A fluorescence-based responsive blood vessel imaging method and system

Abstract

The application relates to the field of imaging technology and discloses a fluorescence-based response blood vessel imaging method and system.The method comprises the following steps: acquiring a fluorescence image sequence of a living target region and a modulation reference sequence corresponding to a periodic modulation condition, constructing a pixel-level fluorescence time sequence matrix according to spatial positions, time aligning, constructing a pixel-level fluorescence time sequence matrix according to spatial positions, determining a target response frequency according to a periodic stimulation signal applied to the living target region, and generating an in-phase reference vector and a quadrature reference vector required for digital frequency domain demodulation based on the target response frequency and an image acquisition time sequence; performing inner product operation on the pixel-level fluorescence time sequence matrix and the in-phase reference vector and the quadrature reference vector respectively, and obtaining response parameters to determine the response amplitude and the response phase of each pixel.The application can be used for extracting response blood vessels related to the periodic modulation condition from a visible light or near-infrared region fluorescence image sequence.

Description

Technical Field

[0001] This invention relates to the field of imaging technology, specifically to a fluorescence-based response vascular imaging method and system. Background Technology

[0002] Visible or near-infrared fluorescence imaging technology can acquire vascular structure information and time-varying optical signals by continuously acquiring fluorescence image sequences of a target area in a living organism. Existing processing methods typically employ single-frame enhancement, time averaging, maximum intensity projection, or spectral analysis to characterize vascular structure or overall signal changes in fluorescence images, thereby improving vascular visualization or analyzing dynamic changes in the target area. However, for weak vascular response signals induced by periodic stimulation, physiological modulation, or other predetermined conditions, the response signals are usually characterized by low amplitude and non-uniform spatial distribution. During actual acquisition, they may still be affected by factors such as static vascular background, tissue scattering, low-frequency signal drift, motion artifacts, and noise in adjacent frequency bands.

[0003] In fluorescence image sequence processing, if an accurate temporal correspondence is not established between the image acquisition time and the periodic stimulus signal, or if the analysis is mainly based on the overall intensity changes of the image, it may be difficult to effectively extract response information that changes synchronously with the stimulus conditions at a specific spatial location. Furthermore, while frequency domain analysis methods can identify specific frequency components to some extent, they may still be affected by noise frequency components and periodic fluctuations in non-target regions when detecting weak response signals. To address these issues, corresponding reference information can be constructed by combining the periodic stimulus signal, and the fluorescence sequence can be further analyzed using response frequency, phase characteristics, and spatial structure information to improve the accuracy of weak response signal extraction and the reliability of response region localization. Summary of the Invention

[0004] This invention aims to address the following problems in existing vascular imaging technologies: Existing fluorescence imaging-based vascular detection methods mainly rely on differences in fluorescence intensity at single time points for vascular identification. During in vivo detection, these methods are easily affected by tissue background fluorescence, optical scattering, probe distribution differences, and random noise, making it difficult to accurately separate weak vascular response signals from background signals. At the same time, existing methods lack technical means to analyze the dynamic response characteristics generated by periodic physiological stimuli or externally modulated stimuli, making it impossible to effectively obtain the time response relationship between vascular regions and stimulation signals, thus reducing the accuracy of vascular function status detection.

[0005] To address the above problems, this application provides a fluorescence-based response vascular imaging method, comprising:

[0006] S1. Under periodic stimulation, the visible light or near-infrared fluorescence image sequence of the target area of ​​the living organism is acquired synchronously, and the stimulation timing information corresponding to the periodic stimulation is recorded. The acquisition time point and the stimulation timing information are synchronized to obtain the synchronous fluorescence sequence.

[0007] S2. Extract the fluorescence intensity of each image frame in the fluorescence sequence according to the same spatial location, and arrange them according to the acquisition time to construct a pixel-level fluorescence time sequence matrix;

[0008] S3. Determine the target response frequency based on the periodic stimulation signal applied to the live target area, and generate the in-phase reference vector and orthogonal reference vector required for digital frequency domain demodulation based on the target response frequency and the image acquisition time series.

[0009] S4. Perform inner product operation between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain response parameters, and determine the response amplitude and response phase of each pixel based on the response parameters;

[0010] S5. Determine the target response frequency based on the periodic stimulation signal applied to the live target area, determine the periodic response consistency based on the response amplitude and response phase of each periodic data segment, and calculate the response score by combining the noise characteristics of the frequency band adjacent to the target frequency.

[0011] S6. Determine the response judgment threshold based on the response score of the background pixels, and determine the pixels that reach the response judgment threshold as response pixels;

[0012] S7. Extract vascular structure features from the visible light or near-infrared fluorescence image sequence to obtain a vascular structure mask, and spatially fuse the response pixels with the vascular structure mask to generate a response vascular image and / or a response vascular mask.

[0013] Specifically, S1 includes:

[0014] Obtain fluorescence image sequences in the visible or near-infrared region and their corresponding image timestamp sequences;

[0015] Obtain the modulation reference sequence and reference timestamp sequence corresponding to the periodic modulation conditions;

[0016] A clock mapping model is established based on the image timestamp sequence and the reference timestamp sequence to obtain the time offset parameter;

[0017] The modulation reference sequence is interpolated and resampled according to the time offset parameter to generate a reference vector corresponding to each image frame;

[0018] The temporal alignment error between the reference vector and each image frame is calculated, and data with temporal alignment errors less than a preset error threshold are combined into the fluorescence sequence.

[0019] Specifically, S2 includes:

[0020] A spatial coordinate index matrix is ​​established based on the image size of the fluorescence sequence, and a pixel index is configured for each spatial location;

[0021] The fluorescence intensity of each image frame is extracted according to the pixel index, and each image frame is converted into a frame intensity vector;

[0022] The frame intensity vector is filtered according to a preset saturation threshold and a missing rate threshold to generate a set of valid pixel indices;

[0023] The intensity vectors of each frame are arranged sequentially according to the acquisition time, and the fluorescence intensity corresponding to the effective pixel index set is extracted to obtain the initial temporal matrix;

[0024] The initial timing matrix is ​​set as a pixel-level fluorescence timing matrix with rows corresponding to spatial locations and columns corresponding to acquisition times.

[0025] Specifically, S3 includes:

[0026] The target response frequency is determined based on the periodic stimulation signals applied to the target area in a living organism;

[0027] Based on the target response frequency and the acquisition time points corresponding to each image frame, generate in-phase reference sequences and orthogonal reference sequences corresponding to the target response frequency, respectively.

[0028] The in-phase reference sequence and the quadrature reference sequence are normalized to obtain the in-phase reference vector and quadrature reference vector required for digital frequency domain demodulation.

[0029] The in-phase reference vector and the quadrature reference vector have a 90° phase difference and are used to extract the in-phase response component and the quadrature response component of the fluorescence signal at the target response frequency, respectively.

[0030] Specifically, S4 includes:

[0031] The pixel-level fluorescence time series matrix is ​​subjected to mean removal and normalization processing row by row to obtain a standard time series matrix;

[0032] The in-phase response vector is obtained by performing an inner product operation between each row vector of the standard timing matrix and the in-phase reference vector.

[0033] The orthogonal response vector is obtained by performing a matrix inner product between the standard time series matrix and the orthogonal reference vector.

[0034] The in-phase response vector and the orthogonal response vector are combined into a pixel response parameter matrix;

[0035] The response amplitude and response phase are calculated based on the in-phase and quadrature response values ​​corresponding to each pixel, and the response amplitude matrix and response phase matrix are obtained by arranging them according to spatial coordinates.

[0036] Specifically, S5 includes:

[0037] Based on the periodic boundaries of the periodic stimulus, the pixel-level fluorescence time series matrix is ​​divided into multiple periodic sub-matrices;

[0038] Calculate the response amplitude and response phase of each pixel in each periodic submatrix to construct the periodic response feature matrix;

[0039] The periodic response consistency is calculated based on the amplitude and phase dispersion of each pixel over multiple periods.

[0040] Extract adjacent frequency components on both sides of the target frequency that are outside the preset guard band, and calculate the adjacent frequency noise characteristics of each pixel;

[0041] The response score for each pixel is obtained by weighting the ratio of the response amplitude to the adjacent-channel noise feature and combining it with the periodic response consistency.

[0042] Specifically, S6 includes:

[0043] The response scores of each pixel are arranged according to their spatial coordinates to obtain the response score matrix;

[0044] Background pixels are selected from a preset background area or a low-response score area to construct a background score sequence;

[0045] Calculate the median and dispersion of the background rating sequence to obtain the background noise statistical parameters;

[0046] Calculate the response judgment threshold based on the background noise statistical parameters and the preset threshold coefficient;

[0047] The pixels in the response scoring matrix that reach the response determination threshold are identified as response pixels, and a response pixel mask is generated.

[0048] Specifically, S7 includes:

[0049] The visible light or near-infrared fluorescence image sequence is time-averaged or projected with maximum intensity to obtain a reference image of the vascular structure.

[0050] Multi-scale vascular enhancement is performed on the vascular structure reference image to obtain a vascular structure feature matrix;

[0051] The vascular structure feature matrix is ​​segmented according to the structural segmentation threshold, and a vascular structure mask is obtained through morphological processing.

[0052] The response pixel mask corresponding to the response pixel is intersected or weighted and fused with the blood vessel structure mask to obtain the response blood vessel feature matrix.

[0053] Based on the response vascular feature matrix, the spatial location and response intensity are recovered to generate a response vascular image and / or a response vascular mask.

[0054] This invention also provides a fluorescence-based response vascular imaging system, applicable to the aforementioned fluorescence-based response vascular imaging method, comprising:

[0055] The periodic stimulation and image acquisition module is used to apply periodic stimulation to the live target area and simultaneously acquire the visible light or near-infrared fluorescence image sequence of the live target area, while recording the stimulation timing information corresponding to the periodic stimulation.

[0056] A time synchronization module, connected to the periodic stimulus and image acquisition module, is used to perform time synchronization based on the acquisition time point corresponding to the visible light or near-infrared fluorescence image sequence and the stimulus timing information to obtain a synchronized fluorescence sequence.

[0057] A pixel-level time-series matrix construction module, connected to the time synchronization module, is used to extract the fluorescence intensity corresponding to each image frame in the synchronized fluorescence sequence according to the same spatial location, and arrange them according to the acquisition time order to construct a pixel-level fluorescence time-series matrix.

[0058] A digital frequency domain demodulation module, connected to the pixel-level time-series matrix construction module, is used to determine the target response frequency based on the periodic stimulation signal applied to the live target region, and generate an in-phase reference vector and an orthogonal reference vector based on the target response frequency and the image acquisition time series; the pixel-level fluorescence time-series matrix is ​​then used to perform inner product operations with the in-phase reference vector and the orthogonal reference vector to obtain response parameters corresponding to the target response frequency, and the response amplitude and response phase of each pixel are determined based on the response parameters;

[0059] The periodic response analysis module, connected to the digital frequency domain demodulation module, divides the pixel-level fluorescence time series matrix into multiple periodic data segments according to the periodic boundaries corresponding to the periodic stimulus. It determines the periodic response consistency based on the response amplitude and response phase corresponding to each periodic data segment, and calculates the response score of each pixel by combining the noise characteristics of the frequency band adjacent to the target response frequency.

[0060] The response pixel determination module is connected to the periodic response analysis module. It is used to determine the response determination threshold based on the response score corresponding to the background pixel, and to determine the pixel that reaches the response determination threshold as the response pixel, and to generate a response pixel mask.

[0061] The vascular structure fusion module, connected to the periodic stimulation and image acquisition module and the response pixel determination module, is used to extract vascular structure features from the visible light or near-infrared fluorescence image sequence, generate a vascular structure mask, and spatially fuse the response pixel mask with the vascular structure mask to generate a response vascular image and / or a response vascular mask.

[0062] Specifically, the time synchronization module includes:

[0063] The timestamp acquisition unit is used to acquire the image timestamp sequence corresponding to the visible light or near-infrared region fluorescence image sequence, as well as the stimulus reference sequence and reference timestamp sequence corresponding to the periodic stimulus;

[0064] A clock mapping unit is used to establish a clock mapping model based on the image timestamp sequence and the reference timestamp sequence to obtain time offset parameters;

[0065] A reference resampling unit is used to interpolate and resample the stimulus reference sequence according to the time offset parameter to generate a reference vector corresponding to each image frame.

[0066] An error filtering unit is used to calculate the time alignment error between the reference vector and each image frame, and to filter and obtain a synchronous fluorescence sequence according to a preset error threshold.

[0067] Specifically, the pixel-level temporal matrix construction module includes:

[0068] A spatial indexing unit is used to establish a spatial coordinate index matrix based on the image size of the synchronized fluorescence sequence;

[0069] A fluorescence intensity extraction unit is used to extract the fluorescence intensity of each image frame at a corresponding spatial location based on the spatial coordinate index matrix.

[0070] The effective pixel filtering unit is used to filter effective pixels based on a preset saturation threshold and a missing rate threshold.

[0071] The time-series arrangement unit is used to arrange the fluorescence intensity data corresponding to each image frame in the order of acquisition time, forming a pixel-level fluorescence time-series matrix.

[0072] Specifically, the digital frequency domain demodulation module includes:

[0073] The target frequency determination unit is used to determine the target response frequency based on the stimulation period of the periodic stimulus signal.

[0074] The reference vector generation unit is used to generate an in-phase reference sequence and an orthogonal reference sequence based on the target response frequency and the acquisition time point corresponding to each image frame, and to normalize the in-phase reference sequence and the orthogonal reference sequence to obtain an in-phase reference vector and an orthogonal reference vector.

[0075] The response component calculation unit is used to perform inner product operations between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain the in-phase response component and the orthogonal response component.

[0076] The response parameter calculation unit is used to calculate the response amplitude and response phase based on the in-phase response component and the quadrature response component.

[0077] Specifically, the periodic response analysis module includes:

[0078] The periodic division unit is used to divide the pixel-level fluorescence time series matrix into multiple periodic sub-matrices according to the periodic boundaries corresponding to the periodic stimulus;

[0079] The consistency analysis unit is used to calculate the periodic response consistency based on the changes in response amplitude and response phase of corresponding pixels within multiple periods.

[0080] The noise analysis unit is used to extract the frequency components of the frequency band adjacent to the target response frequency and calculate the adjacent frequency noise characteristics.

[0081] The scoring calculation unit is used to perform weighted calculations based on the response amplitude, adjacent channel noise characteristics, and periodic response consistency to obtain the response score.

[0082] Specifically, the response pixel determination module includes:

[0083] The background statistics unit is used to obtain the response scores corresponding to the pixels in the background region and calculate the statistical parameters of the background score sequence.

[0084] A threshold generation unit is used to generate a response judgment threshold based on the statistical parameters and a preset threshold coefficient.

[0085] The response mask generation unit is used to filter response pixels according to the response determination threshold and generate a response pixel mask.

[0086] Specifically, the vascular structure fusion module includes:

[0087] The vascular reference image generation unit is used to perform time averaging or maximum intensity projection on visible light or near-infrared fluorescence image sequences to obtain vascular structure reference images.

[0088] The vascular enhancement unit is used to perform multi-scale vascular enhancement on the vascular structure reference image to obtain a vascular structure feature matrix.

[0089] The blood vessel segmentation unit is used to segment the blood vessel structure feature matrix according to the structure segmentation threshold and generate a blood vessel structure mask through morphological processing.

[0090] The fusion unit is used to perform intersection operation or weighted fusion of the response pixel mask and the vascular structure mask to generate a response vascular image and / or a response vascular mask.

[0091] Compared with the prior art, this application has at least the following beneficial effects:

[0092] This invention synchronizes visible or near-infrared fluorescence image sequences acquired under periodic stimulation, determines the target response frequency based on the stimulation signal, and generates in-phase and orthogonal reference vectors using digital frequency domain demodulation. Response analysis is then performed on pixel-level fluorescence time-series signals to extract response components that change synchronously with the periodic stimulation. Simultaneously, the response signal is evaluated by combining response amplitude, response phase, periodic response consistency, and adjacent frequency band noise characteristics. Furthermore, spatial fusion is performed using a vascular structure mask to locate the responding vascular region. This method can, to a certain extent, reduce the impact of static background, random noise, and non-target periodic fluctuations on the detection results, improve the reliability of weak vascular response signal identification, and enhance the spatial interpretation capability of the responding vascular imaging results. Attached Figure Description

[0093] Figure 1 This is a flowchart of the fluorescence-based response vascular imaging method proposed in this invention;

[0094] Figure 2 This is a timing diagram of the fluorescence-based response vascular imaging method proposed in this invention;

[0095] Figure 3 This is a flowchart of the response scoring method for the fluorescence-based response vascular imaging method proposed in this invention.

[0096] Figure 4 This is a timing diagram of periodic modulation, visible light or near-infrared fluorescence image acquisition and data processing according to the present invention;

[0097] Figure 5 These are original visible or near-infrared fluorescent blood vessel images under electrical stimulation according to the present invention;

[0098] Figure 6 This is a vascular response image extracted from the target frequency domain under electrical stimulation according to the present invention;

[0099] Figure 7This is a raw visible or near-infrared fluorescent blood vessel image under light stimulation according to the present invention;

[0100] Figure 8 This is a response vascular image extracted in the frequency domain of the target frequency under light stimulation according to the present invention;

[0101] Figure 9 This is a raw visible or near-infrared fluorescent blood vessel image under acoustic stimulation according to the present invention;

[0102] Figure 10 This is a vascular response image extracted in the frequency domain of the target frequency under acoustic stimulation according to the present invention. Detailed Implementation

[0103] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0104] The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of the invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0105] Example 1:

[0106] like Figures 1-2 As shown, a fluorescence-based response vascular imaging method includes:

[0107] S1. Under periodic stimulation, the visible light or near-infrared fluorescence image sequence of the target area of ​​the living organism is acquired synchronously, and the stimulation timing information corresponding to the periodic stimulation is recorded. The acquisition time point and the stimulation timing information are synchronized to obtain the synchronous fluorescence sequence.

[0108] S2. Extract the fluorescence intensity of each image frame in the fluorescence sequence according to the same spatial location, and arrange them according to the acquisition time to construct a pixel-level fluorescence time sequence matrix;

[0109] S3. Determine the target response frequency based on the periodic stimulation signal applied to the live target area, and generate the in-phase reference vector and orthogonal reference vector required for digital frequency domain demodulation based on the target response frequency and the image acquisition time series.

[0110] S4. Perform inner product operation between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain response parameters, and determine the response amplitude and response phase of each pixel based on the response parameters;

[0111] S5. Divide the data into multiple periodic data segments according to the periodic envelope, determine the periodic response consistency based on the response amplitude and response phase of each periodic data segment, and calculate the response score in combination with the noise characteristics of the frequency band adjacent to the target frequency.

[0112] S6. Determine the response judgment threshold based on the response score of the background pixels, and determine the pixels that reach the response judgment threshold as response pixels;

[0113] S7. Extract vascular structure features from the visible light or near-infrared fluorescence image sequence to obtain a vascular structure mask, and spatially fuse the response pixels with the vascular structure mask to generate a response vascular image and / or a response vascular mask.

[0114] In this embodiment, to address the problems of background fluorescence interference, tissue scattering, and difficulty in accurately identifying vascular response signals from continuous fluorescence changes during in vivo vascular imaging, this method combines periodic stimulation with visible or near-infrared fluorescence imaging. By acquiring the periodic fluorescence changes generated in the vascular region under stimulation, and using digital frequency domain demodulation to extract the response signal corresponding to the stimulation period, the identification of the responding vascular region is achieved.

[0115] In the specific implementation process, the live target area is first continuously acquired using a visible light or near-infrared fluorescence imaging device. The imaging device includes an excitation light source, a fluorescence detection module, an image acquisition module, and a timing control module. The excitation light source provides illumination conditions to the live target area capable of stimulating visible light or near-infrared fluorescence signals. The fluorescence detection module receives the fluorescence signals returned from the target area. The image acquisition module acquires fluorescence images at set time intervals. The timing control module outputs periodic stimulation signals and records the timing of signal changes.

[0116] When periodic stimulation is applied to a target area in vivo, the blood flow, blood distribution, or tissue optical properties within the vascular region change with the stimulation cycle, causing fluorescence intensity changes in the vascular region corresponding to the stimulation cycle. The image acquisition module simultaneously acquires a continuous sequence of fluorescence images during stimulation, while also recording the time information of each stimulation change. Based on the correspondence between image acquisition time and stimulation time, the fluorescence image sequence is time-calibrated so that each fluorescence image corresponds to a specific time position within the stimulation cycle, forming a fluorescence sequence synchronized with the stimulation process.

[0117] Subsequently, the synchronized fluorescence sequence underwent spatial alignment processing to ensure that pixels at the same location in image frames acquired at different times correspond to the same spatial location within the live target region. For each pixel location, fluorescence intensity data corresponding to that location was extracted according to the image acquisition order, and the fluorescence intensity data at different time points were arranged chronologically to form a pixel-level fluorescence temporal matrix. This matrix is ​​used to record the changes in fluorescence intensity over time at each spatial location within the live target region.

[0118] The target response frequency is determined based on the periodic stimulation signal applied to the target area in vivo. Since there is a temporal correspondence between the fluorescence response changes in the vascular region and the periodic stimulation, a reference signal for digital frequency domain demodulation is generated based on the variation pattern of the stimulation signal. Specifically, based on the time position corresponding to each image acquisition, the reference signal value corresponding to that time position is calculated, forming both an in-phase reference sequence synchronized with the stimulation change and an orthogonal reference sequence with a fixed phase difference from the stimulation change.

[0119] The fluorescence intensity variation data corresponding to each pixel position in the pixel-level fluorescence time series matrix is ​​used to calculate the inner product with both in-phase and orthogonal reference sequences. Through this calculation, the response component consistent with the target stimulus frequency in the fluorescence intensity variation data is extracted, yielding response parameters representing the response state at that pixel position. Based on the relationship between the in-phase and orthogonal calculation results, the response amplitude and response phase corresponding to that pixel position are determined. The response amplitude represents the signal intensity corresponding to the periodic stimulus in the fluorescence variation at that position, and the response phase represents the correspondence between the fluorescence variation at that position and the stimulus signal occurrence time.

[0120] To determine whether a pixel location exhibits stable stimulus-response characteristics, continuously acquired fluorescence data is divided according to stimulation cycles, with each cycle corresponding to an independent set of response data. The response amplitude and phase of the same pixel location within each cycle are analyzed, and the consistency of the periodic response is determined based on the changes in response amplitude and phase across different cycles. When the same pixel location consistently exhibits the corresponding response amplitude across different stimulation cycles, and the response phase changes meet predetermined conditions, that pixel location is considered to possess periodic response characteristics.

[0121] Simultaneously, signal analysis is performed on the frequency region near the target response frequency to obtain noise characteristics within this region. Based on the periodic response consistency results and the noise characteristics near the target response frequency, a response score is calculated for each pixel location. The response score is used to comprehensively represent whether the fluorescence change at that pixel location satisfies the periodic stimulus response condition.

[0122] Furthermore, a background region without vascular structures is selected in the fluorescence image, and the response score corresponding to each pixel location within the background region is obtained. A response determination threshold is determined based on the data distribution of the background region response scores. Specifically, the threshold can be determined based on the average value and dispersion of the background region response scores, as well as a preset determination rule. When the response score of the target pixel location reaches the response determination threshold, that pixel location is determined as the response pixel.

[0123] After obtaining the response pixels, vascular structures are extracted from the visible or near-infrared fluorescence image sequences. Vascular regions in the fluorescence images are identified through vascular enhancement, edge feature extraction, image segmentation, or image recognition methods based on trained models, resulting in a vascular structure mask representing the spatial distribution of blood vessels.

[0124] Finally, the positions of the response pixels are spatially matched with the vascular structure mask, and the response pixels located within the vascular structure region are fused to ensure that the response signals correspond to the spatial structure of the blood vessels, generating a response vascular image and / or a response vascular mask. This process allows for the display of fluorescence response information generated by periodic stimuli in the vascular region while preserving the spatial location of the blood vessels, achieving vascular imaging based on fluorescence response features.

[0125] Example 2:

[0126] In one possible implementation, S1 includes: acquiring a visible light or near-infrared fluorescence image sequence and a corresponding image timestamp sequence; acquiring a modulation reference sequence and a reference timestamp sequence corresponding to a periodic modulation condition; establishing a clock mapping model based on the image timestamp sequence and the reference timestamp sequence to obtain a time offset parameter; interpolating and resampling the modulation reference sequence based on the time offset parameter to generate a reference vector corresponding to each image frame; calculating the time alignment error between the reference vector and each image frame, and combining data with a time alignment error less than a preset error threshold into the fluorescence sequence.

[0127] This embodiment addresses the issue of time asynchrony between the periodic stimulus signal and the acquisition process of fluorescence images in the visible or near-infrared regions. By acquiring image acquisition time information and periodic modulation reference time information, a time correspondence between the two is established. The modulation reference sequence is then time-corrected based on the time offset, thereby establishing a correspondence between the periodic modulation signal and each fluorescence image frame, thus improving the time matching accuracy in the subsequent response signal extraction process.

[0128] In the specific implementation process, a continuous sequence of fluorescence images of the live target area is first acquired using a visible light or near-infrared fluorescence imaging device. During each image acquisition, the image acquisition module records the acquisition time corresponding to the current image frame, and the acquisition times of each image frame are stored in the order of image acquisition to form an image timestamp sequence. This image timestamp sequence is used to indicate the actual time and location at which each fluorescence image was acquired.

[0129] Simultaneously, during the application of periodic modulation stimuli, the stimulation control module records the change time of the modulation signal and the corresponding modulation state value. The modulation states corresponding to different time points are arranged in chronological order to form a modulation reference sequence, and the occurrence time corresponding to each modulation state is recorded to obtain a reference timestamp sequence. The modulation reference sequence is used to represent the state information of the periodic stimulus changing over time.

[0130] Because the image acquisition clock and the stimulus control clock may have differences in start time, operating frequency, and triggering delay, a clock mapping relationship needs to be established based on the image timestamp sequence and the reference timestamp sequence. Specifically, multiple time nodes during the image acquisition process are matched with corresponding time nodes in the modulation reference sequence, the time difference of the same event in the two time sequences is calculated, and a time offset parameter is determined based on the changing relationship between the multiple matched time nodes. The time offset parameter is used to represent the time deviation between the image acquisition time and the modulation reference time.

[0131] After obtaining the time offset parameter, the modulation reference sequence is interpolated and resampled based on this parameter. Specifically, based on the acquisition time corresponding to each fluorescence image, the modulation state data corresponding to that time position is searched in the modulation reference sequence. When the image acquisition time falls between two recorded modulation times, the modulation state at that time position is interpolated based on the modulation state change relationship between two adjacent modulation times to obtain modulation reference data consistent with the current image frame time. Through the above processing, each fluorescence image has reference data corresponding to its time position, forming a reference vector corresponding to each image frame.

[0132] Subsequently, the temporal alignment error between the generated reference vector and the corresponding fluorescence image frame is calculated. Specifically, each time position in the reference vector is compared with the acquisition time of the corresponding image frame, the time difference between the two is calculated, and the time difference for each image frame is statistically processed to obtain the temporal alignment error between each set of image data and the reference data. This temporal alignment error is used to represent the degree of deviation between the image acquisition time and the periodic modulation reference time.

[0133] The preset error threshold is determined based on the imaging device's acquisition frequency, stimulus period length, and the time synchronization accuracy requirements of subsequent frequency domain demodulation. Specifically, the actual deviation data between the image trigger time and stimulus trigger time under continuous acquisition conditions is first obtained. Based on this deviation data, the maximum allowable time offset under normal operating conditions is determined, and this maximum allowable time offset is set as the preset error threshold. When the calculated time alignment error is less than the preset error threshold, the corresponding image frame and modulation reference data are considered to meet the time synchronization condition; when the time alignment error reaches or exceeds the preset error threshold, the corresponding data does not participate in the subsequent construction of the synchronized fluorescence sequence.

[0134] Finally, the fluorescence image frames that meet the time alignment conditions are combined with the corresponding reference data in chronological order to form a fluorescence sequence. Each fluorescence image in this sequence corresponds to a specific time state in the periodic modulation process, enabling subsequent analysis based on the accurate corresponding time information when extracting the response signal according to the stimulus frequency. Through the above processing, time synchronization between the visible or near-infrared fluorescence image sequence and the periodic modulation reference signal is achieved.

[0135] Example 3:

[0136] In one possible implementation, S2 includes: establishing a spatial coordinate index matrix based on the image size of the fluorescence sequence, and configuring pixel indices for each spatial location; extracting the fluorescence intensity of each image frame according to the pixel indices, and converting each image frame into a frame intensity vector; filtering the frame intensity vectors according to a preset saturation threshold and a missing rate threshold to generate a set of effective pixel indices; arranging the frame intensity vectors sequentially according to the acquisition time, and extracting the fluorescence intensity corresponding to the set of effective pixel indices to obtain an initial time series matrix; and setting the initial time series matrix as a pixel-level fluorescence time series matrix with rows corresponding to spatial locations and columns corresponding to acquisition times.

[0137] This embodiment is used to convert a fluorescence image sequence after time synchronization processing into a pixel-level fluorescence time series matrix with spatial location correspondence and temporal variation relationship, so that the subsequent digital frequency domain demodulation process can independently analyze the fluorescence changes at each spatial location.

[0138] In the specific implementation process, the image size information of each image frame in the fluorescence sequence is first obtained, including the number of pixel rows and columns. A two-dimensional spatial coordinate index matrix is ​​established based on the number of rows and columns in the image. Specifically, a unique pixel index is assigned to each pixel position according to the pixel arrangement order. The pixel index corresponds to a fixed spatial coordinate position in the image, so that image frames acquired at different acquisition times can correspond to the same spatial position through the same pixel index.

[0139] After establishing the spatial coordinate index matrix, the fluorescence intensity values ​​at corresponding positions in each image frame of the fluorescence sequence are read sequentially according to the pixel index. Specifically, for a single image frame, the fluorescence grayscale value or fluorescence intensity value corresponding to each pixel position is extracted from the image data storage area according to the pixel index, and rearranged according to the pixel index order, converting the pixel intensity data in the two-dimensional image into a one-dimensional data sequence to form a frame intensity vector. Each data point in the frame intensity vector corresponds to the fluorescence intensity at a spatial location.

[0140] To eliminate the influence of saturated and missing pixels on subsequent response analysis, frame intensity vectors are filtered based on preset saturation and missing rate thresholds. The saturation threshold is determined according to the effective dynamic range of the image acquisition device. Specifically, based on the maximum effective response value of the fluorescence detector output signal, fluorescence intensity values ​​close to or reaching the detector's full-scale output are set as the saturation criterion. When the fluorescence intensity value corresponding to a pixel location reaches this saturation criterion, that pixel is marked as a saturated pixel.

[0141] The missing rate threshold is determined based on the number of valid acquisitions of pixels in the image sequence. Specifically, the number of valid fluorescence intensity data for each pixel location in all image frames is counted, and the number of invalid data is compared with the total number of image frames to obtain the data missing ratio for that pixel location. When the data missing ratio reaches the preset missing rate threshold, that pixel location is excluded from subsequent analysis data. The missing rate threshold is determined based on the minimum number of valid time samples required for subsequent response analysis, for example, based on the number of complete stimulus cycles required for digital frequency domain demodulation, ensuring that the retained pixel locations have continuous time sampling data. Based on the above saturated pixel screening results and missing data screening results, the pixel indices in the frame intensity vector are filtered, and the pixel indices that meet the requirements of fluorescence intensity not reaching saturation and time series data integrity are collected and stored to form a valid pixel index set. This valid pixel index set is used to determine the spatial location participating in subsequent fluorescence time series analysis. Subsequently, the frame intensity vectors are arranged according to the image acquisition time order, so that the fluorescence intensity data corresponding to different acquisition times are stored according to the time sequence. Based on the valid pixel index set, the fluorescence intensity data of the corresponding pixel location is extracted from the arranged frame intensity vectors and reorganized according to pixel location and acquisition time to form an initial time series matrix. After forming the initial time series matrix, the data arrangement of the matrix is ​​adjusted so that each row corresponds to a pixel index at a fixed spatial location, and each column corresponds to an image acquisition time point. Specifically, the data in the same row represents the fluorescence intensity change at the same spatial location under different acquisition times, and the data in the same column represents the fluorescence intensity distribution at different spatial locations under the same acquisition time, thus forming a pixel-level fluorescence time series matrix.

[0142] Through the above processing, the original fluorescence image sequence is converted into a data structure with spatial location as the index and acquisition time as the variable dimension, so that each pixel location has an independent fluorescence time.

[0143] Example 4:

[0144] In one possible implementation, S3 specifically includes: determining the target response frequency based on the periodic stimulation signal applied to the live target region; generating an in-phase reference sequence and an orthogonal reference sequence corresponding to the target response frequency based on the target response frequency and the acquisition time point corresponding to each image frame; normalizing the in-phase reference sequence and the orthogonal reference sequence to obtain the in-phase reference vector and the orthogonal reference vector required for digital frequency domain demodulation; wherein the in-phase reference vector and the orthogonal reference vector have a 90° phase difference and are used to extract the in-phase response component and the orthogonal response component of the fluorescence signal at the target response frequency, respectively.

[0145] This embodiment establishes a reference signal sequence corresponding to the target response frequency based on the periodic stimulus signal, and uses in-phase reference vectors and orthogonal reference vectors with a fixed phase difference to perform matching analysis on the fluorescence time series, so that the response component in the fluorescence signal that is consistent with the periodic stimulus frequency can be extracted.

[0146] In the specific implementation process, the periodic stimulation signal applied to the target area of ​​the living organism is first acquired. This periodic stimulation signal is generated by a stimulation control module, which records the data of the stimulation signal changing over time. Based on the temporal change process of the stimulation signal, the time interval between two adjacent positions of the same state change is calculated to determine the time length corresponding to one cycle of the stimulation signal. The target response frequency corresponding to the stimulation signal is determined based on this cycle length; this target response frequency is used to represent the periodic response frequency that needs to be extracted in subsequent fluorescence signal analysis.

[0147] After determining the target response frequency, a reference signal sequence is generated based on the acquisition time points corresponding to each image frame. Specifically, the acquisition time point of each image frame is used as the time position for reference signal generation. Based on the correspondence between this time position and the periodic stimulus signal, the specific position of this acquisition time within the stimulus cycle is determined. Furthermore, according to the periodic variation pattern corresponding to the target response frequency, a corresponding reference signal value is assigned to each acquisition time point, thus forming an in-phase reference sequence with the same number of fluorescence image frames. Simultaneously with generating the in-phase reference sequence, an orthogonal reference sequence is generated based on the same acquisition time points. Specifically, the periodic reference signal corresponding to the same target response frequency is shifted by 90° in the phase direction, so that each reference data in the orthogonal reference sequence corresponds to the phase shift state at the same time position in the in-phase reference sequence. In this way, the in-phase reference sequence and the orthogonal reference sequence respectively represent two mutually perpendicular phase directions corresponding to the target response frequency.

[0148] Subsequently, the generated in-phase and quadrature reference sequences are normalized. Specifically, the sum of squared amplitudes of all reference data in each reference sequence is calculated, and the square root of the result is taken to obtain the amplitude reference value for the corresponding reference sequence. Then, each reference data in the reference sequence is divided by the corresponding amplitude reference value to ensure that the processed reference data maintains a uniform amplitude range. After normalization, the in-phase reference sequence is converted into an in-phase reference vector, and the quadrature reference sequence is converted into a quadrature reference vector. Through normalization, the inner product calculation of the reference vector and pixel-level fluorescence time-series data during subsequent digital frequency domain demodulation is unaffected by changes in the reference signal amplitude, and the response component is extracted only based on the phase correspondence between the fluorescence change signal and the target response frequency.

[0149] In the digital frequency domain demodulation process, the in-phase reference vector is used to extract the response data in the fluorescence signal that is in the same phase direction as the periodic stimulus signal. Specifically, the fluorescence intensity change data corresponding to the pixel position is arranged in the order of acquisition time, and multiplied item by item with the data at the corresponding time position in the in-phase reference vector. Then, all the multiplication results are accumulated to obtain the in-phase response component of the pixel position at the target response frequency.

[0150] Orthogonal reference vectors are used to extract response data in the fluorescence signal that has a 90° phase difference with the periodic stimulus signal. Specifically, the fluorescence intensity change data corresponding to the same pixel position is multiplied item by item with the data at the corresponding time position in the orthogonal reference vector, and all the product results are accumulated to obtain the orthogonal response component at the target response frequency for that pixel position.

[0151] In this design, the in-phase reference vector and the orthogonal reference vector maintain a 90° phase difference, allowing the fluorescence change signal corresponding to the target response frequency to be mapped to two independent response directions. Subsequent combined analysis of the in-phase and orthogonal response components determines the response amplitude and phase at the corresponding pixel location, thereby identifying the fluorescence change region associated with periodic stimulation.

[0152] This implementation generates a matching reference vector based on the stimulus signal frequency, providing clear frequency selection conditions for subsequent fluorescence signal processing and avoiding the influence of non-periodic fluorescence changes when directly relying on fluorescence intensity changes for judgment. Response information in two phase directions is obtained through in-phase and orthogonal reference vectors, providing a data foundation for subsequent response parameter calculation and response pixel identification.

[0153] Example 5:

[0154] In one possible implementation, step S4 includes: performing mean-removal and normalization processing on the pixel-level fluorescence timing matrix row by row to obtain a standard timing matrix; performing an inner product operation on each row vector of the standard timing matrix with the in-phase reference vector to obtain an in-phase response vector; performing a matrix inner product operation on the standard timing matrix with the orthogonal reference vector to obtain an orthogonal response vector; combining the in-phase response vector and the orthogonal response vector into a pixel response parameter matrix; calculating the response amplitude and response phase based on the in-phase response value and orthogonal response value corresponding to each pixel, and arranging them in spatial coordinates to obtain a response amplitude matrix and a response phase matrix.

[0155] This embodiment is used to perform digital frequency domain demodulation processing on pixel-level fluorescence time series matrix. By removing the basic fluorescence intensity differences at different pixel locations, and using in-phase reference vectors and orthogonal reference vectors to extract the fluorescence change components corresponding to the target response frequency, the response amplitude information and response phase information corresponding to each spatial location are obtained.

[0156] In the specific implementation process, the pixel-level fluorescence time series matrix is ​​first processed row by row. Since each row of data corresponds to the fluorescence intensity change of a fixed spatial location at different acquisition times, the average fluorescence intensity value of that spatial location over all acquisition times is calculated for each row of data. Subsequently, the average fluorescence intensity value is subtracted from the fluorescence intensity value corresponding to each acquisition time in the row, converting the fluorescence time series of that spatial location into a data series expressed as a change, thereby removing the fixed background fluorescence component of that spatial location. After the mean removal process is completed, each row of data is normalized. Specifically, the squares of the mean-removed time series data are calculated, the squares of the data corresponding to each acquisition time are accumulated, and the square root of the accumulated result is taken to obtain the normalized reference value corresponding to that spatial location; then, each time point data after mean removal is divided by the normalized reference value, converting the fluorescence change data of different spatial locations to a uniform amplitude scale to obtain a standard time series matrix, and the standard time series matrix is ​​subjected to matrix inner product processing with the in-phase reference vector. Specifically, for each pixel location in the standard time series matrix, the time series data is multiplied by the corresponding data in the in-phase reference vector according to the acquisition time order. All product results are then summed to obtain the in-phase response value for that pixel location. The in-phase response values ​​for all pixel locations are arranged spatially to form an in-phase response vector. Simultaneously, the standard time series matrix and the orthogonal reference vector are subjected to a matrix inner product. Specifically, the standardized fluorescence time series data for each pixel location is multiplied by the corresponding data in the orthogonal reference vector, and the product results for each time location are summed to obtain the orthogonal response value for that pixel location. The orthogonal response values ​​for all pixel locations are arranged in pixel index order to form an orthogonal response vector.

[0157] The in-phase response vector and the quadrature response vector are combined according to their corresponding pixel positions, so that each pixel position has both in-phase and quadrature response values. Specifically, based on the pixel index relationship, the in-phase and quadrature response values ​​corresponding to the same spatial position are stored in the same data location to form a pixel response parameter matrix. The pixel response parameter matrix is ​​used to record the response data of the two phase directions corresponding to each spatial position at the target response frequency.

[0158] The response amplitude at each pixel location is calculated based on the in-phase and quadrature response values ​​in the pixel response parameter matrix. Specifically, the in-phase and quadrature response values ​​corresponding to the same pixel location are squared, the two squared results are added together, and the square root of the sum is taken to obtain the response amplitude corresponding to that pixel location. The response amplitude represents the response intensity in the fluorescence change signal at that pixel location corresponding to the target response frequency. Simultaneously, the response phase is determined based on the in-phase and quadrature response values ​​corresponding to the same pixel location. Specifically, based on the correspondence between the quadrature and in-phase response values, the phase shift of the fluorescence response at that pixel location relative to the periodic stimulus signal is calculated to obtain the response phase data. The response phase is used to represent the correspondence between the fluorescence response at different spatial locations and the occurrence time of the stimulus signal. Finally, based on the spatial coordinate index corresponding to each pixel location, the calculated response amplitude data is arranged according to the original image spatial location to form a response amplitude matrix; simultaneously, the response phase data is arranged according to the same spatial coordinate relationship to form a response phase matrix. Through the above processing, the temporal variation information in the fluorescence image sequence is converted into response parameter data containing spatial distribution characteristics, providing a data foundation for subsequent periodic response consistency analysis and response pixel determination.

[0159] Example 6:

[0160] In one possible implementation, S5 includes: dividing the pixel-level fluorescence time-series matrix into multiple periodic sub-matrices according to the periodic boundaries of each periodic stimulus; calculating the response amplitude and response phase of each pixel in each periodic sub-matrix to construct a periodic response feature matrix; calculating the periodic response consistency based on the amplitude dispersion and phase dispersion of each pixel in multiple periods; extracting adjacent frequency components on both sides of the target frequency and located outside the preset guard band to calculate the adjacent frequency noise features of each pixel; and performing a weighted calculation based on the ratio of the response amplitude to the adjacent frequency noise features and in conjunction with the periodic response consistency to obtain the response score of each pixel.

[0161] This implementation method segments pixel-level fluorescence time-series data according to periodic stimulation boundaries, analyzes the response amplitude and phase changes of pixel positions within each stimulation period, and evaluates the periodic response stability of pixel positions by combining the noise characteristics of non-target frequency regions near the target response frequency, thereby obtaining a response score for screening response pixels.

[0162] In the specific implementation process, the periodic division boundary is first determined based on the periodic envelope corresponding to the periodic stimulus signal. The periodic envelope is obtained based on the target response frequency determined above. Specifically, the duration of a stimulation cycle is determined based on the single complete change time of the periodic stimulus signal. Then, the corresponding time position is found based on the fluorescence image acquisition time series, and image frames that meet the time range of the same stimulation cycle are divided into the same periodic data segment. The pixel-level fluorescence time series matrix is ​​segmented according to each periodic boundary, so that the original time series matrix is ​​converted into multiple periodic sub-matrices. Each periodic sub-matrix corresponds to the fluorescence change data of each pixel position within a stimulation cycle. After the period division is completed, the response parameters are calculated for each periodic sub-matrix. Specifically, for each pixel position in each periodic sub-matrix, the fluorescence time series corresponding to the pixel position is calculated by performing the inner product with the in-phase reference vector and the orthogonal reference vector respectively to obtain the in-phase response value and the orthogonal response value corresponding to the pixel position within that cycle. The response amplitude is calculated based on the in-phase and quadrature response values ​​corresponding to the same pixel location. Each response value is squared, the squared results are summed, and the square root of the sum is taken to obtain the response amplitude of that pixel location within the current period. Simultaneously, the response phase within the current period is determined based on the in-phase and quadrature response values ​​corresponding to the same pixel location. Specifically, the phase shift of the fluorescence response signal relative to the periodic stimulus signal is determined based on the correspondence between the quadrature and in-phase response values, and the response phase of that pixel location within the current period is recorded. The calculated response amplitudes and phases for each period are arranged according to the pixel's spatial location to form a periodic response feature matrix, used to record the response changes of the same pixel location in different stimulus periods.

[0163] Furthermore, the periodic response consistency is calculated based on the response amplitude and response phase of each pixel location in different periods. Specifically, for any pixel location, the average response amplitude of that pixel is first calculated over all periods. Then, the difference between the response amplitude of each period and the average value is calculated. Each difference is squared and accumulated. The degree of dispersion of the response amplitude of that pixel location is determined based on the accumulated result. The calculation process for the response phase dispersion is as follows: the response phase of that pixel location in different periods is compared with the average phase value. The deviation between the response phase of each period and the average phase is calculated. The degree of phase change is determined based on the phase deviation of each period.

[0164] Periodic response consistency is determined based on amplitude and phase dispersion. Specifically, amplitude and phase dispersion are converted into consistency evaluation parameters. When the amplitude and phase changes of a pixel location within different stimulus periods meet set conditions, the consistency evaluation value for that pixel location is increased; when the amplitude or phase changes exceed the set conditions, the consistency evaluation value for that pixel location is decreased. Subsequently, noise features are extracted from adjacent frequency regions near the target response frequency. Specifically, the target response frequency is used as the center frequency, and guard bands are set on both sides of this frequency to avoid the effective response signal region corresponding to the target response frequency. The width of the guard band is determined based on the frequency stability of the stimulus signal and the image sampling time interval, ensuring that the effective response signal near the target response frequency does not participate in noise calculation.

[0165] Outside the guard band, fluorescence signal components in adjacent frequency regions on both sides of the target response frequency are extracted. Specifically, the fluorescence time series corresponding to the pixel location is frequency decomposed to obtain frequency components located in the region adjacent to the target response frequency but not within the guard band, and the signal intensity corresponding to these frequency components is calculated. This signal intensity is used as the adjacent frequency noise feature of the pixel location.

[0166] The response intensity evaluation parameters are calculated based on the response amplitude and adjacent-channel noise characteristics. Specifically, the response amplitude obtained at the target response frequency is taken as the effective response signal, and the adjacent-channel noise characteristics are taken as the non-target frequency interference signal. By calculating the ratio between the effective response signal and the adjacent-channel noise signal, the evaluation parameters reflecting the signal-to-noise relationship of the response at that pixel location are obtained.

[0167] Finally, the response intensity evaluation parameters and periodic response consistency are weighted and calculated to obtain the response score for each pixel location. Specifically, the influence ratios of the response intensity evaluation parameters and periodic response consistency are configured according to preset weights. The two evaluation parameters are multiplied by their respective weights and then summed to obtain the final response score. The weights are determined based on the background noise level and the stability of the periodic stimulus in the imaging experiment.

[0168] The response score threshold can be determined based on the response scores of pixels in the background region. Specifically, the response scores of each pixel location are acquired within a background region that does not contain vascular structures. The background score data is statistically analyzed to determine the score range corresponding to normal fluctuations in the background region. Score values ​​exceeding the upper limit of this range are set as the response score threshold. When the response score of the pixel to be detected reaches this response score threshold, the pixel is identified as a response pixel with effective periodic response characteristics.

[0169] Through the above processing, the periodic stability, target frequency response intensity, and adjacent channel noise interference factors are comprehensively evaluated, so that the subsequent response pixel selection process can be judged based on the time response characteristics of the pixel itself, providing a data basis for generating response vascular images.

[0170] Example 7:

[0171] In one possible implementation, step S6 includes: arranging the response scores of each pixel according to spatial coordinates to obtain a response score matrix; selecting background pixels from a preset background region or a low response score region to construct a background score sequence; calculating the median and dispersion of the background score sequence to obtain background noise statistical parameters; calculating a response determination threshold based on the background noise statistical parameters and a preset threshold coefficient; determining the pixels in the response score matrix that reach the response determination threshold as response pixels, and generating a response pixel mask.

[0172] This implementation spatially arranges the response scores of each pixel and establishes an adaptive judgment criterion using the response scores of the background region. It determines the response judgment threshold based on the background noise level, thereby extracting pixels that meet the periodic response characteristics from the entire imaging area and generating a response pixel mask for subsequent vascular structure fusion.

[0173] In the specific implementation process, the pixel response score data calculated above is first obtained. Based on the spatial coordinate information corresponding to each pixel, the response scores are reorganized according to the pixel arrangement in the original image. Specifically, based on the image row and column positions corresponding to each pixel index, the corresponding response scores are stored in the same spatial coordinate positions, forming a two-dimensional response score matrix with spatial positional relationships. Each data point in the response score matrix corresponds to a pixel position in the live target region, representing the periodic stimulus response characteristics at that position.

[0174] Subsequently, a background pixel region for background noise analysis is determined from the imaging area. This background region can be set before image acquisition based on locations without vascular structures, or it can be automatically selected based on a response scoring matrix, identifying regions with low response scores. Specifically, the response scores of all pixels are sorted, and corresponding pixel positions are extracted from data regions with low response scores according to a preset selection ratio as background pixels, or background pixels are extracted based on the coordinates of marked non-vascular regions. The response scores corresponding to the selected background pixels are arranged in the extraction order to form a background score sequence.

[0175] After obtaining the background rating sequence, statistical calculations are performed on it. First, all rating values ​​in the background rating sequence are sorted according to their numerical value. When the number of data points in the sequence is odd, the middle rating value after sorting is selected as the median; when the number of data points is even, the average of the two middle rating values ​​is selected as the median. The median is used to represent the central level of the background region's response rating.

[0176] Further, the dispersion of the background rating sequence is calculated. Specifically, the difference between each rating value in the background rating sequence and the median is calculated. The absolute values ​​of each difference are then summed, and the average is calculated based on the number of data points in the background rating sequence to obtain a dispersion parameter representing the degree of fluctuation in the background rating. This dispersion parameter is used to reflect the range of change in the response rating in the non-response region.

[0177] The background noise statistical parameters are determined based on the calculated median and dispersion parameters of the background score. Specifically, the median background score is used as the basic background response level, and the dispersion parameter is used as the change in background score. The background noise statistical parameters used for calculating the response determination threshold are obtained by combining the two.

[0178] Subsequently, the response judgment threshold is calculated based on background noise statistical parameters and a preset threshold coefficient. The preset threshold coefficient is set according to the actual imaging conditions, and its determination process is as follows: First, background score data of the target region for aperiodic stimulus response is acquired; under these conditions, the number of falsely detected pixels corresponding to different threshold coefficients is counted; the threshold coefficient is determined based on the allowed number of background false responses, ensuring that response scores in the background region are not judged as valid response pixels. After determining the threshold coefficient, the median background score and dispersion parameter are combined according to a preset calculation rule and multiplied by the threshold coefficient to obtain the final response judgment threshold.

[0179] After obtaining the response determination threshold, each pixel in the response score matrix is ​​compared point by point. Specifically, the response score corresponding to each pixel position is compared with the response determination threshold. When the response score of the pixel position reaches or exceeds the response determination threshold, the pixel is determined as a response pixel; when the response score is lower than the response determination threshold, the pixel is determined as a non-response pixel.

[0180] Finally, a response pixel mask is generated based on the positional relationship of the response pixels. Specifically, positions identified as response pixels are assigned valid identifiers, while positions of pixels that do not meet the response determination threshold are assigned invalid identifiers. These are then arranged according to the spatial coordinates of the original image to form a response pixel mask corresponding to the size of the fluorescence image. This response pixel mask represents the spatial positions within the target region that satisfy the periodic response condition, providing a data basis for subsequent fusion processing with the vascular structure mask.

[0181] Through the above processing, the response judgment conditions are determined according to the actual background noise level, so that the response pixel extraction process under different imaging conditions can be adjusted according to the changes in background signal, while retaining the periodic response features contained in the response score, thereby improving the accuracy of response blood vessel region identification.

[0182] Example 8:

[0183] In one possible implementation, S7 includes: performing time averaging or maximum intensity projection on the visible light or near-infrared fluorescence image sequence to obtain a vascular structure reference image; performing multi-scale vascular enhancement on the vascular structure reference image to obtain a vascular structure feature matrix; segmenting the vascular structure feature matrix according to a structure segmentation threshold and obtaining a vascular structure mask through morphological processing; performing intersection operation or weighted fusion on the response pixel mask corresponding to the response pixel and the vascular structure mask to obtain a response vascular feature matrix; and recovering the spatial position and response intensity according to the response vascular feature matrix to generate a response vascular image and / or a response vascular mask.

[0184] This embodiment extracts the spatial structure of fluorescence image sequences in the visible or near-infrared regions, and fuses the vascular morphology information with the periodic response pixel information, so that the response pixels can correspond to the actual vascular structure location, forming a responsive vascular image that simultaneously contains vascular spatial distribution and stimulus response information.

[0185] In the specific implementation process, structural information is first extracted from the visible light or near-infrared fluorescence image sequence after time synchronization processing. Since the vascular signal in a single fluorescence image may be affected by instantaneous fluorescence changes and random noise, intensity fusion processing is performed on consecutive image frames in the temporal direction. When generating a vascular structure reference image using time averaging, the fluorescence intensity in all image frames is accumulated at the same spatial location, and the accumulated result is divided by the number of image frames involved in the calculation to obtain the average fluorescence intensity corresponding to that spatial location. The calculation results for all spatial locations are rearranged to form a time-averaged vascular structure reference image. When generating a vascular structure reference image using maximum intensity projection, for each spatial location, the fluorescence intensity values ​​at that location are compared across all image frames, and the highest fluorescence intensity value is taken as the projection intensity for that spatial location. These values ​​are then arranged according to the original image coordinates to form a maximum intensity projection image. Through this method, the fluorescence structural information corresponding to the vascular region is displayed centrally. Subsequently, multi-scale vascular enhancement processing is performed on the vascular structure reference image. Specifically, based on the directional and scale characteristics of the vascular structure in the image, different scale filtering parameters are used to process the image. At each scale, the grayscale changes of local areas of the image are analyzed, and linear structural response values ​​with continuous directionality are extracted. The structural response values ​​obtained at different scales are then combined according to their spatial location to form a vascular structure feature matrix.

[0186] Each data point in the vascular structure feature matrix corresponds to the vascular structure response intensity at a spatial location. For pixels located within a vascular region, since their grayscale change direction corresponds to the vascular extension direction, the corresponding structural response value is obtained after multi-scale enhancement processing. For non-vascular regions, their structural response values ​​are generated based on local image changes. After obtaining the vascular structure feature matrix, vascular structure regions are identified according to a structural segmentation threshold. This structural segmentation threshold is used to distinguish between vascular structure response regions and non-vascular structure regions. Specifically, firstly, the structural feature matrix data corresponding to the background region that does not contain vascular structures is obtained, the average level and fluctuation range of the background region's structural response value are calculated, and the initial segmentation threshold is determined based on the normal structural response range of the background region.

[0187] Furthermore, the initial segmentation threshold is adjusted based on the actual vascular imaging results. The specific adjustment process involves comparing the segmentation results obtained under different threshold conditions with the labeled vascular regions, counting the number of missed vascular regions and the number of misidentified non-vascular regions, and determining the final structure segmentation threshold based on a preset allowable error range. When the pixel value in the vascular structure feature matrix reaches the structure segmentation threshold, the corresponding pixel is identified as a vascular structure pixel.

[0188] After thresholding, the binarized segmentation results are subjected to morphological processing. Specifically, based on the requirement of vascular structure continuity, hole-filling processing is performed on the segmented image to fill in the broken areas caused by grayscale changes within the vascular region; simultaneously, isolated small regions are removed to reduce interference from non-vascular structures. After morphological processing, a vascular structure mask that continuously represents the spatial location of blood vessels is obtained.

[0189] Subsequently, the generated response pixel mask and vascular structure mask are spatially fused. When using intersection operations, the pixel states at the same spatial location in both masks are judged, and only locations that simultaneously satisfy both the response pixel condition and the vascular structure condition are retained; these locations are designated as the response vascular region. When using a weighted fusion method, the data at the same spatial location are combined and calculated based on the response score corresponding to the response pixel mask and the structural response value corresponding to the vascular structure mask. Specifically, the response score data and the vascular structure response data are multiplied according to preset weights, and the results are summed to obtain the response vascular feature value corresponding to that spatial location, thus forming a response vascular feature matrix. Finally, based on the spatial index information in the response vascular feature matrix, the response vascular feature values ​​corresponding to each spatial location are restored to their original image coordinates. For locations where the response vascular feature value meets the set display conditions, the image grayscale value is configured according to its corresponding response intensity to generate a response vascular image; or, binarization processing is performed based on the existence of valid response vascular feature values ​​to generate a response vascular mask.

[0190] Through the above processing, the fluorescence response information generated by periodic stimulation is correlated with the spatial structure information of blood vessels, so that the final response vascular image can simultaneously reflect the distribution location of blood vessels and the periodic response characteristics corresponding to the vascular regions. This processing avoids the problem of misidentification of non-vascular regions caused by spatial judgment based solely on fluorescence response scores, thus improving the accuracy of vascular region localization.

[0191] Example 9:

[0192] In one possible implementation, a fluorescence-based response vascular imaging system, applicable to a fluorescence-based response vascular imaging method, includes: a periodic stimulation and image acquisition module, used to apply periodic stimulation to a live target region and simultaneously acquire a visible or near-infrared fluorescence image sequence of the live target region, while recording stimulation timing information corresponding to the periodic stimulation; a time synchronization module, connected to the periodic stimulation and image acquisition module, used to perform time synchronization based on the acquisition time points corresponding to the visible or near-infrared fluorescence image sequence and the stimulation timing information to obtain a synchronized fluorescence sequence; a pixel-level time-series matrix construction module, connected to the time synchronization module, used to extract the fluorescence intensity corresponding to each image frame in the synchronized fluorescence sequence according to the same spatial position, and arrange them according to the acquisition time order to construct a pixel-level fluorescence time-series matrix; a digital frequency domain demodulation module, connected to the pixel-level time-series matrix construction module, used to determine the target response frequency based on the periodic stimulation signal applied to the live target region, and generate an in-phase reference vector and an orthogonal reference vector based on the target response frequency and the image acquisition time sequence; and to convert the pixel-level fluorescence time sequence... The matrix is ​​subjected to inner product operation with the in-phase reference vector and the orthogonal reference vector respectively to obtain the response parameters corresponding to the target response frequency, and the response amplitude and response phase of each pixel are determined according to the response parameters; the periodic response analysis module, connected to the digital frequency domain demodulation module, divides the pixel-level fluorescence time series matrix into multiple periodic data segments according to the periodic boundary corresponding to the periodic stimulus, determines the periodic response consistency according to the response amplitude and response phase corresponding to each periodic data segment, and calculates the response score of each pixel in combination with the noise characteristics of the frequency band adjacent to the target response frequency; the response pixel determination module, connected to the periodic response analysis module, is used to determine the response determination threshold according to the response score corresponding to the background pixel, and determines the pixel that reaches the response determination threshold as the response pixel, generating a response pixel mask; the vascular structure fusion module, connected to the periodic stimulus and image acquisition module and the response pixel determination module, is used to extract vascular structure features from the visible light or near-infrared fluorescence image sequence, generate a vascular structure mask, and spatially fuse the response pixel mask with the vascular structure mask to generate a response vascular image and / or a response vascular mask. This embodiment provides a system for vascular imaging based on fluorescence response characteristics. It acquires time-series fluorescence data of a live target region through a periodic stimulation and image acquisition module, eliminates the time deviation between the stimulation signal and image acquisition through a time synchronization module, establishes a data structure corresponding to spatial location and time changes through a pixel-level temporal matrix construction module, extracts fluorescence change information corresponding to the target response frequency using a digital frequency domain demodulation module, determines the effective response region by combining a periodic response analysis module and a response pixel determination module, and finally matches the response information with the vascular spatial structure through a vascular structure fusion module to form a response vascular image.

[0193] In practical implementation, the periodic stimulation and image acquisition module includes a periodic stimulation unit and a visible or near-infrared fluorescence image acquisition unit. The periodic stimulation unit applies stimulation signals to the live target area according to a preset period; these stimulation signals can be generated by a mechanical modulation device, an optical modulation device, an electrical stimulation device, or other periodic control equipment. The visible or near-infrared fluorescence image acquisition unit continuously acquires fluorescence images generated in the live target area during stimulation and records the corresponding acquisition time point for each image acquisition.

[0194] When the periodic stimulation and image acquisition module is working, the stimulation control time and image acquisition time are recorded separately. Stimulation timing information is used to represent the state changes of the periodic stimulation signal at different time positions, and the fluorescence image sequence is used to represent the fluorescence distribution of the live target region at different acquisition times. By synchronously acquiring these two types of data, a data foundation is provided for subsequently establishing the time correspondence between stimulation state and fluorescence response. The time synchronization module is connected to the periodic stimulation and image acquisition module and is used to handle the time offset between stimulation time and image acquisition time. Specifically, the time synchronization module acquires the acquisition time point of each image frame in the fluorescence image sequence and obtains the reference time point corresponding to the periodic stimulation, calculating the time difference between corresponding events in the two time series.

[0195] When establishing the temporal correspondence, the time synchronization module first selects multiple stimulus change nodes and matches them with image acquisition nodes, calculates the time difference between each matched node, and determines the offset of the image acquisition time relative to the stimulus time based on all time differences. Then, it adjusts the stimulus reference sequence according to this offset, ensuring the adjusted stimulus reference data corresponds to the acquisition time of each image frame, thus obtaining a synchronized fluorescence sequence. The pixel-level temporal matrix construction module, connected to the time synchronization module, converts the synchronized fluorescence sequence into pixel-level temporal change data. Specifically, it establishes spatial coordinate indices based on the pixel size of the image frames in the synchronized fluorescence sequence, with each pixel location corresponding to a unique spatial index. For each image frame, it reads the fluorescence intensity value at the corresponding location based on the spatial index and arranges them according to the pixel index order, forming the intensity data sequence corresponding to that image frame. Subsequently, the pixel-level temporal matrix construction module arranges all intensity data sequences according to the image acquisition time order, ensuring that the fluorescence intensity data corresponding to the same spatial location is stored continuously in temporal order. After matrix recombination, each row in the matrix corresponds to a spatial location, and each column corresponds to an acquisition time point, forming a pixel-level fluorescence temporal matrix.

[0196] The digital frequency domain demodulation module is connected to the pixel-level time-series matrix construction module and is used to extract the response signal corresponding to the periodic stimulus frequency from the fluorescence time series. Specifically, the digital frequency domain demodulation module first obtains the variation period of the periodic stimulus signal, and determines the target response frequency based on the time interval between two adjacent complete periodic variations.

[0197] After determining the target response frequency, the digital frequency domain demodulation module generates a reference vector based on the image acquisition time series. Specifically, for each image acquisition time point, the corresponding reference signal state is determined according to the position of that time point in the stimulation cycle, generating both an in-phase reference vector and a quadrature reference vector. The in-phase reference vector corresponds to the direction with the same phase as the stimulation signal, while the quadrature reference vector corresponds to the direction with a 90° phase difference from the in-phase reference vector.

[0198] The digital frequency domain demodulation module further calculates the response of the pixel-level fluorescence time series matrix. Specifically, for each pixel location, the fluorescence time series corresponding to that pixel location is multiplied item by item with the data at the corresponding time position in the in-phase reference vector, and all the product results are accumulated to obtain the in-phase response value corresponding to that pixel location; then, the same fluorescence time series is multiplied item by item with the data at the corresponding time position in the orthogonal reference vector, and all the product results are accumulated to obtain the orthogonal response value corresponding to that pixel location.

[0199] Based on the in-phase and quadrature response values, the digital frequency domain demodulation module calculates the response amplitude and phase corresponding to each pixel position. The response amplitude is obtained by summing the squares of the corresponding in-phase and quadrature response values ​​and then taking the square root of the sum. The response phase is determined based on the correspondence between the in-phase and quadrature response values ​​and is used to represent the time offset of the fluorescence response relative to the periodic stimulus signal.

[0200] The periodic response analysis module is connected to the digital frequency domain demodulation module to analyze the response stability of each pixel location within different stimulus periods. Specifically, the periodic response analysis module divides the pixel-level fluorescence time series matrix into multiple periodic data segments based on the periodic boundaries corresponding to the periodic stimulus signals.

[0201] For each periodic data segment, the periodic response analysis module calculates the response amplitude and response phase corresponding to each pixel position, and arranges them according to spatial location to form periodic response feature data. Subsequently, the changes in response amplitude at the same pixel position in different periods are statistically analyzed, and the response amplitude of each period is compared with the average response amplitude of all periods at that pixel position to calculate the degree of response amplitude change; at the same time, the deviation between the response phase of each period and the average response phase is calculated to obtain the degree of phase change.

[0202] Based on the degree of change in response amplitude and phase, the periodic response analysis module determines the consistency of the periodic response. When the change in response amplitude and phase at the same pixel location in different stimulus cycles meets the preset stability condition, the consistency evaluation value corresponding to that pixel location is increased; when the change exceeds the preset stability condition, the corresponding evaluation value is decreased. Simultaneously, the periodic response analysis module performs noise analysis on the frequency region near the target response frequency. Specifically, guard bands are set on both sides of the target response frequency to exclude valid response signals corresponding to the target response frequency. Then, neighboring frequency components outside the guard bands are extracted, and the intensity of the neighboring frequency components corresponding to each pixel location is calculated to obtain the adjacent frequency noise characteristics.

[0203] The response intensity evaluation parameter is determined based on the proportional relationship between the response amplitude and the adjacent-channel noise characteristics, and this parameter is then weighted and calculated in conjunction with the periodic response consistency. Specifically, the weighted results for the response intensity evaluation parameter and the periodic response consistency are calculated separately according to preset weights, and then the two weighted results are added together to obtain the response score for each pixel position. The response pixel determination module is connected to the periodic response analysis module and is used to determine the effective response region based on the response score. Specifically, the response pixel determination module first obtains the response scores corresponding to background pixels from a preset background region or a low response score region, and arranges them in numerical order to form a background score sequence.

[0204] Statistical calculations were performed on the background rating sequence to obtain the median and dispersion of the background ratings. The median was obtained by selecting the middle data point after sorting the background rating sequence; the dispersion was obtained by calculating the difference between each background rating and the median, and then averaging the absolute values ​​of the differences.

[0205] The response judgment threshold is calculated based on the median and dispersion of the background score and a preset threshold coefficient. The preset threshold coefficient is determined according to the actual imaging conditions. Specifically, background score data is acquired under no-target response conditions, the number of false response pixels corresponding to different threshold conditions is counted, and the threshold coefficient is determined based on the allowable number of false responses. The response pixel judgment module compares the score of each pixel in the response score matrix with the response judgment threshold, identifies pixels that reach the response judgment threshold as response pixels, and generates a response pixel mask. The vascular structure fusion module is connected to the periodic stimulation and image acquisition module and the response pixel judgment module, respectively, to fuse response information with vascular structure information. Specifically, the vascular structure fusion module acquires a sequence of fluorescence images in the visible or near-infrared region, performs time averaging or maximum intensity projection processing on consecutive image frames to obtain a vascular structure reference image. Subsequently, multi-scale vascular enhancement processing is performed on the vascular structure reference image, and structural response values ​​are calculated based on the vascular edge and continuity features at different scales to form a vascular structure feature matrix. A structural segmentation threshold is determined based on the structural response values ​​of the background region, and the positions that reach the structural segmentation threshold are identified as vascular structure regions. Morphological processing such as hole filling and isolated region removal is then performed to generate a vascular structure mask.

[0206] Finally, the vascular structure fusion module fuses the response pixel mask and the vascular structure mask. When using the intersection operation method, only the pixel positions in both masks that simultaneously satisfy the response condition and the vascular structure condition are retained to form a response vascular feature matrix; when using the weighted fusion method, the response score and the vascular structure response value are calculated according to preset weights to obtain the fused response vascular feature value.

[0207] Based on the spatial coordinate information in the response vessel feature matrix, the fused data is restored to the original image spatial position, and a response vessel image is generated according to the corresponding response intensity; or a response vessel mask is generated according to the effective response position in the response vessel feature matrix, so as to realize the imaging of the vessel region with periodic stimulus response characteristics in the live target region.

[0208] Example 10:

[0209] In one possible implementation, the time synchronization module includes: a timestamp acquisition unit, used to acquire image timestamp sequences corresponding to visible light or near-infrared fluorescence image sequences, and stimulus reference sequences and reference timestamp sequences corresponding to periodic stimuli; a clock mapping unit, used to establish a clock mapping model based on the image timestamp sequences and the reference timestamp sequences to obtain time offset parameters; a reference resampling unit, used to interpolate and resample the stimulus reference sequences based on the time offset parameters to generate reference vectors corresponding to each image frame; and an error filtering unit, used to calculate the time alignment error between the reference vectors and each image frame, and filter to obtain synchronized fluorescence sequences according to a preset error threshold.

[0210] In this embodiment, the time synchronization module is used to establish the correspondence between the acquisition time of fluorescence images in the visible or near-infrared region and the periodic stimulation time. Through timestamp acquisition, clock mapping, reference sequence resampling, and error screening, each fluorescence image can correspond to a specific time state in the stimulation process, thereby ensuring that the subsequent extraction of response signals based on stimulation frequency has an accurate time reference.

[0211] In the specific implementation process, the timestamp acquisition unit is connected to the periodic stimulus and image acquisition modules respectively, and is used to acquire the time information corresponding to the fluorescence image sequence and the reference time information corresponding to the periodic stimulus.

[0212] Specifically, during the acquisition of fluorescence images in the visible or near-infrared region, the image acquisition device records the current acquisition time at each exposure trigger and stores the time data corresponding to each image frame according to the acquisition order, forming an image timestamp sequence. Each time data in the image timestamp sequence corresponds to one visible or near-infrared fluorescence image. Simultaneously, the periodic stimulation control unit records the time position of the stimulation signal change and the corresponding stimulation state when outputting periodic stimulation signals. The stimulation states corresponding to each time position are arranged in chronological order to form a stimulation reference sequence, and the occurrence time of the stimulation state is recorded to form a reference timestamp sequence.

[0213] The clock mapping unit is connected to the timestamp acquisition unit and is used to establish a time correspondence between the image timestamp sequence and the reference timestamp sequence. Since there may be a start time difference between the image acquisition clock and the stimulus control clock, it is necessary to perform matching calculations on the two time series.

[0214] Specifically, the clock mapping unit selects multiple image acquisition time points from the image timestamp sequence and finds their corresponding stimulus time points in the reference timestamp sequence. For each set of corresponding time points, the time difference between the image acquisition time and the stimulus reference time is calculated. The time differences are arranged in chronological order, and the time offset parameter is determined based on the relationship between the changes in the time differences.

[0215] When there is only a fixed time offset between the image acquisition clock and the stimulus control clock, the average value of each group of time differences is used as the time offset parameter; when there is a time variation error between the two clocks, a time mapping relationship is established based on the correspondence between multiple matching time points, and the time offset parameter used for time correction is obtained by calculating the offset change corresponding to different time positions.

[0216] The reference resampling unit is connected to the clock mapping unit and is used to adjust the stimulus reference sequence according to the time offset parameter so that the stimulus reference data corresponds to the image acquisition time.

[0217] Specifically, the reference resampling unit first corrects the reference timestamp sequence according to the time offset parameter, moving the time position in the stimulus reference sequence to the position corresponding to the image acquisition time. Then, based on the acquisition time of each image frame, the corresponding reference data is determined in the corrected stimulus reference sequence.

[0218] When the image acquisition time perfectly corresponds to the stimulus reference time, the stimulus state at that time position is directly read as the reference data. When the image acquisition time lies between two adjacent stimulus reference times, the stimulus state at the current acquisition time position is interpolated based on the stimulus state change relationship between the two adjacent reference times to obtain the corresponding reference data. Through the above interpolation and resampling process, each image frame has a stimulus reference value corresponding to its time position, and reference vectors corresponding to each image frame are formed according to the image frame arrangement order. The error filtering unit is connected to the reference resampling unit and is used to determine the degree of time matching between the image acquisition time and the reference vector.

[0219] Specifically, the error filtering unit compares the acquisition time corresponding to each image frame with the time position of the corresponding reference data in the reference vector, calculates the time difference between the two, and statistically analyzes the time differences corresponding to all image frames to obtain the time alignment error corresponding to each image frame.

[0220] The preset error threshold is determined based on the sampling frequency of the imaging equipment, the periodic stimulation frequency, and the time synchronization accuracy requirements of subsequent digital frequency domain demodulation. Specifically, during the determination process, multiple sets of image trigger time and stimulus trigger time data are first acquired under stable equipment operation, and the actual deviation range between the image trigger time and the stimulus trigger time is calculated. Then, the allowable error range is determined based on the maximum acceptable time offset for digital frequency domain demodulation, and the maximum time deviation value corresponding to this allowable error range is set as the preset error threshold. The error filtering unit compares the calculated time alignment error with the preset error threshold. When the time alignment error corresponding to a certain image frame is less than the preset error threshold, the image frame and its corresponding reference data are retained; when the time alignment error reaches or exceeds the preset error threshold, the data set is excluded. Finally, the image frames that meet the time error requirements are rearranged according to the acquisition time order and combined with the corresponding reference vector to form a synchronized fluorescence sequence. Each fluorescence image in this synchronized fluorescence sequence has a time marker corresponding to the periodic stimulation state, providing a synchronized data foundation for subsequent pixel-level fluorescence timing matrix construction, digital frequency domain demodulation, and responsive vessel region identification.

[0221] Example 11:

[0222] In one possible implementation, the pixel-level time-series matrix construction module includes: a spatial index establishment unit, used to establish a spatial coordinate index matrix based on the image size of the synchronized fluorescence sequence; a fluorescence intensity extraction unit, used to extract the fluorescence intensity of each image frame at the corresponding spatial location based on the spatial coordinate index matrix; an effective pixel filtering unit, used to filter effective pixels based on a preset saturation threshold and a missing rate threshold; and a time-series arrangement unit, used to arrange the fluorescence intensity data corresponding to each image frame in the order of acquisition time to form a pixel-level fluorescence time-series matrix.

[0223] The pixel-level time-series matrix construction module in this embodiment is used to convert two-dimensional image data in the synchronous fluorescence sequence into pixel-level time-series data arranged according to spatial location and acquisition time. Through spatial indexing, fluorescence intensity extraction, invalid pixel screening, and time arrangement processing, each valid pixel position corresponds to complete fluorescence intensity change information, providing a data foundation for subsequent digital frequency domain demodulation.

[0224] In the specific implementation process, the spatial indexing unit is connected to the time synchronization module and is used to establish a spatial coordinate index matrix based on the image size in the synchronized fluorescence sequence.

[0225] Specifically, the spatial indexing unit obtains the number of pixel rows and columns of any image frame in the synchronous fluorescence sequence, and establishes a two-dimensional spatial coordinate relationship based on the actual arrangement of the image. For each pixel position in the image, a unique index identifier is assigned according to its row and column positions, so that the same index corresponds to the same spatial position in image frames obtained at different acquisition times.

[0226] For example, when all image frames in a synchronized fluorescence sequence have the same size, the spatial indexing unit establishes a corresponding index for each pixel location according to the arrangement order from the first row to the last row and from the first column to the last column, and generates a spatial coordinate index matrix. This spatial coordinate index matrix is ​​used to record the correspondence between pixel indices and actual image spatial locations.

[0227] The fluorescence intensity extraction unit is connected to the spatial indexing unit and is used to extract the fluorescence intensity of each image frame at the corresponding spatial location based on the spatial coordinate index matrix.

[0228] Specifically, the fluorescence intensity extraction unit sequentially reads each image frame in the synchronous fluorescence sequence, determines the storage location of each pixel in the current image frame according to the spatial coordinate index matrix, and extracts the fluorescence intensity data at the corresponding location.

[0229] For a single image frame, the fluorescence intensity extraction unit reads the fluorescence intensity values ​​corresponding to all pixel positions in pixel index order and converts the two-dimensional fluorescence intensity data into a one-dimensional intensity data sequence arranged by pixel index. Each data point in this intensity data sequence corresponds to a fluorescence intensity at a specific spatial location. An effective pixel filtering unit is connected to the fluorescence intensity extraction unit and is used to filter pixel positions based on preset saturation thresholds and missing rate thresholds.

[0230] Specifically, the saturation threshold is first determined based on the dynamic response range of the fluorescence image acquisition device. In this process, the maximum fluorescence intensity value that the image sensor can accurately respond to under normal exposure conditions is obtained, and the fluorescence intensity that reaches or exceeds this maximum effective response value is set as the saturation criterion. When the fluorescence intensity at a certain pixel location reaches the saturation threshold during image acquisition, the data corresponding to that pixel location is marked as a saturated pixel.

[0231] To determine the missing data rate threshold, the number of valid fluorescence intensities at each pixel location across all synchronized fluorescence image frames is first calculated. Specifically, time points that cannot be read, have abnormal data, or were not fully acquired are marked as missing data, and the ratio between the number of missing data at that pixel location and the total number of image frames is calculated.

[0232] The missing rate threshold is determined based on the number of effective time samples required for subsequent response analysis. Specifically, the number of periods corresponding to the periodic stimulus and the time series length required for digital frequency domain demodulation are considered to determine the minimum number of effective samples that must be retained at each pixel location. The missing rate threshold is then set based on the maximum allowable missing proportion corresponding to this number. When the data missing proportion at a pixel location reaches or exceeds the missing rate threshold, that pixel location is determined to be an invalid pixel. Based on the saturation threshold and missing rate threshold selection results, the effective pixel selection unit retains pixel locations that have not experienced saturation and meet the data integrity requirements, generating an effective pixel index set. This effective pixel index set is used to determine the spatial location involved in the subsequent temporal matrix construction process. The temporal arrangement unit is connected to the effective pixel selection unit and is used to arrange the fluorescence intensity data corresponding to each image frame according to the acquisition time sequence.

[0233] Specifically, the temporal arrangement unit reads the effective pixel fluorescence intensity data of each frame according to the acquisition time sequence of the image frames in the synchronous fluorescence sequence. For the same effective pixel index, the fluorescence intensity values ​​corresponding to the pixel position at different acquisition times are arranged in chronological order to form the fluorescence time series corresponding to that spatial position.

[0234] Subsequently, all valid pixel time series are matrixed according to spatial location index and acquisition time order. Specifically, each valid pixel location is treated as a row in the matrix, each image acquisition time point is treated as a column in the matrix, and the fluorescence intensity value of the corresponding time location is stored at the intersection of the row and column, forming a pixel-level fluorescence time series matrix.

[0235] In the pixel-level fluorescence timing matrix, each row of data represents the fluorescence response process at the same spatial location as the acquisition time changes, and each column of data represents the fluorescence distribution at different spatial locations at the same acquisition time. This matrix structure enables the subsequent digital frequency domain demodulation module to independently extract the fluorescence change signal corresponding to the target response frequency for each pixel location.

[0236] Example 12:

[0237] In one possible implementation, the digital frequency domain demodulation module includes: a target frequency determination unit, configured to determine a target response frequency based on the stimulation period of a periodic stimulus signal; a reference vector generation unit, configured to generate an in-phase reference sequence and an orthogonal reference sequence based on the target response frequency and the acquisition time points corresponding to each image frame, and to normalize the in-phase reference sequence and the orthogonal reference sequence to obtain an in-phase reference vector and an orthogonal reference vector; a response component calculation unit, configured to perform inner product operations between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain an in-phase response component and an orthogonal response component; and a response parameter calculation unit, configured to calculate the response amplitude and response phase based on the in-phase response component and the orthogonal response component.

[0238] In this embodiment, the digital frequency domain demodulation module is used to extract the fluorescence response components corresponding to the periodic stimulation frequency from the pixel-level fluorescence time series matrix. By determining the target response frequency, generating the phase reference vector, calculating the response components, and obtaining the response parameters, the fluorescence time change at each pixel position is converted into the response amplitude and response phase information at the corresponding frequency.

[0239] In the specific implementation process, the target frequency determination unit is connected to the periodic stimulus and image acquisition module to determine the target response frequency based on the periodic stimulus signal.

[0240] Specifically, the target frequency determination unit acquires the stimulus time series corresponding to the periodic stimulus signal and determines the stimulus period based on the time interval between two consecutive completions of the same periodic state change in the stimulus signal. In the specific calculation process, the difference between the starting time points of adjacent periods is calculated to obtain the time length corresponding to each periodic change, and the periodic time parameter is determined based on the corresponding time lengths of consecutive periods. When the periodic time length remains consistent, the frequency corresponding to that period is used as the target response frequency; when the periodic time length varies, statistical processing is performed on all periodic time lengths to obtain the target periodic time for response analysis, and the target response frequency is determined based on this target periodic time.

[0241] The reference vector generation unit is connected to the target frequency determination unit and is used to generate in-phase reference sequences and orthogonal reference sequences based on the target response frequency and the image frame acquisition time.

[0242] Specifically, the reference vector generation unit acquires the image acquisition time points corresponding to the pixel-level fluorescence timing matrix. For each image acquisition time point, the stimulation phase state corresponding to that time point is determined based on its positional relationship during the periodic stimulation process.

[0243] Based on the periodic variation pattern corresponding to the target response frequency, corresponding reference data is configured for each image acquisition time point to form an in-phase reference sequence. Each data point in the in-phase reference sequence corresponds to an image acquisition time point and is used to represent the reference state in the in-phase direction with the stimulus signal at that time point.

[0244] While generating the in-phase reference sequence, the reference vector generation unit generates an orthogonal reference sequence based on the same target response frequency. Specifically, the phase state corresponding to the in-phase reference sequence is shifted by 90° along the periodic change direction to obtain a reference data sequence with a fixed phase difference from the in-phase reference sequence. The orthogonal reference sequence is used to represent the response reference state in the direction with a 90° phase difference from the stimulus signal.

[0245] Subsequently, the reference vector generation unit normalizes the in-phase reference sequence and the orthogonal reference sequence. Specifically, it calculates the sum of squares of all reference data in the in-phase reference sequence and takes the square root of the sum of squares to obtain the amplitude reference value corresponding to the in-phase reference sequence; then, it divides each reference data in the in-phase reference sequence by the amplitude reference value to convert the in-phase reference sequence into an in-phase reference vector with uniform amplitude.

[0246] The orthogonal reference sequence is processed in the same way. Specifically, the sum of squares of all reference data in the orthogonal reference sequence is calculated, the result is square rooted to obtain the amplitude reference value corresponding to the orthogonal reference sequence, and then each reference data in the orthogonal reference sequence is divided by the amplitude reference value to obtain the normalized orthogonal reference vector.

[0247] After the above processing, the in-phase reference vector and the orthogonal reference vector have the same data length and maintain a 90° phase difference relationship, so that the subsequent calculation process can extract the response signals of different phase directions corresponding to the target response frequency.

[0248] The response component calculation unit is connected to the pixel-level temporal matrix construction module and is used to calculate the response components corresponding to each pixel position.

[0249] Specifically, the response component calculation unit reads the fluorescence time series corresponding to each pixel position in the pixel-level fluorescence time series matrix. For any pixel position, the fluorescence intensity data corresponding to that pixel position at different acquisition times are multiplied with the corresponding data in the in-phase reference vector according to the acquisition time order.

[0250] After completing the time-by-time multiplication calculation, the product results corresponding to all time points of the pixel location are accumulated to obtain the in-phase response component corresponding to that pixel location. This in-phase response component represents the signal component in the fluorescence change at that pixel location that is in phase with the target response frequency.

[0251] Subsequently, the response component calculation unit calculates the orthogonal response components using the same processing method. Specifically, the fluorescence time series corresponding to the same pixel location is multiplied with the data at the corresponding time position in the orthogonal reference vector, and all product results are summed to obtain the orthogonal response component corresponding to that pixel location.

[0252] Through the above calculation process, the response component calculation unit can obtain the corresponding in-phase response component and quadrature response component for each spatial location in the pixel-level fluorescence time series matrix.

[0253] The response parameter calculation unit is connected to the response component calculation unit and is used to calculate the response amplitude and response phase based on the in-phase response component and the quadrature response component.

[0254] Specifically, the response parameter calculation unit obtains the corresponding in-phase response component and quadrature response component for each pixel location. First, the in-phase response component is squared, and the quadrature response component is squared. Then, the two squared results are summed, and the sum is squared to obtain the response amplitude corresponding to that pixel location.

[0255] During response phase calculation, the phase shift of the fluorescence response relative to the periodic stimulus signal is determined based on the proportional relationship between the in-phase and quadrature response components. Specifically, by comparing the directional relationships between the in-phase and quadrature response components, the phase position of the response signal at that pixel location relative to the stimulus reference signal is determined, thus obtaining the response phase.

[0256] Finally, the response parameter calculation unit outputs the response amplitude data and response phase data corresponding to each pixel position, and arranges them according to the pixel spatial coordinates to form a response amplitude matrix and a response phase matrix. The response amplitude matrix is ​​used to represent the fluorescence response intensity corresponding to the target response frequency at different spatial locations, and the response phase matrix is ​​used to represent the time correspondence between the fluorescence response and the periodic stimulus at different spatial locations, providing a parameter basis for subsequent periodic response analysis and response pixel determination.

[0257] Example 13:

[0258] In one possible implementation, the periodic response analysis module includes: a periodic division unit, used to divide the pixel-level fluorescence time series matrix into multiple periodic sub-matrices according to the periodic boundaries corresponding to the periodic stimulus; a consistency analysis unit, used to calculate the periodic response consistency based on the response amplitude and response phase changes of corresponding pixels within multiple periods; a noise analysis unit, used to extract frequency components of the frequency band adjacent to the target response frequency and calculate adjacent-channel noise characteristics; and a scoring calculation unit, used to perform weighted calculation based on the response amplitude, adjacent-channel noise characteristics, and periodic response consistency to obtain a response score. In this embodiment, the periodic response analysis module is used to perform periodic stability analysis on the response parameters output by the digital frequency domain demodulation module. By dividing the continuous fluorescence time series according to the stimulus period, comparing the response changes of the same pixel position in different periods, and combining the noise level of non-target frequency components near the target response frequency, the effective response degree of each pixel position is evaluated to obtain a response score for determining the response pixel.

[0259] In the specific implementation process, the period division unit is connected to the digital frequency domain demodulation module and is used to segment the pixel-level fluorescence time sequence matrix according to the period boundary corresponding to the periodic stimulus signal.

[0260] Specifically, the periodization unit acquires the start and end times of the periodic stimulus signal, and determines the length of a single stimulus period based on the time interval between two adjacent start times. Subsequently, the image acquisition time points in the pixel-level fluorescence time series matrix are matched with the boundaries of each period, dividing the pixel fluorescence time data within the same stimulus period time range into the same data segment. After segmenting the continuous fluorescence time series according to all period boundaries, multiple periodic sub-matrices are obtained. Each periodic sub-matrix contains fluorescence intensity change data corresponding to all effective pixel positions within a stimulus period, enabling subsequent analysis to compare response characteristics within different stimulus periods for the same pixel position.

[0261] The consistency analysis unit is connected to the period division unit and is used to calculate the response consistency of corresponding pixels between different stimulus periods.

[0262] Specifically, the consistency analysis unit calculates the response parameters for each pixel position in each periodic submatrix. For the same pixel position, it obtains the response amplitude data corresponding to that pixel in different periods and calculates the average value of the response amplitude across all periods.

[0263] Subsequently, the difference between the response amplitude corresponding to each cycle and the average value is calculated to obtain the change in response amplitude for each cycle. The squares of each change are then calculated, and the squared results for all cycles are summed and averaged according to the number of cycles to obtain data representing the degree of change in response amplitude at that pixel location.

[0264] For response phase changes, the consistency analysis unit obtains the response phase data corresponding to the same pixel location in different periods. First, it calculates the average response phase of all periods, and then calculates the deviation between the response phase of each period and the average response phase. Based on the phase deviation values ​​of each period, it determines the degree of phase change corresponding to that pixel location.

[0265] Based on the degree of change in response amplitude and the degree of change in response phase, the consistency analysis unit calculates the periodic response consistency. Specifically, the degree of change in response amplitude is converted into an amplitude stability evaluation value, and the degree of change in response phase is converted into a phase stability evaluation value. Then, the two evaluation values ​​are weighted according to preset weights to obtain the periodic response consistency.

[0266] The preset weights are determined based on the periodic stimulation method and response signal characteristics. In applications where the stimulation period is stable and phase changes significantly influence response judgment, the weight corresponding to the phase stability evaluation value is increased; in applications where fluorescence intensity changes significantly influence response judgment, the weight corresponding to the amplitude stability evaluation value is increased. The noise analysis unit is connected to the digital frequency domain demodulation module to obtain noise characteristics corresponding to frequency bands adjacent to the target response frequency.

[0267] Specifically, the noise analysis unit acquires the fluorescence time series corresponding to each pixel position and performs frequency decomposition processing on the fluorescence time series to obtain the signal components corresponding to different frequency positions.

[0268] During frequency decomposition, the target response frequency is used as the center frequency, and guard bands are set on both sides of the target response frequency. The guard bands are used to exclude the effective response signal corresponding to the target response frequency, so as to prevent the effective response components at the target response frequency from participating in noise calculation.

[0269] Subsequently, the noise analysis unit extracts frequency components located outside the guard band but adjacent to the target response frequency, and calculates the signal strength corresponding to these frequency components. Specifically, the amplitude of the extracted adjacent frequency components is calculated, and the amplitudes of each adjacent frequency component are accumulated to obtain the adjacent frequency noise features at the corresponding pixel location.

[0270] The scoring calculation unit is connected to the consistency analysis unit and the noise analysis unit respectively, and is used to calculate the response score based on the response amplitude, adjacent frequency noise characteristics and periodic response consistency.

[0271] Specifically, the scoring calculation unit acquires the response amplitude data output by the digital frequency domain demodulation module, the adjacent-channel noise characteristics output by the noise analysis unit, and the periodic response consistency data output by the consistency analysis unit.

[0272] First, response strength evaluation parameters are calculated based on the response amplitude and adjacent channel noise characteristics. Specifically, the response amplitude corresponding to the target response frequency is used as the effective response data, and the adjacent channel noise characteristics are used as the interference data. By calculating the ratio between the effective response data and the interference data, the data representing the target frequency response strength is obtained.

[0273] Subsequently, the response intensity evaluation parameter and the periodic response consistency are multiplied by their respective weights, and the two weighted results are summed to obtain the response score corresponding to the pixel location. The weights are determined based on the actual detection conditions. Specifically, in the determination process, response score data corresponding to background pixels are acquired under conditions of no target response. By adjusting different weight combinations, the number of false response pixels in the background area and the effective response recognition results in the target area are compared, and the final weight is determined based on the preset allowable number of false responses.

[0274] Through the above processing, the periodic response analysis module can simultaneously evaluate the target response frequency signal intensity, periodic repetition characteristics, and adjacent-channel noise influence at the pixel location, so that the response score can reflect the periodic stimulus response characteristics corresponding to the pixel location, providing an evaluation basis for the subsequent response pixel determination module to determine the effective response area.

[0275] Example 14:

[0276] In one possible implementation, the response pixel determination module includes: a background statistics unit, used to obtain the response score corresponding to the pixel in the background region and calculate the statistical parameters of the background score sequence; a threshold generation unit, used to generate a response determination threshold according to the statistical parameters and a preset threshold coefficient; and a response mask generation unit, used to filter response pixels according to the response determination threshold and generate a response pixel mask.

[0277] The response pixel determination module in this embodiment is used to determine the effective response area based on the pixel response score. By performing statistical analysis on the response score of the background area, a determination standard based on the background noise level is established, and the pixel positions that meet the periodic response conditions are selected according to the determination standard to generate a response pixel mask for vascular structure fusion.

[0278] In the specific implementation process, the background statistics unit is connected to the periodic response analysis module to obtain the response score corresponding to the pixels in the background area.

[0279] Specifically, the background statistics unit first obtains the response scoring matrix and extracts the response scores for corresponding pixel positions based on pre-defined background region coordinate information. The background region is a region that does not contain the target vascular structure, which can be pre-defined based on the experimental acquisition area or selected based on the vascular structure recognition results.

[0280] When using a preset background region method, the background statistics unit reads the response score data of the corresponding position from the response score matrix based on the stored spatial coordinates of the background region. When using a low response score region method to determine the background region, the background statistics unit sorts all pixels according to their response score values, selects pixel positions that meet the background filtering criteria, and extracts the corresponding response scores. The extracted background pixel response scores are arranged in the order they were read to form a background score sequence. Subsequently, statistical calculations are performed on the background score sequence to obtain statistical parameters describing the background response level.

[0281] Specifically, the background statistics unit first sorts all the rating data in the background rating sequence. When the background rating sequence contains an odd number of data points, the rating value in the middle position after sorting is selected as the median statistical parameter; when the background rating sequence contains an even number of data points, the two rating values ​​in the middle position after sorting are selected, and the average of the two rating values ​​is calculated as the median statistical parameter.

[0282] Furthermore, the background statistics unit calculates the dispersion of the background rating sequence. Specifically, the difference between each rating value in the background rating sequence and the median statistical parameter is calculated. The absolute values ​​of these differences are then processed, and all absolute values ​​are summed. Finally, the average value is calculated based on the number of ratings in the background rating sequence to obtain the background rating dispersion parameter.

[0283] Through the above statistical processing, the median statistical parameter representing the background response level and the discrete parameter representing the degree of change in the background score are obtained, which serve as the data basis for subsequent response determination threshold calculation.

[0284] The threshold generation unit is connected to the background statistics unit and is used to generate a response judgment threshold based on the background statistical parameters.

[0285] Specifically, the threshold generation unit obtains the median statistical parameters of the background score, the discrete parameters of the background score, and the preset threshold coefficient, and generates the response judgment threshold according to the preset calculation rules.

[0286] The preset threshold coefficient is determined based on the actual imaging conditions. Specifically, fluorescence image data is acquired under conditions where no target periodic response signal is present, and the background region response score is calculated according to the aforementioned processing flow. Subsequently, different threshold coefficients are set to filter response pixels, and the number of misjudged background pixels corresponding to each threshold coefficient is counted.

[0287] A threshold coefficient is determined based on a preset allowable number of misjudged pixels. When the number of misjudged background pixels corresponding to the threshold coefficient meets the set requirements, this threshold coefficient is used as the preset threshold coefficient for subsequent response determination. This method allows the response determination threshold to be determined based on the actual background noise level. The threshold generation unit calculates the response determination threshold by combining the median statistical parameter and discrete parameter of the background score based on the determined threshold coefficient. Specifically, the median statistical parameter of the background score is used as the basic score reference, and the discrete parameter of the background score is used as the background fluctuation correction amount. The influence ratio corresponding to the background fluctuation correction amount is adjusted according to the threshold coefficient, and the adjusted result is combined with the basic score reference to obtain the final response determination threshold. The response mask generation unit is connected to the threshold generation unit and is used to filter response pixels and generate response pixel masks based on the response determination threshold.

[0288] Specifically, the response mask generation unit obtains the response score corresponding to each pixel position in the response score matrix and compares it with the response judgment threshold pixel by pixel.

[0289] When the response score corresponding to a certain pixel location reaches or exceeds the response determination threshold, the pixel location is marked as a response pixel; when the response score corresponding to the pixel location is lower than the response determination threshold, the pixel location is marked as a non-response pixel.

[0290] After completing the comparison of all pixels, the response mask generation unit arranges the judgment results of each pixel according to the spatial coordinates of the original image. Specifically, the positions corresponding to the response pixels are set as valid markers, and the positions corresponding to the non-response pixels are set as invalid markers, generating a response pixel mask with the same size as the fluorescence image.

[0291] The response pixel mask records the spatial locations that satisfy the periodic stimulus response conditions, and is used for spatial fusion with the vascular structure mask. Through the above processing, the response pixel determination module can automatically determine the determination criteria based on the actual background noise level, avoiding deviations in response region identification results under different imaging conditions caused by using a fixed threshold.

[0292] Example 15:

[0293] In one possible implementation, the vascular structure fusion module includes: a vascular reference image generation unit, used to perform time averaging or maximum intensity projection on a visible light or near-infrared fluorescence image sequence to obtain a vascular structure reference image; a vascular enhancement unit, used to perform multi-scale vascular enhancement on the vascular structure reference image to obtain a vascular structure feature matrix; a vascular segmentation unit, used to segment the vascular structure feature matrix according to a structure segmentation threshold and generate a vascular structure mask through morphological processing; and a fusion unit, used to perform intersection operation or weighted fusion of the response pixel mask and the vascular structure mask to generate a response vascular image and / or a response vascular mask.

[0294] The vascular structure fusion module in this embodiment is used to map the response pixel region obtained from the periodic stimulus response to the vascular spatial structure in the visible light or near-infrared fluorescence image. Through vascular structure extraction, vascular region segmentation and response information fusion processing, the final output result contains both the vascular spatial location and the corresponding response information.

[0295] In the specific implementation process, the vascular reference image generation unit is connected to the periodic stimulation and image acquisition module to extract vascular structure reference information from visible light or near-infrared fluorescence image sequences.

[0296] Specifically, the vascular reference image generation unit acquires a sequence of continuously acquired visible or near-infrared fluorescence images and reads the fluorescence intensity data corresponding to each image frame in chronological order of image acquisition. When generating a vascular structure reference image using time averaging, for each spatial pixel location, the fluorescence intensity data of that location across all image frames is extracted. The fluorescence intensity values ​​at each time point are accumulated, and the accumulated result is divided by the number of image frames involved in the calculation to obtain the average fluorescence intensity value corresponding to that spatial location. These average fluorescence intensity values ​​are then arranged according to all spatial locations to form a time-averaged vascular structure reference image. When generating a vascular structure reference image using maximum intensity projection, for each spatial pixel location, the fluorescence intensity values ​​of that location across all image frames are compared. The fluorescence intensity value with the largest value is selected as the projection intensity value for that spatial location, and these values ​​are arranged according to the original image spatial coordinates to form a maximum intensity projection vascular structure reference image.

[0297] Through the above processing, the blood vessel-related fluorescence information in continuous fluorescence images is fused in the temporal direction, converting the structural information corresponding to the blood vessel region into stable image feature data. The blood vessel enhancement unit is connected to the blood vessel reference image generation unit and is used to perform multi-scale blood vessel enhancement processing on the blood vessel structure reference image.

[0298] Specifically, the vascular enhancement unit establishes image processing windows at multiple scales based on preset scale parameters and performs local structural analysis on the vascular structure reference image for each scale. At each scale, it calculates the gray-level change features in different directions within the local image region and extracts linear structural response values ​​with continuous extension directions. Subsequently, it performs spatial correspondence analysis on the structural response values ​​obtained at different scales, combines the structural response values ​​at each scale corresponding to the same pixel location, obtains the comprehensive structural response data corresponding to that pixel location, and arranges them according to the image spatial coordinates to form a vascular structure feature matrix.

[0299] Each data point in the vascular structure feature matrix represents the degree of vascular structure response at the corresponding spatial location, used to distinguish between vascular and non-vascular regions.

[0300] The vessel segmentation unit is connected to the vessel enhancement unit and is used to segment the vessel structure feature matrix into regions based on the structure segmentation threshold.

[0301] Specifically, the vessel segmentation unit first acquires the structural response data corresponding to the non-vascular background region and performs statistical analysis on the structural response values ​​of each pixel within the background region. By calculating the average level and range of variation of the structural response values ​​in the background region, the distribution of the background structural response is determined.

[0302] The initial structure segmentation threshold is determined based on the background structure response distribution. Specifically, different candidate thresholds are applied to the vascular structure feature matrix for segmentation. The number of misidentified non-vascular regions and the number of missed vascular regions corresponding to each candidate threshold are counted. The final structure segmentation threshold is then determined based on a preset allowable error condition.

[0303] When a pixel value in the vascular structure feature matrix reaches the structural segmentation threshold, that pixel location is identified as a vascular structure region; when the pixel value is below the structural segmentation threshold, that pixel location is identified as a non-vascular region. An initial vascular segmentation result is generated based on the determination results of all pixel locations.

[0304] Subsequently, the vessel segmentation unit performs morphological processing on the initial vessel segmentation results. Specifically, it fills in any broken areas within the vessel region to maintain the integrity of continuous vessel regions; and removes isolated regions with an area smaller than a preset area threshold to reduce the influence of non-vascular structures. After morphological processing, a vessel structure mask representing the spatial location of the vessels is generated. The fusion unit is connected to both the response pixel determination module and the vessel segmentation unit to fuse the response pixel mask and the vessel structure mask. When using the intersection operation, the fusion unit reads the pixel states at the same spatial location in both the response pixel mask and the vessel structure mask. For the same spatial location, if both the response pixel mask and the vessel structure mask represent valid regions, that location is determined as a responsive vessel region; if either mask represents an invalid region, that location is determined as a non-responsive vessel region. A response vessel feature matrix is ​​obtained through pixel-by-pixel judgment. When using a weighted fusion method, the fusion unit obtains the response score corresponding to the response pixel and the structural response value corresponding to the vessel structure. For the same spatial location, the response score and structural response value are multiplied by a preset fusion weight, and then the two calculation results are summed to obtain the fusion feature value corresponding to that location.

[0305] The fusion weights are determined based on actual imaging requirements. Specifically, the fusion results under different weight combinations are obtained from the labeled vascular regions, and the fusion weights are determined based on the accuracy of vascular region identification and the number of misidentifications of non-vascular regions. Finally, the fusion unit restores the fusion feature values ​​corresponding to each location to the original image spatial location based on the spatial coordinate information in the response vascular feature matrix, generating a response vascular image; or it performs binarization processing based on whether each spatial location meets the response vascular determination criteria, generating a response vascular mask. Through the above processing, the vascular structure fusion module maps the fluorescence response information induced by periodic stimulation to the vascular spatial structure information, enabling the output result to simultaneously represent the vascular distribution location and the response features corresponding to the vascular region, providing imaging data for subsequent vascular function analysis and response state evaluation.

[0306] The near-infrared region of the present invention that responds to vascular imaging includes near-infrared region I and near-infrared region II.

[0307] This embodiment uses the near-infrared II region for testing. The flowchart of the operation for vascular imaging based on near-infrared II fluorescence is as follows:

[0308] Step 1: Preliminary Preparation. Turn on the deep-cooled SWIR camera, excitation source, water-cooling module, and computer, and complete the equipment self-test. Select the appropriate filter from the filter turntable according to the probe's emission wavelength. The filter turntable is driven by a stepper motor and can be configured with various long-pass filters covering the range of 900nm to 1700nm. Set the camera's exposure time, frame rate, gain, acquisition field of view, and data storage path.

[0309] Step two: Intravenous injection of near-infrared dye. The dye includes, but is not limited to, ICG, AIE-type probes, quantum dots, polymer dots, IR-820, IR-783, or other near-infrared fluorescent probes. Approximately one minute after injection, once the dye has circulated throughout the body or target vascular system via the bloodstream, the excitation light source is activated, and the appropriate illumination power density and spot size are adjusted; camera focusing and field of view localization are then completed.

[0310] Step 3: Select a peripheral stimulation paradigm, such as leg electrical stimulation, hand electrical stimulation, mechanical stimulation, tactile stimulation, visual stimulation, auditory stimulation, thermal stimulation, or other evoked conditions; select stimulation parameters, including stimulation duration of 1-10 seconds, interval of 1-10 seconds after stimulation, and duration of 1-10 cycles. At the start of stimulation, simultaneously activate camera acquisition and frequency domain calculation to ensure a clear temporal correspondence between the stimulation event and the image sequence.

[0311] Step four: Save the data after calculation. The system generates a response vascular map, a response vascular mask, the original image sequence, processing parameters, stimulation parameters, and experimental records, and saves the results to a computer or external storage device.

[0312] Hardware and parameter configuration:

[0313] Excitation light source module: Equipped with three excitation wavelengths of 808nm, 980nm, and 1064nm, or configured with one or more wavelengths according to the dye absorption characteristics; the illumination power density can be adjusted from 1 to 200mW / cm², more preferably from 10 to 60mW / cm². External water cooling is used for heat dissipation to ensure stability under high power or long-term output conditions.

[0314] Beam homogenization module: The excitation light undergoes beam homogenization, which can be completed within the optical fiber, and then a uniform beam is directly output by the collimator. By changing the beam homogenization component, the presence or absence of speckle in the beam spot can be adjusted. The beam spot size can be adjusted by changing the illumination distance.

[0315] Acquisition Module: Equipped with a camera for acquiring near-infrared II fluorescence signals from 900nm to 1700nm. The camera can be externally water-cooled to maintain a low operating temperature, thereby keeping dark current and thermal noise levels low.

[0316] Imaging module: It can be equipped with SWIR lenses of different magnification to achieve imaging of different fields of view. Lens magnifications such as 0.1×, 0.2×, 0.5×, and 1× can cover the target areas of small animal brains, large animals, human surgical areas, or peripheral limbs. By changing the lens and adjusting the working distance, different fields of view and spatial resolutions can be switched.

[0317] Filter module: It can be equipped with a filter turntable, which can carry filters of different wavelengths and is driven by a stepper motor to switch between them; the filter combination can cover a variety of long-pass filters or bandpass filters in the range of 900nm to 1700nm.

[0318] Data processing and display module: The signal processing module performs frequency domain extraction calculations, and the processing results are displayed on the host computer in real time.

[0319] (3) Theoretical expression for frequency domain extraction

[0320] Assume the acquisition frame rate is The stimulation modulation frequency is The stimulation cycle is:

[0321] ;

[0322] Let the total number of stimulation cycles be... The stimulation and camera acquisition started and ended synchronously, with a total acquisition time of:

[0323] ;

[0324] The number of frames acquired within each stimulation cycle is:

[0325] ;

[0326] The total number of frames captured is:

[0327] ;

[0328] No. The frame acquisition time is:

[0329] ;

[0330] Let coordinates First The fluorescence intensity of the frame is .

[0331] Fourier frequency domain extraction is performed after each stimulation cycle. At the end of each stimulation cycle, the number of frames acquired was:

[0332] ;

[0333] For each pixel In the already collected Intra-frame calculation of stimulus frequency Fourier coefficients at:

[0334] ;

[0335] Take the intensity at that frequency and assign it to that pixel:

[0336] ;

[0337] in, For the first After each stimulation cycle, the pixel The response intensity. This process is performed on all pixels across the entire field of view to obtain the [response intensity]. Response angiography after each cycle. Final result output after stimulation completion:

[0338] .

[0339] Specifically, mice (C57 / B6 male, 7 weeks old, without craniotomy, without anesthesia) were placed in a brain fixation device, and the fixed brain was positioned directly below the camera lens.

[0340] Turn on the camera water chiller (set the water flow temperature to 20°C), the camera (turn on low temperature mode, do not store data stream), the laser (do not turn on), and the peripheral stimulator (in this embodiment, the electrical stimulator) in sequence.

[0341] 0.04 mg ICG was injected into the mouse via the tail vein; the laser wavelength of 808 nm was selected and the laser was turned on, and the optical power density was set to 30 mW / cm2; the filter turntable was rotated to select a 1200 nm long-pass filter; the camera exposure time was set to 100 ms and the frame rate to 10 Hz; a 1× lens was selected; the electrical stimulation paradigm of the left leg was selected (+0.04 mA 100 μs, -0.01 mA 400 μs, 10 Hz bidirectional electrical stimulation, stimulation duration 1 s, interval 9 s, for a total of 10 cycles) and other parameters.

[0342] One minute after injection, the dye has spread throughout the body's blood vessels. Simultaneously, stimulation, acquisition, and calculation (extracting signals at a frequency of 0.1Hz) are initiated, and the test results are displayed on the screen at the end of each stimulation cycle.

[0343] Figure 5 The image shows near-infrared fluorescence of blood vessels in the unextracted rat brain region II, mostly superficial vessels. Figure 6 This is a response vascular map extracted from the target frequency domain, showing a few blood vessels and... Figure 5 The blood vessels in the corresponding locations overlap, while most blood vessels are in Figure 5 Unidentifiable, which means these blood vessels originate from deep brain regions; Figure 6 Blood vessels with a diameter of 40 μm can be distinguished. Figure 5 , 6 It's electrical stimulation. Figure 7 , 8 It is light stimulation. Figure 9 , 10 It is an auditory stimulus, among which Figure 5 , 7 Image 9 is the original image obtained directly from fluorescence imaging before extraction; it mostly shows superficial blood vessels. Figure 6 , 8 Image 10 is the image after frequency domain extraction. Some blood vessel routes are not present in the original image because these blood vessels are in deeper layers, and the light they emit is scattered by the tissue, so the structure of these deep blood vessels cannot be distinguished in the original image.

[0344] like Figures 5 to 10 As shown, electrical, light, and acoustic stimulation were applied to the target area in vivo, and images of the vascular response under each stimulation condition were acquired to characterize the vascular structure and deep vascular components within the living tissue. The imaging field of view for each test group was 1cm × 1cm.

[0345] like Figure 5 As shown, Figure 5 The original images of the target area in vivo, obtained before or during electrical stimulation testing, are shown. These images include visible vascular structures and the corresponding tissue background within the target area. The electrical stimulation employs a bidirectional pulsed stimulation method, where the positive pulse has a current amplitude of +0.04 mA and a pulse width of 100 μs; the negative pulse has a current amplitude of -0.01 mA and a pulse width of 400 μs. The positive and negative pulses constitute a bidirectional stimulation pulse group, which is periodically output at a repetition frequency of 10 Hz, i.e., repeated once every 100 ms. Based on the aforementioned current amplitude and pulse width, the positive and negative pulses have equal charge magnitudes but opposite polarities, thereby reducing the possibility of continuous unidirectional current polarizing the living tissue.

[0346] In this embodiment, each electrical stimulation cycle includes a 1-second electrical stimulation phase and a 9-second interval phase. The duration of a single electrical stimulation cycle is 10 seconds, and a total of 10 electrical stimulation cycles are performed, with a total test time of 100 seconds.

[0347] like Figure 6 As shown, Figure 6 The vascular response results obtained under the above-described electrical stimulation conditions are shown. Compared to Figure 5 The original image shown, Figure 6 Several vascular branches that were not obvious, not prominently displayed, or difficult to identify from the tissue background appeared in the original image. These vascular branches can be used to characterize the deep vascular components that respond to electrical stimulation. By comparing the original vascular structure with the stimulation response results, the ability to identify deep vascular branches and their spatial distribution within the target area in vivo can be improved.

[0348] like Figure 7 As shown, Figure 7The original images of the target area in vivo, obtained before or during light stimulation testing, are shown. These original images include visible vascular structures within the target area and the corresponding tissue background. Light stimulation employs a black and white checkerboard pattern, consisting of 64 squares arranged in 8 rows and 8 columns, with adjacent squares alternating between black and white.

[0349] The display screen used to display the black and white checkerboard pattern has a refresh rate of 60Hz, and the black and white squares exchange their brightness states at a reversal frequency of 6Hz. During light stimulation, the original black squares turn white, and the original white squares turn black, and this reversal occurs periodically at a predetermined frequency to create visual light stimulation.

[0350] In this embodiment, each light stimulation cycle includes a light stimulation phase lasting 1 second and an interval phase lasting 9 seconds. The duration of a single light stimulation cycle is 10 seconds, and a total of 10 light stimulation cycles are performed, with a total test time of 100 seconds.

[0351] like Figure 8 As shown, Figure 8 The results of the vascular response obtained under the above-described light stimulation conditions are shown. Figure 8 The responsive vascular structures shown in Figure 7 The tissue is not clearly shown in the original image, or cannot be directly identified from the original tissue background. Therefore, Figure 8 The vascular structures shown can be used to characterize deep vascular components that respond to light stimulation. This result demonstrates that by applying periodic black-and-white checkerboard light stimulation, vascular information related to the hemodynamic response of a target region in vivo can be obtained, which is beneficial for identifying deep vascular structures not shown in the original image.

[0352] like Figure 9 As shown, Figure 9 The original images of the target area in vivo, obtained before or during acoustic stimulation testing, are shown. These original images include visible vascular structures within the target area and the corresponding tissue background. The acoustic stimulation frequency was 16 kHz, and the sound pressure level was 68 dB.

[0353] In this embodiment, each sound stimulation cycle includes a 1-second sound stimulation phase and a 9-second interval phase. The duration of a single sound stimulation cycle is 10 seconds, and a total of 10 sound stimulation cycles are performed, with a total test time of 100 seconds.

[0354] like Figure 10 As shown, Figure 10 The vascular response results obtained under the above acoustic stimulation conditions are shown. Compared to Figure 9 The original image shown, Figure 10Newly added, enhanced, or poorly identifiable vascular structures in the original images can be characterized as deep vascular components that respond to acoustic stimulation. By comparing vascular images before and after acoustic stimulation, information on the distribution of blood vessels related to the acoustic stimulation response within the target area in vivo can be obtained.

[0355] In the above embodiments, electrical stimulation, light stimulation, and acoustic stimulation all employ a periodic stimulation pattern of 1 second duration followed by 9 seconds intervals, repeated for 10 cycles each. This ensures that different stimulation methods have the same stimulation time sequence and total test duration, facilitating comparison of vascular responses induced by different types of external stimuli. By comparing the vascular response results obtained under each stimulation condition with the corresponding original vascular images, the interference of tissue background and superficial static vascular structures on the recognition results can be reduced, and the vascular branches that respond to stimulation can be highlighted, providing image evidence for the detection of deep vascular structures and their functional responses within the target area in vivo.

[0356] The method described in this application utilizes the temporal synchronization between a periodic stimulation signal and a near-infrared II fluorescence image sequence to extract the vascular fluorescence response at the target stimulation envelope frequency. This reduces interference from static tissue background, non-responsive vascular signals, and non-periodic noise, thereby obtaining stimulus-related responsive vascular images. This method enhances weakly responsive vascular branches that are not prominent in the original fluorescence image, exhibits high detection capability for responsive vessels that may be located in deeper tissue regions, and is applicable to different stimulation forms such as electrical, visual, and auditory stimulation. Within a 1cm × 1cm field of view, responsive vessels with a diameter of approximately 40μm can be resolved, and response analysis results can be output after each stimulation cycle.

[0357] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A fluorescence-based response vascular imaging method, characterized in that, include: S1. Under periodic stimulation, the visible light or near-infrared fluorescence image sequence of the target area of ​​the living organism is acquired synchronously, and the stimulation time sequence information corresponding to the periodic stimulation is recorded. The acquisition time point and the stimulation time sequence information are synchronized to obtain the synchronous fluorescence sequence. S2. Extract the fluorescence intensity of each image frame in the synchronous fluorescence sequence according to the same spatial location, and arrange them according to the acquisition time to construct a pixel-level fluorescence time sequence matrix; S3. Determine the target response frequency based on the periodic stimulation signal applied to the live target area, and generate the in-phase reference vector and orthogonal reference vector required for digital frequency domain demodulation based on the target response frequency and the image acquisition time series. S4. Perform inner product operation between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain response parameters, and determine the response amplitude and response phase of each pixel based on the response parameters; S5. According to the periodic boundary of the periodic stimulus, the pixel-level fluorescence time series matrix is ​​divided into multiple periodic data segments. The periodic response consistency is determined based on the response amplitude and response phase of each periodic data segment, and the response score is calculated in combination with the noise characteristics of the frequency band adjacent to the target frequency. S6. Determine the response judgment threshold based on the response score of the background pixels, and determine the pixels that reach the response judgment threshold as response pixels; S7. Extract vascular structure features from the visible light or near-infrared fluorescence image sequence to obtain a vascular structure mask, and spatially fuse the response pixels with the vascular structure mask to generate a response vascular image and / or a response vascular mask.

2. The fluorescence-based vascular imaging method according to claim 1, characterized in that, S1 includes: Obtain fluorescence image sequences in the visible or near-infrared region and their corresponding image timestamp sequences; Obtain the modulation reference sequence and reference timestamp sequence corresponding to the periodic modulation conditions; A clock mapping model is established based on the image timestamp sequence and the reference timestamp sequence to obtain the time offset parameter; The modulation reference sequence is interpolated and resampled according to the time offset parameter to generate a reference vector corresponding to each image frame; The temporal alignment error between the reference vector and each image frame is calculated, and data with temporal alignment errors less than a preset error threshold are combined into the fluorescence sequence.

3. The fluorescence-based vascular imaging method according to claim 1, characterized in that, S2 includes: A spatial coordinate index matrix is ​​established based on the image size of the fluorescence sequence, and a pixel index is assigned to each spatial location. The fluorescence intensity of each image frame is extracted according to the pixel index, and each image frame is converted into a frame intensity vector; The frame intensity vector is filtered according to a preset saturation threshold and a missing rate threshold to generate a set of valid pixel indices; The intensity vectors of each frame are arranged sequentially according to the acquisition time, and the fluorescence intensity corresponding to the effective pixel index set is extracted to obtain the initial temporal matrix; The initial timing matrix is ​​set as a pixel-level fluorescence timing matrix with rows corresponding to spatial locations and columns corresponding to acquisition times.

4. The fluorescence-based response vascular imaging method according to claim 1, characterized in that, S3 specifically includes: The target response frequency is determined based on the periodic stimulation signals applied to the target area in a living organism; Based on the target response frequency and the acquisition time points corresponding to each image frame, generate in-phase reference sequences and orthogonal reference sequences corresponding to the target response frequency, respectively. The in-phase reference sequence and the quadrature reference sequence are normalized to obtain the in-phase reference vector and quadrature reference vector required for digital frequency domain demodulation. The in-phase reference vector and the quadrature reference vector have a 90° phase difference and are used to extract the in-phase response component and the quadrature response component of the fluorescence signal at the target response frequency, respectively.

5. The fluorescence-based response vascular imaging method according to claim 1, characterized in that, S4 includes: The pixel-level fluorescence time series matrix is ​​subjected to mean removal and normalization processing row by row to obtain a standard time series matrix; The in-phase response vector is obtained by performing an inner product operation between each row vector of the standard timing matrix and the in-phase reference vector. The orthogonal response vector is obtained by performing a matrix inner product between the standard time series matrix and the orthogonal reference vector. The in-phase response vector and the orthogonal response vector are combined into a pixel response parameter matrix; The response amplitude and response phase are calculated based on the in-phase and quadrature response values ​​corresponding to each pixel, and the response amplitude matrix and response phase matrix are obtained by arranging them according to spatial coordinates.

6. The fluorescence-based vascular imaging method according to claim 1, characterized in that, S5 includes: Based on the periodic boundaries of the periodic stimulus, the pixel-level fluorescence time series matrix is ​​divided into multiple periodic sub-matrices; Calculate the response amplitude and response phase of each pixel in each periodic submatrix to construct the periodic response feature matrix; The periodic response consistency is calculated based on the amplitude and phase dispersion of each pixel over multiple periods. Extract adjacent frequency components on both sides of the target frequency that are outside the preset guard band, and calculate the adjacent frequency noise characteristics of each pixel; The response score for each pixel is obtained by weighting the ratio of the response amplitude to the adjacent-channel noise feature and combining it with the periodic response consistency.

7. The fluorescence-based response vascular imaging method according to claim 1, characterized in that, S6 includes: The response scores of each pixel are arranged according to their spatial coordinates to obtain the response score matrix; Background pixels are selected from a preset background area or a low-response score area to construct a background score sequence; Calculate the median and dispersion of the background rating sequence to obtain the background noise statistical parameters; Calculate the response judgment threshold based on the background noise statistical parameters and the preset threshold coefficient; The pixels in the response scoring matrix that reach the response determination threshold are identified as response pixels, and a response pixel mask is generated.

8. The fluorescence-based response vascular imaging method according to claim 1, characterized in that, S7 includes: The visible light or near-infrared fluorescence image sequence is time-averaged or projected with maximum intensity to obtain a reference image of the vascular structure. Multi-scale vascular enhancement is performed on the vascular structure reference image to obtain a vascular structure feature matrix; The vascular structure feature matrix is ​​segmented according to the structural segmentation threshold, and a vascular structure mask is obtained through morphological processing. The response pixel mask corresponding to the response pixel is intersected or weighted and fused with the blood vessel structure mask to obtain the response blood vessel feature matrix. Based on the response vascular feature matrix, the spatial location and response intensity are recovered to generate a response vascular image and / or a response vascular mask.

9. A fluorescence-based responsive vascular imaging system, applicable to the fluorescence-based responsive vascular imaging method according to any one of claims 1-8, characterized in that, include: The periodic stimulation and image acquisition module is used to apply periodic stimulation to the live target area and simultaneously acquire the visible light or near-infrared fluorescence image sequence of the live target area, while recording the stimulation timing information corresponding to the periodic stimulation. A time synchronization module, connected to the periodic stimulus and image acquisition module, is used to perform time synchronization based on the acquisition time point corresponding to the visible light or near-infrared fluorescence image sequence and the stimulus timing information to obtain a synchronized fluorescence sequence. A pixel-level time-series matrix construction module, connected to the time synchronization module, is used to extract the fluorescence intensity corresponding to each image frame in the synchronized fluorescence sequence according to the same spatial location, and arrange them according to the acquisition time order to construct a pixel-level fluorescence time-series matrix. A digital frequency domain demodulation module, connected to the pixel-level time-series matrix construction module, is used to determine the target response frequency based on the periodic stimulation signal applied to the live target region, and generate an in-phase reference vector and an orthogonal reference vector based on the target response frequency and the image acquisition time series; the pixel-level fluorescence time-series matrix is ​​then used to perform inner product operations with the in-phase reference vector and the orthogonal reference vector to obtain response parameters corresponding to the target response frequency, and the response amplitude and response phase of each pixel are determined based on the response parameters; The periodic response analysis module, connected to the digital frequency domain demodulation module, divides the pixel-level fluorescence time series matrix into multiple periodic data segments according to the periodic boundaries corresponding to the periodic stimulus. It determines the periodic response consistency based on the response amplitude and response phase corresponding to each periodic data segment, and calculates the response score of each pixel by combining the noise characteristics of the frequency band adjacent to the target response frequency. The response pixel determination module is connected to the periodic response analysis module. It is used to determine the response determination threshold based on the response score corresponding to the background pixel, and to determine the pixel that reaches the response determination threshold as the response pixel, and to generate a response pixel mask. The vascular structure fusion module, connected to the periodic stimulation and image acquisition module and the response pixel determination module, is used to extract vascular structure features from the visible light or near-infrared fluorescence image sequence, generate a vascular structure mask, and spatially fuse the response pixel mask with the vascular structure mask to generate a response vascular image and / or a response vascular mask.

10. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The time synchronization module includes: The timestamp acquisition unit is used to acquire the image timestamp sequence corresponding to the visible light or near-infrared region fluorescence image sequence, as well as the stimulus reference sequence and reference timestamp sequence corresponding to the periodic stimulus; A clock mapping unit is used to establish a clock mapping model based on the image timestamp sequence and the reference timestamp sequence to obtain time offset parameters; A reference resampling unit is used to interpolate and resample the stimulus reference sequence according to the time offset parameter to generate a reference vector corresponding to each image frame. An error filtering unit is used to calculate the time alignment error between the reference vector and each image frame, and to filter and obtain a synchronous fluorescence sequence according to a preset error threshold.

11. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The pixel-level temporal matrix construction module includes: A spatial indexing unit is used to establish a spatial coordinate index matrix based on the image size of the synchronized fluorescence sequence; A fluorescence intensity extraction unit is used to extract the fluorescence intensity of each image frame at a corresponding spatial location based on the spatial coordinate index matrix. The effective pixel filtering unit is used to filter effective pixels based on a preset saturation threshold and a missing rate threshold. The time-series arrangement unit is used to arrange the fluorescence intensity data corresponding to each image frame in the order of acquisition time, forming a pixel-level fluorescence time-series matrix.

12. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The digital frequency domain demodulation module includes: The target frequency determination unit is used to determine the target response frequency based on the stimulation period of the periodic stimulus signal. The reference vector generation unit is used to generate an in-phase reference sequence and an orthogonal reference sequence based on the target response frequency and the acquisition time point corresponding to each image frame, and to normalize the in-phase reference sequence and the orthogonal reference sequence to obtain an in-phase reference vector and an orthogonal reference vector. The response component calculation unit is used to perform inner product operations between the pixel-level fluorescence timing matrix and the in-phase reference vector and the orthogonal reference vector respectively to obtain the in-phase response component and the orthogonal response component. The response parameter calculation unit is used to calculate the response amplitude and response phase based on the in-phase response component and the quadrature response component.

13. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The periodic response analysis module includes: The periodic division unit is used to divide the pixel-level fluorescence time series matrix into multiple periodic sub-matrices according to the periodic boundaries corresponding to the periodic stimulus; The consistency analysis unit is used to calculate the periodic response consistency based on the changes in response amplitude and response phase of corresponding pixels within multiple periods. The noise analysis unit is used to extract the frequency components of the frequency band adjacent to the target response frequency and calculate the adjacent frequency noise characteristics. The scoring calculation unit is used to perform weighted calculations based on the response amplitude, adjacent channel noise characteristics, and periodic response consistency to obtain the response score.

14. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The response pixel determination module includes: The background statistics unit is used to obtain the response scores corresponding to the pixels in the background region and calculate the statistical parameters of the background score sequence. A threshold generation unit is used to generate a response judgment threshold based on the statistical parameters and a preset threshold coefficient. The response mask generation unit is used to filter response pixels according to the response determination threshold and generate a response pixel mask.

15. The fluorescence-based responsive vascular imaging system according to claim 9, characterized in that, The vascular structure fusion module includes: The vascular reference image generation unit is used to perform time averaging or maximum intensity projection on visible light or near-infrared fluorescence image sequences to obtain vascular structure reference images. The vascular enhancement unit is used to perform multi-scale vascular enhancement on the vascular structure reference image to obtain a vascular structure feature matrix. The blood vessel segmentation unit is used to segment the blood vessel structure feature matrix according to the structure segmentation threshold and generate a blood vessel structure mask through morphological processing. The fusion unit is used to perform intersection operation or weighted fusion of the response pixel mask and the vascular structure mask to generate a response vascular image and / or a response vascular mask.

Images