DESCRIPTION

[0058] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0059] A fluorescence lifetime measurement method based on low frame rate video imaging comprises the following steps:

[0060] S1, using an excitation light source to illuminate the sample to be tested, while simultaneously capturing a video of the periodic luminescence decay process of the sample to be tested, and separating the video into a number of images with frame number marks;

[0061] Wherein, the excitation frequency F of the excitation light source is LED and the frame rate of video capture F cam Satisfied: F LED =F cam +n, n is less than F cam A positive integer;

[0062] S2. Convert all images into grayscale images, and use the average grayscale value of the luminescent area of ​​the grayscale image as the fluorescence signal intensity value;

[0063] S3. Save the fluorescence signal intensity values ​​according to the frame number index, and screen out a signal decay sequence therefrom; fit the signal decay sequence using the least squares method to obtain a fitting curve of the signal intensity value changing with time, and obtain the fluorescence lifetime of the sample to be tested from the fitting curve.

[0064] The details are as follows:

[0065] (1) Select a suitable camera frame rate F cam and shutter time T Shutter , and then select the appropriate LED excitation light source switching frequency F LED , LED duty cycle D = (high level time / total time) * 100%, by adjusting the frame rate F of the ordinary camera cam and LED frequency F LED A fixed time delay between frames is achieved by adding a gap between the two frames to obtain a video of a repetitive fluorescence decay cycle.

[0066] First, according to the possible range of fluorescence lifetime τ, choose an appropriate camera frame rate F cam, shutter time (T Shutter ), try to obtain a higher intensity fluorescence signal, but it should be shorter than the LED off time to avoid capturing signals outside the decay period. After setting the camera parameters, select the appropriate LED parameters. By adjusting the LED frequency F LED and the camera frame rate F cam The frequency difference between the two frames can be used to precisely adjust the phase change between the time-gated frames and thus modulate the delay time. In order to increase the amount of data and accurately measure the fluorescence lifetime (τ) of the sample, the off time of each LED cycle (given by (1-D) / F LED The fluorescence lifetime (defined as τ) should be significantly longer than the entire decay curve, and here it can be selected to be more than 5 times the fluorescence lifetime τ.

[0067] (1-D) / F LED ≥5τ

[0068] Video frequency difference sampling: In the general settings, set the LED frequency F LED Set to F cam , the camera can collect n data points in each LED cycle, but in this case, the phase of the data collected in each LED cycle remains unchanged, and we cannot effectively collect data during the fluorescence decay period, as shown in the simulation diagram Figure 1 However, if the LED frequency F LED Set to F cam +n, we can see that the data sampled every other frame will have a fixed phase difference, that is, a fixed delay time will be added to the LED cycle every other frame. By continuously increasing the delay time, the data sampling points can traverse the entire LED cycle, forming a time-gated sequence. We can effectively extract the data of the fluorescence decay period in the time-gated frame sequence. The delay time Δt between each frame n It can be calculated by the following formula:

[0069]

[0070] From the formula we can see that F cam The larger the value, the smaller the value of n, and the step time Δt generated. n The smaller the value, the more precise the step sampling interval of time gating can be controlled. Therefore, the camera frame rate is increased as much as possible and the frequency difference is reduced to improve the system resolution. However, considering the accuracy of frequency control, n is set to 1 to achieve a frequency difference of 1Hz. At this time, the delay time Δt1 between each frame will be equal to:

[0071]

[0072] If you set the camera frame rate F cam is 30FPS, LED frequency F LED If the frequency is 31Hz, the delay time Δt1 can be calculated by the formula:

[0073]

[0074] Use Python code to simulate and verify the video frequency difference process. Set the fluorescence excitation process to instantaneous excitation. Keep the high level in the set LED high level range to simulate the fluorescence being continuously excited by the LED light source. Start the single exponential decay in the low level range to simulate the fluorescence decay process after the excitation stops. Use a periodic integration window to simulate the camera sampling process, record the sampling value of each integration window, and use the Matplotlib library to realize the overall simulation image display. Set the camera F cam is 30FPS, exposure time is 1 / 1000s; set LED frequency F LED The frequency was set to 31 Hz, the duty cycle was set to 20%, and the fluorescence lifetime was set to 1 ms.

[0075] Through simulation Figure 1 、 2 It can be seen that if the video frequency difference sampling principle is not applied, the sampling window in each signal period is in a repeated position, while the sampling window that applies the video frequency difference sampling principle presents a state of step-by-step continuous sampling within the signal period, which can verify the feasibility of the basic principle of video frequency difference sampling.

[0076] Multiplying the index of the signal sequence obtained by sampling the video frequency difference by Δt1, we get a signal sequence indexed by the relative time within the fluorescence cycle, as shown in the simulation diagram. Figure 3 As shown, the attenuation segment data in the signal sequence is fitted with a least squares single exponential decay method, and the results are shown in the simulation diagram. Figure 4 As shown in the figure, the fitted fluorescence lifetime result is 1ms, which is consistent with the set fluorescence lifetime. It can be proved that the frequency difference time gating system can be used to detect the fluorescence decay signal. The fluorescence lifetime can be obtained by single exponential decay fitting, which verifies the principle feasibility of the time-gated fluorescence lifetime detection system based on video frequency difference sampling.

[0077] (2) The actual fluorescence attenuation video obtained by sampling the video frequency difference is processed as follows: Figure 5 First, the video is separated into individual images labeled with frame numbers, and then all images are converted to grayscale. After converting the fluorescence image to grayscale, it is found that the image noise is high at low brightness, which significantly affects the fluorescence signal. Gaussian noise reduction is highly versatile and achieves optimal average metrics such as no-reference SNR, PSNR, and SSIM for fluorescence images of varying intensities. The algorithm also has a short runtime. Therefore, the Gaussian noise reduction algorithm is used for all fluorescence image processing.

[0078] The processed grayscale image is segmented into a region of interest (ROI). The OTSU threshold segmentation algorithm is used to perform preliminary segmentation of the ROI region to obtain the centroid of the target region. A region growing algorithm is then used, using the detected ROI center point as the initial seed point. Based on the pixel intensity difference threshold, the region is expanded outward from the initial seed point to generate a segmentation result for the luminous region of the ROI. The segmentation result is then applied to the original image. Finally, the FloodFill algorithm is used to fill holes in the segmented region, maintaining a uniformity in the extracted signal values ​​and ensuring the integrity and accuracy of the segmentation result. The image is post-processed, and the segmentation result is subjected to a morphological dilation operation to fill in incomplete edges and ensure clear boundaries of the segmented region.

[0079] The segmentation result is used as a mask to extract the ROI luminous area in the LED grayscale image. The average grayscale value of the ROI luminous area is calculated as the simulated fluorescence signal intensity value. The average grayscale value extracted from all images is saved as a CSV file by frame number for subsequent analysis and processing.

[0080] (3) From the extracted signal, the continuous decay phase is automatically identified and the decay sequence that meets the requirements is selected. Due to the movement of the camera sampling window, not the entire decay phase is fluorescence decay data. When the sampling window leaves the LED excitation plateau, the signal value decreases at the highest rate. This time point can generally be used as the starting point of fluorescence decay. At the same time, the minimum sequence length and the maximum signal increase allowed number of points are set to avoid noise interference and the introduction of abnormal data. The decay sequence is fitted using the least squares method (LSM). An improved initial parameter estimation method is used to dynamically adjust the initial value of the fitting through the maximum and minimum values ​​of the signal. Combined with the optimization parameters of the curve fitting, the convergence and stability of the algorithm under low signal-to-noise ratio conditions are improved.

[0081] (4) Multi-cycle data fusion algorithm: Figure 6 For short-lifetime fluorescence detection under 100μs, the impact of insufficient single-cycle sampling data is far greater than the impact of abnormal data and random errors. Therefore, it is proposed to fuse multi-cycle data to increase the amount of effective data and thus enhance the accuracy of data fitting. Given that the derivative of the exponential decay function is monotonically increasing and always less than 0, it can be inferred that the absolute value of the derivative at the starting point of the decay function is the largest. Since the time intervals between adjacent data points are equal, the algorithm calculates the decay values ​​of adjacent points within each cycle of data and identifies the point with the fastest decay rate as the starting point of the exponential decay sequence.

[0082] After determining the starting point, the continuous decay sequence detection algorithm is used to automatically extract the decay sequence of each cycle in the data. The LSM algorithm is used to fit each cycle, and the fitting effect is evaluated by indicators such as R2 value and residual standard deviation. The cycle with the best fitting effect is selected as the benchmark data. In the fitting process, the reasonable range of parameters is set, such as τ min =10, τ max = 1500 to ensure that the fitting results are within a reasonable range. For example, for the decay sequence of each cycle, the LSM algorithm is used to fit different lifetime estimates, and the cycle with the highest comprehensive score is finally selected as the benchmark.

[0083] Then, the theoretical starting time is deduced based on the decay curve of the best fitting effect, and the starting point t start , adjust the time axis of the benchmark period to align with the theoretical decay curve. For decay sequences of other periods, extract data starting from the point of maximum drop and perform time offset optimization to improve the overall fit. During the optimization process, the core idea is to merge the data of the candidate period with the benchmark data by adjusting the time offset, then re-fit the data and calculate the new R2 value. In the function used to adjust the time offset, set the time offset range to within 100μs before and after the benchmark period, and search for the optimal time offset in steps of 0.5μs to ensure a better fit for the merged data.

[0084] Throughout the entire process, data from each cycle is continuously iterated, gradually merged and optimized, ultimately resulting in an optimized result that integrates information from multiple cycles. At each step, corresponding judgment conditions and processing logic are set to ensure the rationality and effectiveness of the process. The merging progress and fitting results are displayed visually for easy observation and analysis. For example, after merging the data from each cycle, a scatter plot and fitting curve are drawn to display the current merging status and fitting parameters.

[0085] Example 1: Measurement of LED brightness decay life

[0086] Fluorescence intensity typically decays exponentially over time, and the discharge process of an RC circuit also follows this exponential decay law. Specifically, when the capacitor is fully charged to an initial voltage of V0 and the power is turned off, the voltage across the capacitor changes as shown in the following formula, where R is the resistance and C is the capacitance. This mathematical form is equivalent to the exponential decay of fluorescence intensity.

[0087]

[0088] Based on this law, consider the feasibility of simulating the exponential decay of fluorescence lifetime by driving a light-emitting diode (LED) through an RC circuit. An LED is a semiconductor that allows current to flow in only one direction. When powered in the forward direction, the LED emits incoherent, narrow-spectrum light. The wavelength of the light emitted by the LED depends on the type of semiconductor material used. The LED semiconductor chip consists of a p-region with excess positive charge, an n-region with excess negative charge, and a junction between the two. A certain voltage is required for current to flow through the LED, only then will the LED emit light. If there is not enough voltage supplied, no current will flow and the LED will not emit light because the junction acts as a potential barrier. When current flows through the LED, electrons jump from the n-region to the p-region. When electrons move through the junction to the p-region, a recombination process occurs. The potential energy is converted into electromagnetic energy and emitted in the form of photons. The power supplied to the LED will be determined by measuring the voltage and current supplied to the LED because:

[0089] P=IV

[0090] Where P is power, I is current, and V is voltage. Light-emitting diodes are not 100% efficient at converting electrical potential energy into electromagnetic energy. Some of the initial electrical potential energy is always converted into heat, which can increase the temperature of the LED. As the LED junction temperature rises, its luminous output decreases. Therefore, at a constant input voltage, luminous output is expected to decrease over time due to junction heating. Sean King experimentally discovered that below a certain current, the current through an LED is proportional to the light intensity:

[0091] I e =k1I

[0092] Therefore, the change in LED brightness should also conform to a single exponential decay function, and the decay coefficient is related to the RC time constant, which provides theoretical support for using capacitor discharge to drive LEDs to simulate fluorescence decay.

[0093] The exponential decay characteristics of the RC circuit are used to simulate the exponential decay process of the fluorescence intensity. That is, when the capacitor is fully charged, the initial voltage across the capacitor is U0. After power failure / low level, the voltage change formula across the capacitor is:

[0094]

[0095] Like the fluorescence decay curve, it is a single exponential decay function. The LED intensity is basically linearly related to the voltage, that is, the LED brightness is also a single exponential decay function and the decay coefficient is approximately equal to RC. Using a PWM signal to drive the LED, the frequency and duty cycle of the LED can be set.

[0096] The experiment set up 10 RC circuits with capacitance values ​​of 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 μF. To eliminate idealized deviations from the theoretical model, a Microsig tBook mini oscilloscope was used to measure the voltage across a fixed resistor connected in series with the LED, indirectly obtaining the LED current signal. A low-frame-rate rolling camera was used to sample the video frequency difference of the LED driven by the RC circuit. To maximize data acquisition and facilitate subsequent processing, the camera frame rate was set to 200 FPS (with a frame rate error within ±0.02 FPS) and the exposure time was set to 1 / 5000s. The frequency of the LED excitation (lighting) drive signal was 201 Hz, with a duty cycle of 20%, to ensure that the LED light decayed as completely as possible.

[0097] Select the frequency difference parameter to calculate the step delay time:

[0098]

[0099] By calculating the frequency difference, we know that at 200FPS and 201Hz, the delay time of each frame relative to the previous frame is about 24.875μs. The video captured by the camera is separated into frames, and a sequence of LED brightness change images indexed by frame number is obtained, such as Figure 7 Calculate the step delay time generated by the video frequency difference sampling and multiply it by the index number to represent the time change. The relative time offset t of the nth frame n It can be characterized by the following formula:

[0100] t n =n×Δt(n=0,1,2,3……N)

[0101] The LED attenuation image sequence is preprocessed and segmented to extract the average grayscale signal value of the LED luminous area, and the relative offset time corresponding to each frame number and the extracted signal value are saved. The multi-cycle fusion algorithm described in (4) is used to fit the obtained attenuation data. The fitting effect is as follows: Figure 8 As shown, the fitting effect of short fluorescence lifetime data can be greatly improved.

[0102] The relative standard deviation (RSD) and relative error (RE) indicators were used to evaluate the algorithm.

[0103]

[0104]

[0105] SD is the relative standard deviation, is the average value of the measurement results; τ i is the independent fitting result for each time, and τ is the fluorescence lifetime benchmark value measured by the oscilloscope. The RSD and RE were calculated using the results of four replicate experiments at each capacitance. The calculated indicators are shown in Table 1.

[0106] Table 1

[0107]

[0108] The results show that after using the multi-cycle fusion algorithm, this method can basically achieve the fitting of simulated fluorescence lifetime at the 20μs level relatively stably.

[0109] Example 2: Fluorescence lifetime detection of long-lived fluorescent dyes

[0110] Eu(TTA)3 is a typical long-lived fluorescent dye from the lanthanide series of rare earth elements, with an excitation wavelength of 365 nm and an emission wavelength of approximately 615 nm. In this experiment, Eu(TTA)3 was used as a representative fluorescent compound and dissolved in anhydrous ethanol to prepare a 1 mg / mL solution. To facilitate image capture and sample excitation, a 96-well microtiter plate was used to hold the fluorescent solution. The time-gated fluorescence lifetime detection setup is shown in the figure below. The main components include:

[0111] (1) Fluorescence excitation light source: Common excitation light sources include lasers, tungsten lamps, and LEDs. Taking into account factors such as cost, size, ease of use, and the characteristics of fluorescent dyes, a UV LED with an emission peak of 365nm and a maximum power of 0.25W was selected as the excitation light source. A convex lens with adjustable focal length was placed in front of the LED to adjust the size of the LED spot.

[0112] (2) Filter: Placed in front of the light source to filter out light with wavelengths other than 365nm and wrapped with light-shielding tape to remove the influence of ambient light. To eliminate the influence of scattered light from the LED light source and ambient stray light on fluorescence imaging detection, a 450nm high-pass filter is also placed in the fluorescence detection window (in front of the camera lens).

[0113] (3) Dichroic beamsplitter: A dichroic beamsplitter (also called a dichroic mirror) is a special optical element that can simultaneously reflect and transmit light. In this system, a dichroic beamsplitter is placed at a 45-degree angle. Its function is to reflect the horizontally incident LED excitation light onto the surface of the object to be tested, while allowing the fluorescence generated by the fluorescent material to pass vertically, making it easier for the camera to capture the image.

[0114] In order to place the above optical components, a pure black fluorescence detection chamber was designed and 3D printed. The schematic diagram of the device structure is shown in the figure. Figure 9 Shown on the right.

[0115] The combination of the maximum integer frame rate supported by the camera, 200 FPS, and the excitation frequency of 201 Hz was selected for subsequent frequency difference time gating detection, and the exposure time was set to 1000 μs. Similarly, the fluorescence decay image sequence obtained by framing is as follows Figure 10 shown.

[0116] The same image processing process as in Example 1 was used to obtain the fluorescence signal sequence corresponding to the sampled video, and the multi-cycle data fusion algorithm was used for fitting, which was basically similar to the TCSPC fitting result, as shown in Figure 1. Figure 11 and 12 The results of three repeated experiments were 354.1μs, 350.9μs, and 366.5μs, with an average fluorescence lifetime of 357μs, close to the TCSPC fluorescence lifetime of 322μs. The relative error (RE) was approximately 10.7%, demonstrating the excellent fluorescence lifetime measurement capabilities of our self-developed frequency-difference time-gated system.

[0117] Example 3: Detecting two fluorescent materials with different lifetimes using time-resolved fluorescence

[0118] In order to verify the time resolution capability of the self-developed frequency-difference time-gated system for long-life and short-life fluorescence, two commercial microspheres for fluorescent immunolabeling provided by Guangna Dakang (Guangzhou) Biotechnology Co., Ltd. were selected: one is a rare earth fluorescent microsphere with a long fluorescence lifetime, denoted as MB-L; the other is a chemical dye fluorescent microsphere with a short fluorescence lifetime, denoted as MB-S. According to the fluorescence intensity of the original solution, the MB-L solution was diluted 50 times for standby use, and the MB-S solution was diluted 400 times for standby use. At this time, the fluorescence intensity generated by the two solutions after excitation by a 365nm monochromatic light source is similar.

[0119] Take three separate wells of a microtiter plate, labeled A, B, and C. Add 25 μL of pre-diluted MB-S stock solution and 25 μL of PBS buffer to well A; add 25 μL of MB-L stock solution and 25 μL of PBS buffer to well B; and add 25 μL each of MB-S and MB-L stock solutions to well C and mix thoroughly. This ensures that the concentrations of the two fluorescent substances in well C are consistent with those in wells A and B. Simultaneously place all three wells of the microtiter plate into a custom-developed frequency-difference time-gated system for video frequency-difference measurement.

[0120] After dividing the collected video into frames, select the representative fluorescence decay sequence images, such as Figure 13As shown. It can be seen intuitively from the image sequence that the intensity decay rates of the two fluorescent solutions MB-S and MB-L are different, and the MB-S fluorescence lifetime is short and the decay rate is fast. At the initial moment, the fluorescence of the three holes is very strong, and the C hole containing two fluorescent microspheres has the highest brightness. As time goes by, the short-lived MB-S fluorescence fades, and the A hole becomes invisible; at 200μs, in the C hole mixed with two fluorescent microspheres, only the long-lived MB-L still has fluorescence, and the brightness of the C hole is similar to that of the B hole. This very intuitively shows that the self-developed frequency difference time gating system has good time-resolved fluorescence imaging (TRFI) capabilities, which can effectively remove the background signal of short-lived fluorescence.

[0121] Extract the fluorescence signal sequences of the three wells respectively and draw Figure 14 . It can be clearly seen that the fluorescence signal change trends of the three solutions are different. The fluorescence of MB-S with a short fluorescence lifetime decays very quickly, and the fluorescence lifetime fitting result is about 40μs, while the fluorescence lifetime fitting result of MB-L with a long fluorescence lifetime is about 639μs, and the TCSPC detection result is 578μs. For well C, which is a mixture of two fluorescent microspheres, it can be seen that the early fluorescence intensity is roughly the sum of the fluorescence intensities of wells A and B; after about 200μs, with the rapid decay of MB-S fluorescence, the fluorescence trend of the mixed solution in well C becomes basically consistent with that of MB-L in well B. Time-resolved fluorescence imaging is best performed 200μs after fluorescence excitation.

[0122] The test results of two fluorescent microsphere solutions, MB-S and MB-L, and their mixtures show that the self-developed frequency-difference time-gated system can not only directly measure the fluorescence lifetime, and the detection results are well consistent with the TCSPC detection results; it can also reasonably set the fluorescence detection delay time (Delay Time) according to the background fluorescence lifetime, thereby avoiding the influence of short-lived background fluorescence signals and realizing time-resolved fluorescence imaging function.

[0123] Example 4: Detection of fluorescent immunochromatographic test strips

[0124] In order to verify the applicability of the frequency difference time gating system to the time-resolved fluorescence test paper card, this study selected the fluorescence immunochromatography test paper card (quality control card) provided by Nadakang Biotechnology Co., Ltd. as the test card. 3+ The MB-S test paper is labeled with fluorescent microspheres of short-fluorescence-lifetime chemical dyes. The T lines of the two sets of test paper cards have a certain concentration gradient.

[0125] like Figure 15The homemade frequency difference time gating device used in the test is the same as that in Examples 2 and 3. The two commercial test paper card detectors used in the comparative test are both provided by Guangna Dakang (Guangzhou) Biotechnology Co., Ltd., namely the scanning BH1000 dry fluorescence instrument and the imaging BH101 handheld fluorescence instrument.

[0126] Two sets of test strips were tested using a self-developed frequency difference time gating system. The test device and detection method were the same as in Example 3. The representative fluorescence decay image sequence of the T line of the fluorescent test strip is shown in Figure 2. Figure 16 As shown, the darkening process of the T line can be clearly observed.

[0127] The fluorescence signal sequences of the T-line position of the MB-S test paper and the MB-L test paper were extracted respectively, and the fluorescence lifetime fitting of the two test paper cards was performed using the multi-cycle fusion fitting algorithm, as shown in FIG. Figure 17 The fluorescence lifetime of the MB-L test paper card was approximately 606.4 μs, and the fluorescence lifetime of the MB-S test paper card was approximately 48.9 μs. Compared to the fluorescence lifetime of the liquid state in Example 3, the fluorescence lifetime of the microspheres fixed on the reagent card did not change much.

[0128] In order to verify the time-resolved fluorescence quantitative detection capability of the self-developed frequency-difference time-gated system, it is assumed that the long-lifetime MB-L is the actual fluorescence signal that needs to be measured, and the short-lifetime MB-S is used to simulate the background fluorescence signal. The delay time of the fluorescence detection is set to 200μs, and the fluorescence intensity signal at a delay of 200μs is obtained using the self-developed system. Then, a linear correlation comparison is performed with the test data of two commercial fluorescence card readers. The test results are as follows: Figure 18 and 19 .

[0129] From the above linear analysis results, it can be seen that after using the time-resolved fluorescence measurement method, for the MB-L series test strips with long fluorescence lifetime, the delayed fluorescence value of the self-developed system and the non-delayed fluorescence value of the commercial card reader still have a good correlation, R 2 The fluorescence signal of MB-S, which is used to simulate background fluorescence interference, is close to zero on the self-developed frequency-difference time-gated system, indicating that the self-developed system has successfully eliminated the interference of background fluorescence through the time-resolved fluorescence detection mode.

[0130] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A fluorescence lifetime measurement method based on video imaging, characterized in that: The following steps are involved: S1, using an excitation light source to illuminate the sample to be tested, while simultaneously capturing a video of the periodic luminescence decay process of the sample to be tested, and separating the video into a number of images with frame number marks; Wherein, the excitation frequency F of the excitation light source is LED and the frame rate of video capture F cam Satisfied: F LED =F cam +n, n is less than F cam A positive integer; S2. Convert all images into grayscale images, and use the average grayscale value of the luminescent area of ​​the grayscale image as the fluorescence signal intensity value; S3. Save the fluorescence signal intensity values ​​according to the frame number index, and screen out a signal decay sequence therefrom; fit the signal decay sequence using the least squares method to obtain a fitting curve of the signal intensity value changing with time, and obtain the fluorescence lifetime of the sample to be tested from the fitting curve.

2. The fluorescence lifetime measurement method based on video imaging according to claim 1, characterized in that: In step S2, the grayscale image is first subjected to noise reduction processing, and then image segmentation is performed to preliminarily extract the ROI area. Then, a region growing algorithm is used to use the detected center point of the ROI area as the initial seed point, and based on the pixel intensity difference threshold, the area is expanded outward from the initial seed point to generate the segmentation result of the ROI luminous area. Finally, the segmented area is filled with holes by the flooding algorithm, and the edges are filled by morphological dilation to finally obtain the luminous area.

3. The fluorescence lifetime measurement method based on video imaging according to claim 1, characterized in that: Step S3 obtains the best fitting result by fusing multi-cycle data, specifically including: extracting the signal attenuation sequence of each cycle, fitting the signal attenuation sequence of each cycle using the least squares method, and selecting the cycle with the best fitting effect as the reference cycle; The theoretical starting time is deduced from the decay curve based on the best fitting effect, and the starting point t of the theoretical peak of the signal intensity is calculated. start , adjust the time axis of the reference period to align with the theoretical attenuation curve; then insert the signal attenuation sequences of other periods in sequence, and re-fit, evaluate the fitting results, if they meet the preset fitting effect, complete the multi-period data fusion and obtain the fitting curve; if not, the signal attenuation sequences of other periods are time-shifted and then inserted, and the time offset is continuously adjusted until the fitting result meets the preset fitting effect to obtain the final fitting curve.

4. The fluorescence lifetime measurement method based on video imaging according to claim 3, characterized in that: The adjustment range of the time offset is within 100 μs before and after the reference period, and each time offset is 0.1-1 μs.

5. The fluorescence lifetime measurement method based on low frame rate video imaging according to claim 3, characterized in that: The fitting effect of the signal attenuation sequence of each cycle is evaluated by R2 value and / or residual standard deviation, R2≥0.8, and the residual standard deviation ranges from -2 to 2.

6. The fluorescence lifetime measurement method based on video imaging according to claim 1, characterized in that: The sample to be tested is a fluorescent material or a light emitting diode driven by an RC circuit; When the sample to be tested is a fluorescent material, the excitation light source is a laser, a tungsten lamp or an LED lamp; When the sample to be tested is a light emitting diode driven by an RC circuit, the excitation light source is the light emitting diode, that is, the video of the light emitting diode's light attenuation process is directly collected.

7. The fluorescence lifetime measurement method based on video imaging according to claim 1, characterized in that: The duty cycle D and excitation frequency F of the excitation light source LED and the fluorescence lifetime τ satisfy the following relationship: (1-D) / F LED ≥5τ; and / or, the F cam ≥30FPS, n≤10FPS.

8. A fluorescence lifetime measurement system based on video imaging, characterized in that: include: The excitation module is used to excite the sample to be tested to produce periodic luminescence decay; The video acquisition module is used to collect the video of the periodic luminescence decay process of the sample to be tested and separate the video into several images with frame number marks; the excitation frequency F of the excitation module LED and the frame rate F of the video acquisition module cam Satisfied: F LED =F cam +n, n is less than F cam A positive integer; An image processing module is used to convert all images into grayscale images, and use the average grayscale value of the luminescent area of ​​the grayscale image as the fluorescence signal intensity value; The lifetime calculation module is used to save the fluorescence signal intensity values ​​according to the frame number index and filter out the signal decay sequence; use the least squares method to fit the signal decay sequence to obtain a fitting curve of the change of the signal intensity value over time, and obtain the fluorescence lifetime of the sample to be tested from the fitting curve.

9. The fluorescence lifetime measurement system based on video imaging according to claim 8, characterized in that: The image processing module also includes: a noise reduction processing unit, configured to perform noise reduction processing on the grayscale image; An image segmentation unit is used to perform image segmentation on the grayscale image after noise reduction to extract the ROI area; The region growing segmentation unit is used to use the region growing algorithm to take the center point of the detected ROI region as the initial seed point, and based on the pixel intensity difference threshold, expand the region outward from the initial seed point to generate the segmentation result of the ROI luminous area; The segmented area filling unit is used to fill the holes in the segmented area by using a flooding algorithm and to fill the edges by using morphological dilation, so as to finally obtain the luminous area.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fluorescence lifetime measurement method based on video imaging are implemented according to any one of claims 1 to 7.

Images