Multi-immunofluorescence detection method and equipment based on spatial reconstruction and medium

Abstract

The invention discloses a multiple immunofluorescence detection method and equipment based on spatial reconstruction and a medium, and relates to the technical field of histopathology analys.The multiple immunofluorescence detection method comprises the steps that a tissue sample is prepared, fluorescent reference microbeads are added, and continuous slices are obtained; circularly preprocessing the tissue sample, and performing intensity normalization and bleaching time drift correction by using fluorescent reference beads to obtain a corrected image sequence; performing spectral unmixing under the three-dimensional voxel coordinate system, and introducing spatial regular constraint to obtain a voxel-level fluorescence intensity vector; reconstructing a high-resolution voxel stack and segmenting a cell voxel set by combining a nuclear dye signal and a fluorescent reference microbead and adopting rigid and non-rigid registration and assisted by three-dimensional deconvolution; and calculating a proximity relation and a co-localization index of different phenotype cells, and generating a detection result. According to the invention, high-resolution tissue and cell reconstruction in a three-dimensional space is realized, the space continuity and accuracy are improved, and the stability and accuracy of voxel-level fluorescence signal quantification are improved.

Description

Technical Field

[0001] This invention relates to the field of histopathological analysis technology, and in particular to a method, device and medium for multiplex immunofluorescence detection based on spatial reconstruction. Background Technology

[0002] In histopathology and cell biology research, multiplex immunofluorescence assays, as important molecular markers and spatial distribution analysis tools, have been widely used in cell phenotype identification, molecular interaction studies, and tissue microenvironment analysis. Conventional methods typically involve fixing, sectioning, and immunofluorescence staining of tissue samples; labeling specific molecules or cells with different fluorescent markers; and then acquiring image data using an optical microscopy system. Subsequently, researchers use spectral separation or channel separation techniques to identify and quantify different fluorescence signals to obtain the molecular distribution at the tissue or cell level. These methods can present the distribution characteristics of cells and their markers in two-dimensional images and provide important data support for histopathological diagnosis, immune function research, and precision medicine. Due to their reproducibility and strong universality, this type of detection method has become one of the fundamental tools for related experimental research and clinical analysis.

[0003] However, conventional immunofluorescence detection methods based on two-dimensional images have certain limitations in terms of spatial information representation. On the one hand, they often rely on local reference signals during image stitching and cross-slice registration, which may lead to insufficient continuity of the overall spatial structure of the tissue, thus affecting the accurate determination of cell localization and proximity relationships at the three-dimensional scale. On the other hand, the spectral demixing and background correction of multi-channel fluorescence signals in two-dimensional detection methods are mainly performed on the plane, making it difficult to fully consider the signal continuity and spatial constraints at the voxel level, which may affect the stable extraction and quantitative accuracy of weak fluorescence signals in complex tissues. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a spatial reconstruction-based multiplex immunofluorescence detection method to solve the problems of insufficient continuity of three-dimensional spatial registration between tissue sections and limited quantitative accuracy of voxel-level fluorescence signals.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a multiplex immunofluorescence detection method based on spatial reconstruction, comprising: preparing a tissue sample and adding fluorescent reference microbeads to obtain serial sections; A baseline of autofluorescence was established by acquiring signals from blank slices, and monochromatic spectra of each fluorescent dye were calibrated to obtain a complete dataset of spectral mixing matrix and baseline parameters. The tissue samples were preprocessed cyclically, and the intensity was normalized and the bleaching time drift was corrected using the signal emitted by the fluorescent reference beads to obtain the corrected multi-channel image sequence. In a three-dimensional voxel coordinate system, spectral unmixing of multi-channel image sequences is performed, and the unmixed result vector is constrained by three-dimensional spatial regularization and continuous spatial constraints to obtain voxel-level fluorescence intensity vectors. Based on the components of the voxel-level fluorescence intensity vector and the fluorescence reference beads, rigid and non-rigid registration methods are used to perform unified spatial alignment on multi-channel images acquired in different cycles. Combined with three-dimensional deconvolution processing, high-resolution reconstructed voxel values ​​are obtained, and cell segmentation is performed in a unified coordinate system to obtain a set of cell voxels. The proximity relationships and co-localization indices of cells with different phenotypes in three-dimensional space were calculated for cytochrome sets to obtain multiplex immunofluorescence detection results.

[0007] As a preferred embodiment of the spatial reconstruction-based multiplex immunofluorescence detection method of the present invention, the specific steps of preparing tissue samples and adding fluorescent reference microbeads to obtain serial sections are as follows: The tissue to be tested was fixed and dehydrated to prepare a tissue sample, and fluorescent reference beads were added to the tissue sample. Tissue samples containing fluorescent reference beads were sectioned to obtain serial sections.

[0008] As a preferred embodiment of the spatial reconstruction-based multiplex immunofluorescence detection method of the present invention, the steps include: establishing an autofluorescence baseline by acquiring signals from blank slides, and performing monochromatic spectral calibration on each fluorescent dye to obtain a complete dataset of spectral mixing matrix and baseline parameters. Signals were acquired from blank slides, and fluorescence calculations were performed on the signals from the blank slides to obtain the autofluorescence baseline. Monochromatic excitation and acquisition were performed on each fluorescent dye individually, and spectral mixing calculations were performed using the monochromatic signals of each fluorescent dye to obtain the spectral mixing matrix; By combining the autofluorescence baseline with the spectral mixing matrix, a complete dataset of the spectral mixing matrix and baseline parameters is obtained.

[0009] As a preferred embodiment of the multiplex immunofluorescence detection method based on spatial reconstruction described in this invention, the blank section refers to a tissue section that has not been treated with any fluorescent dye.

[0010] As a preferred embodiment of the spatial reconstruction-based multiplex immunofluorescence detection method of the present invention, the steps include: sequentially performing cyclic preprocessing on tissue samples, and using the signal emitted by fluorescent reference microbeads for intensity normalization and bleaching time drift correction to obtain a corrected multi-channel image sequence. Tissue samples were subjected to cyclic pretreatment to obtain cyclic immunofluorescence acquisition sequences; The pretreatment includes immunofluorescence staining, imaging, and bleaching; The signal emitted by the fluorescent reference microbeads was collected, and the intensity normalization coefficient was calculated. The intensity normalization coefficient was used to perform numerical normalization on the cyclic immunofluorescence acquisition sequence channel by channel to obtain the intensity-normalized multi-channel image sequence. The bleaching time drift correction parameters were calculated using the signal emitted by the fluorescent reference microbeads. The intensity-normalized multi-channel image sequence is corrected by using drift correction parameters to obtain the corrected multi-channel image sequence.

[0011] As a preferred embodiment of the spatial reconstruction-based multiplex immunofluorescence detection method of the present invention, the steps include: performing spectral unmixing on the multi-channel image sequence in a three-dimensional voxel coordinate system, and constraining the unmixed result vector through three-dimensional spatial regularization constraints and continuity spatial constraints to obtain voxel-level fluorescence intensity vectors. The corrected multi-channel image sequence is mapped to a three-dimensional voxel coordinate system; Spectral unmixing calculations are performed using multi-channel images mapped to a three-dimensional voxel coordinate system and spectral mixing matrices to obtain an initial unmixing result vector. The initial unmixed result vector is constrained by three-dimensional space regularization constraints, and the unmixed result vector after constraint processing is constrained by continuous space constraints to obtain the unmixed result vector. The unmixed result vector is used to calculate the voxel-level intensity to obtain the voxel-level fluorescence intensity vector.

[0012] As a preferred embodiment of the spatial reconstruction-based multiplex immunofluorescence detection method of the present invention, the following steps are described: Based on the components of the voxel-level fluorescence intensity vector and the fluorescence reference beads, rigid and non-rigid registration methods are used to uniformly align the multi-channel images acquired in different cycles, and combined with three-dimensional deconvolution processing to obtain high-resolution reconstructed voxel values. Cell segmentation is then performed in a unified coordinate system to obtain a set of cell voxels. The specific steps are as follows: The nuclear dye signal and the signal emitted by the fluorescent reference microbeads were extracted from the calibrated multichannel image sequence and used as registration references; Rigid and non-rigid registration are performed on multi-channel images acquired in different cycles using the established registration reference to obtain unified spatial alignment data; Three-dimensional deconvolution processing is performed using uniform spatially aligned data to generate high-resolution reconstructed voxel values. High-resolution reconstructed voxel values ​​are used for cell segmentation in a unified coordinate system to extract cell voxel sets.

[0013] As a preferred embodiment of the multiplex immunofluorescence detection method based on spatial reconstruction described in this invention, the specific steps for calculating the proximity relationships and co-localization indices of cells with different phenotypes in three-dimensional space on a set of cytochromes to obtain multiplex immunofluorescence detection results are as follows: Phenotypic determination of cell voxel sets is performed by combining voxel-level fluorescence intensity vectors to obtain cell sets with different phenotypes; Proximity relationships are calculated using the positional relationships of cell assemblies with different phenotypes in three-dimensional space; Colocalization indices were calculated using the signal distribution of cell assemblies with different phenotypes in three-dimensional space. Multiplex immunofluorescence detection results are generated based on voxel-level fluorescence intensity vectors, proximity relationships, and colocalization indices. The results of the multiplex immunofluorescence assay include voxel-level fluorescence signals, cell-level quantitative parameters, and spatial microenvironment distribution characteristics.

[0014] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the spatial reconstruction-based multiplex immunofluorescence detection method as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the spatial reconstruction-based multiplex immunofluorescence detection method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by adding fluorescent reference microbeads to tissue samples and combining nuclear dye signals with rigid and non-rigid registration, high-resolution tissue and cell reconstruction in three-dimensional space is achieved, improving the continuity and accuracy of three-dimensional reconstruction; by performing spectral unmixing in three-dimensional voxel coordinates and optimizing through three-dimensional spatial regularization constraints, not only is the detection capability of weak fluorescence signals enhanced, but the stability and accuracy of voxel-level fluorescence signal quantification are also improved. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a multiplex immunofluorescence detection method based on spatial reconstruction.

[0019] Figure 2 A flowchart for establishing an autofluorescence baseline and monochromatic spectral calibration.

[0020] Figure 3 This is a flowchart for intensity normalization and bleaching time drift correction.

[0021] Figure 4 This is a flowchart for cell segmentation, spatial analysis, and result generation. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a multiplex immunofluorescence detection method based on spatial reconstruction, comprising the following steps: S1: Prepare tissue samples and add fluorescent reference beads to obtain serial sections.

[0026] The tissue to be tested was fixed and dehydrated to prepare a tissue sample, and fluorescent reference beads were added to the tissue sample.

[0027] Furthermore, the tissue to be tested is subjected to routine histological pretreatment procedures, which include fixation and dehydration. Specifically, the tissue to be tested is placed in a fixative, such as a solution containing 10% neutral buffered formaldehyde, to stabilize the proteins and cell structures within the tissue.

[0028] Furthermore, the fixed tissue to be tested was placed in ethanol solutions of different concentrations for gradient dehydration treatment to gradually remove water from the tissue and avoid structural deformation caused by the presence of water during the sectioning process.

[0029] Furthermore, after completing the fixation and dehydration process, tissue samples were prepared, and fluorescent reference microbeads with known concentrations and particle size distributions were added to the tissue samples. The fluorescent reference microbeads have stable and traceable fluorescence emission characteristics in different channels and can be used as signal intensity normalization and spatial anchors during fluorescence detection.

[0030] Fluorescent reference beads are uniformly distributed in the treatment area by mixing with or adding dropwise the tissue sample to be tested.

[0031] Tissue samples containing fluorescent reference beads were sectioned to obtain serial sections.

[0032] Furthermore, the tissue sample containing fluorescent reference microbeads was paraffin-embedded, and the tissue to be tested was sectioned using a paraffin microtome. The section thickness was controlled to ensure high resolution under optical imaging conditions, such as between 3 and 5 μm, thereby obtaining continuous sections of the tissue sample.

[0033] It should be noted that the continuity of the original tissue spatial structure should be maintained during the continuous slicing process to ensure that effective superposition and unification between different slices can be achieved during spatial reconstruction and three-dimensional registration.

[0034] S2: Collect signals from blank slices to establish an autofluorescence baseline, and perform monochromatic spectral calibration on each fluorescent dye to obtain a complete dataset of spectral mixing matrix and baseline parameters.

[0035] Signals were acquired from blank slides, and fluorescence calculations were performed on the signals from the blank slides to obtain the autofluorescence baseline.

[0036] Furthermore, signal acquisition was performed on the blank sections. Specifically, blank tissue sample sections that had not undergone any fluorescent dye treatment were selected, and fluorescence imagers were used to excite and acquire signals in each detection channel to obtain the multi-channel raw signals of the blank sections.

[0037] Furthermore, fluorescence calculation processing was performed on the original multi-channel signals of the blank slices to remove random noise components and obtain a stable background signal curve. The stable background signal curve was used as the autofluorescence baseline to reflect the endogenous fluorescence characteristics generated by the tissue under different channels.

[0038] Each fluorescent dye was excited and collected in monochromatic form, and the monochromatic signals of each fluorescent dye were used to perform spectral mixing calculations to obtain the spectral mixing matrix.

[0039] Furthermore, to obtain the true responses of different fluorescent dyes in each channel, monochromatic excitation and imaging were performed on each fluorescent dye individually. Specifically, under the same experimental conditions, control samples containing only a single fluorescent dye were prepared, and excitation and imaging were performed sequentially in each channel to obtain the monochromatic signal corresponding to the single fluorescent dye. The monochromatic signals were then processed and normalized according to the channel dimension to construct the response relationship of each fluorescent dye in different channels. Based on the monochromatic signals, the normalized average intensity of each fluorescent dye under monochromatic excitation in each detection channel was fitted into a linear response coefficient, and a dimension of [missing information] was generated sequentially with channels as rows and fluorescent dyes as columns. spectral mixing matrix .

[0040] By combining the autofluorescence baseline with the spectral mixing matrix, a complete dataset of the spectral mixing matrix and baseline parameters is obtained.

[0041] Furthermore, the complete dataset of spectral mixing matrices and baseline parameters includes baseline parameters and spectral response parameters.

[0042] S3: Perform cyclic preprocessing on the tissue samples sequentially, and use the signal emitted by the fluorescent reference beads to perform intensity normalization and bleaching time drift correction to obtain the corrected multi-channel image sequence.

[0043] Tissue samples were subjected to cyclic pretreatment to obtain cyclic immunofluorescence acquisition sequences.

[0044] Pretreatment includes immunofluorescence staining, imaging, and bleaching.

[0045] Furthermore, fluorescent antibodies targeting different targets are added sequentially to the tissue samples. After staining once, microscopic imaging is performed, and the bound fluorescent signals are then eliminated by photobleaching or chemical methods, thus providing conditions for the next cycle of fluorescent labeling. This process is repeated to obtain a cyclic immunofluorescence acquisition sequence containing the results of multiple rounds of staining.

[0046] Among them, the bound fluorescent signal refers to the signal generated by the fluorescent label that remains on the tissue sample after one immunofluorescence staining and imaging.

[0047] The signal emitted by the fluorescent reference microbeads was collected, and the intensity normalization coefficient was calculated.

[0048] Furthermore, the signals emitted by the fluorescent reference beads embedded in the tissue sample were collected individually, and the intensity normalization coefficient was calculated by using the ratio of the average intensity of the reference beads in each channel to the reference baseline intensity of the fluorescent reference beads.

[0049] It should be noted that the reference intensity of the fluorescent reference microbeads refers to the average signal intensity of the standard fluorescent reference microbeads in the corresponding channel.

[0050] The standard fluorescent reference beads are commercially available fluorescent beads with known concentration and particle size distribution, stable and traceable emission characteristics in each detection channel. They are prepared according to the instructions, mixed with tissue samples, and embedded in tissue samples to obtain the beads, which are used as reference benchmarks for fluorescent reference bead intensity and spatial anchors.

[0051] The intensity normalization coefficient was used to perform numerical normalization on each channel of the cyclic immunofluorescence acquisition sequence to obtain the intensity-normalized multi-channel image sequence.

[0052] The bleaching time drift correction parameters were calculated using the signal emitted by the fluorescent reference microbeads.

[0053] Furthermore, the bleaching time drift correction parameter was calculated by utilizing the signal attenuation law of the fluorescent reference microbeads in the time series; the bleaching time drift correction parameter was obtained by exponential regression fitting.

[0054] The intensity-normalized multi-channel image sequence is corrected by using drift correction parameters to obtain the corrected multi-channel image sequence.

[0055] Furthermore, the drift correction parameters are used to perform channel-by-channel correction on the intensity-normalized multi-channel image sequence.

[0056] The correction calculation formula is expressed as follows: ; in, Indicates in the channel time The image signal value after intensity normalization. Indicates in the channel time The image correction signal value below, Indicates channel The bleaching rate constant, Indicates time.

[0057] It should be noted that the bleaching rate constant is greater than or equal to 0.

[0058] S4: Perform spectral unmixing on the multi-channel image sequence in a three-dimensional voxel coordinate system, and constrain the unmixed result vector and the continuity space constraint through three-dimensional spatial regularization constraints to obtain the voxel-level fluorescence intensity vector.

[0059] The corrected multi-channel image sequence is mapped to a three-dimensional voxel coordinate system.

[0060] Furthermore, the multi-channel image sequence after bleaching time-drift correction is uniformly mapped to a three-dimensional voxel coordinate system. Specifically, two-dimensional images from different cycles and channels are spatially stacked and registered using slice indexing and imaging parameters to obtain a three-dimensional multi-channel data volume with spatial coordinate identifiers, thus ensuring that each pixel corresponds to a unique three-dimensional voxel coordinate. .

[0061] in, This is a three-dimensional voxel coordinate vector used to represent the position of a voxel in three-dimensional space. Represents the coordinate components of a voxel in the horizontal direction. Represents the coordinate components of a voxel in the vertical direction. This represents the coordinate components of a voxel in the depth direction.

[0062] Spectral unmixing is performed using a multi-channel image mapped to a three-dimensional voxel coordinate system and a spectral mixing matrix to obtain an initial unmixing result vector.

[0063] Furthermore, the spectral unmixing optimization formula is expressed as: ; in, Voxel representation The initial unmixed result vector, Voxel representation The multi-channel observed signal vector, i.e., the actual intensity value acquired by each channel. Represents the spectral mixing matrix. Let represent the intensity vector of the fluorescence component to be determined. Represents the spontaneous fluorescence baseline vector. Let represent the intensity vector of the fluorescence component to be determined.

[0064] It should be noted that the fluorescence component intensity vector to be determined in the spectral unmixing optimization formula corresponds to the intensity value of different fluorescent dyes at the current voxel, and is used as an optimization variable for solving, and satisfies the non-negativity constraint condition ≥ 0; the initial unmixing result vector of the voxel in the spectral unmixing optimization formula represents the optimal solution obtained by minimizing the objective function at the three-dimensional voxel coordinates, i.e., the initial unmixing result vector; in short, the fluorescence component intensity vector to be determined is the candidate variable in the optimization problem, while the initial unmixing result vector of the voxel is the fluorescence component intensity vector obtained by optimization calculation at a specific voxel position, and has practical numerical significance.

[0065] The initial unmixed result vector is constrained by three-dimensional spatial regularity constraints, and the unmixed result vector after constraint processing is constrained by continuous spatial constraints to obtain the unmixed result vector.

[0066] Furthermore, based on the spectral unmixing optimization formula, noise is suppressed and the stability of weak observation signals is improved by three-dimensional spatial regularization constraints. Specifically, a three-dimensional gradient regularization term is added to the spectral unmixing optimization formula to perform spatial constraint processing on the unmixing result vector.

[0067] The spectral unmixing optimization formula incorporating three-dimensional spatial regularization constraints is expressed as: ; in, Voxel representation The constraint unmixing result vector, Indicates the first The gradient of each fluorescence component in three-dimensional space. This represents the spatial regularity coefficient, with a value range of

[10] . -4 10 -1 ].

[0068] It should be noted that the multi-channel images of tissue sample slices were unmixed under different spatial regularization coefficients, and the range of values ​​for the spatial regularization coefficients was obtained by using the proportion of positive cells as a parameter.

[0069] It should be noted that the fluorescence component refers to the component in the voxel-level fluorescence intensity vector obtained after spectral unmixing, representing the intensity of a specific fluorescent dye at the voxel.

[0070] Furthermore, to ensure the continuity of the unmixed result vector among adjacent voxels, a neighborhood smoothing condition of spatial neighborhood difference is applied to the unmixed result vector after spatial constraint processing through a continuity space constraint to obtain the final unmixed result vector.

[0071] The unmixed result vector is used to calculate the voxel-level intensity to obtain the voxel-level fluorescence intensity vector.

[0072] Furthermore, the unmixing result vector obtained through spectral unmixing, three-dimensional spatial regularization constraints, and neighborhood smoothing constraints is directly used as the fluorescence intensity vector of the voxel.

[0073] It should be noted that the components of the final unmixed result vector refer to the fluorescence intensity of various fluorescent dyes at voxels.

[0074] S5: Based on the components of the voxel-level fluorescence intensity vector and the fluorescence reference beads, rigid and non-rigid registration methods are used to perform unified spatial alignment on multi-channel images acquired in different cycles. Combined with three-dimensional deconvolution processing, high-resolution reconstructed voxel values ​​are obtained, and cell segmentation is performed in a unified coordinate system to obtain a set of cell voxels.

[0075] The nuclear dye signal and the signal emitted by the fluorescent reference beads were extracted from the calibrated multichannel image sequence and used as registration references.

[0076] Furthermore, nuclear dye signals are extracted from the calibrated multichannel image sequence, and DAPI nuclear dye is usually selected as a reference. Since nuclear dye is uniformly distributed throughout the tissue sample and can stably identify the location of cell nuclei, it has strong consistency in spatial morphology. Therefore, nuclear dye is used as the registration benchmark for global structure to ensure the consistency of the overall geometric morphology of multichannel images from different cycles.

[0077] Furthermore, the signal emitted by the fluorescent reference beads is extracted from the calibrated multi-channel image sequence. Since the fluorescent reference beads exhibit a regularly distributed dot-like bright signal in the tissue sample, and maintain a fixed fluorescence intensity, temperature, and spatial position under different cycles and different channels, the signal emitted by the fluorescent reference beads is used as a local precise anchor point to correct translation, rotation, scaling, and local deformation during different acquisition cycles.

[0078] By using the established registration reference, rigid and non-rigid registration is performed on multi-channel images acquired in different cycles to obtain unified spatial alignment data.

[0079] Furthermore, rigid registration is performed based on the nuclear dye signal of the registration reference. By extracting the nuclear dye intensity distribution in each cyclic multi-channel image, a matching algorithm based on the mutual information maximization method is used to calculate the geometric transformation parameters between the cyclic multi-channel images. The geometric transformation parameters include translation vector, rotation angle and scaling ratio. The geometric transformation parameters are applied to each cyclic multi-channel image to achieve the consistency of the overall structure in the three-dimensional coordinate system.

[0080] Furthermore, a matching algorithm employing the mutual information maximization method is used to calculate the geometric transformation parameters between cyclic multi-channel images. Specifically, from the multi-channel image sequence after intensity normalization and bleaching drift correction, the nuclear dye channel image is extracted as a registration reference for the global structure. The coordinates of the fluorescent reference beads in each multi-channel image are extracted as local spatial anchor points from the multi-channel image sequence after intensity normalization and bleaching drift correction. Assuming the multi-channel images to be registered are the reference image and the image to be transformed, the mutual information objective of the mutual information maximization matching algorithm is defined as the optimal geometric transformation function, ensuring that the transformed multi-channel images achieve maximum mutual information in their gray-level statistical distribution. Initial geometric transformation parameters are defined, where the translation vector is the displacement along the X, Y, and Z axes, the rotation angle is the Euler angle rotation around the three axes, and the scaling ratio is the three-axis scaling ratio. Based on the defined initial geometric transformation parameters, mutual information is used as the objective function, and stochastic gradient descent is used for iterative parameter updates to obtain the optimal geometric transformation parameters. The optimal geometric transformation parameters are then applied to the image to be registered, performing a consistent geometric transformation on all channel images.

[0081] Furthermore, based on rigid registration, non-rigid registration is performed using the signals emitted by fluorescent reference beads in the registration reference. Specifically, by detecting the spatial coordinates of the signals emitted by fluorescent beads in each cycle of multi-channel images, an anchor point set is constructed. Using the coordinates of the fluorescent beads in the reference cycle as the target, the B-spline free deformation method is used to match the coordinates of the fluorescent beads in the image to be registered with the coordinates of the coordinates in the reference multi-channel image through nearest neighbor matching, ensuring that each anchor point generates a matching pair with the anchor points in the reference image that are close in position.

[0082] A regular three-dimensional grid consisting of control points is constructed within the three-dimensional space of the image to be registered. Each control point corresponds to a spatial position and has an adjustable three-dimensional displacement vector, which is initially set to zero.

[0083] Based on the constructed regular 3D mesh, a free deformation function is defined. The defined free deformation function is based on the B-spline interpolation principle, which determines the position offset of any control point by the displacement values ​​of 16 adjacent control points in the control point grid.

[0084] To find the optimal displacement value of the control points, an optimization objective function is defined, which consists of two parts: an anchor point matching error term and a regularization term. The anchor point matching error term refers to the sum of squared Euclidean distances between the anchor points of the fluorescent microbeads in the image to be registered after deformation and the corresponding anchor points in the reference multi-channel image, used to minimize the spatial deviation between the anchor points. The regularization term uses the sum of squared gradients of the control point displacements as a smoothness constraint to limit the variation in deformation intensity of the control points to suppress image distortion caused by excessive deformation. The overall objective of the optimization is to minimize the total error of the anchor point matching error term and the regularization term.

[0085] By iteratively optimizing the three-dimensional displacement vector of the control point, the optimal control point displacement vector is finally obtained.

[0086] The optimal control point displacement vector is applied to all channels of the image to be registered. Specifically, the B-spline deformation function is applied to the coordinates of each voxel in the multi-channel image sequence to obtain a new spatial position. Based on the new spatial position after transformation, the pixel value of the voxel is recalculated through trilinear interpolation. This process is repeated to generate image data for each channel in the transformed three-dimensional space, thus obtaining a unified spatially aligned dataset.

[0087] It should be noted that a unified spatially aligned dataset is obtained through rigid and non-rigid registration. This unified spatially aligned dataset achieves global consistency and local accuracy among different cyclic multi-channel images in a three-dimensional coordinate system.

[0088] Three-dimensional deconvolution is performed using uniformly spatially aligned data to generate high-resolution reconstructed voxel values.

[0089] Furthermore, three-dimensional deconvolution processing is performed using uniform spatially aligned data to eliminate the point spread effect caused by optical instruments during imaging. Specifically, based on the point spread function of the microscopic imager, the Richardson-Lucy iterative deconvolution algorithm is used to perform convolution processing on the uniform spatially aligned data. Through multiple iterations, the iteration stops when the rate of change between voxel intensities of two iterations is less than the set convergence threshold, and high-resolution reconstructed voxel values ​​are obtained.

[0090] The iterative update formula is expressed as: ; in, Indicates the first The reconstructed voxel values ​​of the next iteration Indicates the first The reconstructed voxel values ​​of the next iteration This represents the observed signal value of the uniformly spatially aligned data. Represents the point spread function. For convolution operations, This represents the flip function of the point spread function.

[0091] It should be noted that the convergence threshold setting for 3D deconvolution is as follows: convergence is determined when the rate of change of voxel intensity between two consecutive iterations is less than 1%; if convergence is not achieved after 40 iterations, the result of the 40th iteration is taken as the final result to avoid excessive noise caused by excessive iteration.

[0092] High-resolution reconstructed voxel values ​​are used for cell segmentation in a unified coordinate system to extract cell voxel sets.

[0093] Furthermore, based on the high-resolution reconstructed voxel values, cell segmentation is performed to extract cell voxel sets. Specifically, the nuclear dye signal channels in the multi-channel image sequence are extracted and smoothed using a three-dimensional Gaussian filter to suppress interference. The nuclear dye signal is then segmented using a three-dimensional threshold method based on the Otsu threshold method to obtain a preliminary mask of the cell nucleus. Three-dimensional morphological reconstruction of the nuclear connected domain is used to remove isolated noise points from the preliminary mask and fill in local missing regions to obtain a continuous and complete nuclear region. Finally, using the nuclear region as a seed point, a three-dimensional watershed algorithm is used to extend to the cytoplasm region to extract a complete set of cell voxels, and each set of cell voxels is assigned a unique identifier in a unified coordinate system.

[0094] S6: Calculate the proximity relationships and co-localization indices of cells with different phenotypes in three-dimensional space for cytochrome sets, and obtain multiplex immunofluorescence detection results.

[0095] Phenotypic determination of cell voxel sets is performed by combining voxel-level fluorescence intensity vectors to obtain cell sets with different phenotypes.

[0096] Furthermore, the cell voxel set is associated with the corresponding voxel-level fluorescence intensity vector. Specifically, for each cell voxel set, the voxel-level fluorescence intensity vector of all voxels in the cell voxel set is counted, and the cell-level fluorescence intensity feature vector is obtained by averaging.

[0097] Furthermore, a preset phenotypic determination threshold is used to determine the cell-level fluorescence intensity feature vector. The phenotypic determination threshold is determined by the fluorescence intensity of negative control cells and is used to distinguish the expression state of different targets. Specifically, each fluorescence component signal is normalized, and the normalized fluorescence component signal is compared with the corresponding phenotypic determination threshold. When the intensity of a fluorescence component signal exceeds the phenotypic determination threshold, the cell corresponding to the fluorescence component signal is determined to be positive on the marker. The positive or negative determination results of each cell under each fluorescence component are combined bit-by-bit into different phenotypic feature vectors. Cells with the same phenotypic feature vector are grouped into the same phenotype, thus obtaining a set of cells with different phenotypes.

[0098] In this context, "bit" in the bitwise combination refers to each fluorescence component corresponding to a Boolean value, i.e., 0 or 1. For example, if the determination result of Y cells under three fluorescence components, such as CD3, CD4, and CD8, is [1, 0, 1], then Y cells are combined into a phenotypic code, i.e., 101. All cells with the 101 phenotypic code are classified into the same phenotypic cell set.

[0099] The phenotypic determination formula is expressed as follows: ; in, Indicates the first The cells in the first The phenotypic determination result for each fluorescence component is either 0 or 1. Indicates the first The number of voxels contained in a cell. Indicates the first A collection of voxels in a single cell Voxel representation In fluorescence component The unmixing intensity below, Indicates the first Phenotypic threshold for each fluorescence component This indicates an indicator function, which takes the value 0 or 1.

[0100] It should be noted that the phenotypic determination threshold is set by statistically analyzing the fluorescence intensity distribution of negative control cells and setting the phenotypic determination threshold to be the average fluorescence intensity of negative control cells ± 2 standard deviations, so as to ensure that the false positive rate is controlled below 5%.

[0101] In another alternative implementation, the phenotypic determination threshold can be set based on the Youden index calculated from the ROC curve, and the optimal cutoff point can be selected as the phenotypic determination threshold, thereby balancing sensitivity and specificity.

[0102] Proximity relationships are calculated using the positional relationships of cell assemblies with different phenotypes in three-dimensional space.

[0103] Furthermore, the three-dimensional coordinates of the centroids of cells in each phenotypic cell set are extracted, and the Euclidean distance between the centroids of cells of different phenotypic types is used as a spatial metric; a proximity threshold is set based on cell diameter. For example, 6~20μm is the judgment condition. If the distance between a phenotype cell and another phenotype cell is less than or equal to the proximity threshold, it is judged as a proximity relationship. By traversing the centroid coordinates of all cells, the proximity relationship coefficient between different phenotype cells is calculated to obtain the proximity matrix that reflects the spatial contact characteristics of cells.

[0104] Colocalization indices were calculated using the signal distribution of cell assemblies with different phenotypes in three-dimensional space.

[0105] Furthermore, under a unified three-dimensional coordinate system, the overlapping region of the voxel-level fluorescence intensity vectors of two cell sets with different phenotypes is statistically analyzed, and the colocalization index is calculated using the normalized intersection method to obtain the colocalization index set.

[0106] Multiplex immunofluorescence detection results are generated based on voxel-level fluorescence intensity vectors, proximity relationships, and co-localization indices.

[0107] Furthermore, the number distribution of cells of each phenotype, the proximity coefficient matrix, and the set of colocalization indicators derived from the voxel-level fluorescence intensity vector are uniformly integrated to generate a detection report containing quantitative parameters and spatial structural features. Specifically, the number of cells in each phenotype cell set is counted, the proportion of cells in all cells is calculated, and a table is generated for output. Based on the three-dimensional centroid coordinates of different phenotype cell sets, a proximity matrix is ​​output to reflect the spatial contact probability of different phenotype cells under the proximity threshold. Colocalization indicators are output in the form of a histogram. Voxel-level fluorescence intensity distribution map and phenotype cell distribution map are generated in a unified three-dimensional coordinate system. Different phenotype cells are marked with pseudo-color, and quality control parameters such as unmixing residual, registration error, and normalization coefficient of fluorescence reference beads are output simultaneously.

[0108] This embodiment also provides a computer device applicable to the multiplex immunofluorescence detection method based on spatial reconstruction, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the multiplex immunofluorescence detection method based on spatial reconstruction as proposed in the above embodiment.

[0109] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0110] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the spatial reconstruction-based multiplex immunofluorescence detection method as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0111] In summary, this invention achieves high-resolution tissue and cell reconstruction in three-dimensional space by adding fluorescent reference beads to tissue samples and combining nuclear dye signals with rigid and non-rigid registration, thereby improving the continuity and accuracy of three-dimensional reconstruction. By performing spectral unmixing in a three-dimensional voxel coordinate system and optimizing through three-dimensional spatial regularization constraints, it not only enhances the detection capability of weak fluorescence signals but also improves the stability and accuracy of voxel-level fluorescence signal quantification.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multiplex immunofluorescence detection method based on spatial reconstruction, characterized in that: include, Tissue samples were prepared and fluorescent reference beads were added to obtain serial sections; A baseline of autofluorescence was established by acquiring signals from blank slices, and monochromatic spectra of each fluorescent dye were calibrated to obtain a complete dataset of spectral mixing matrix and baseline parameters. The tissue samples were preprocessed cyclically, and the intensity was normalized and the bleaching time drift was corrected using the signal emitted by the fluorescent reference beads to obtain the corrected multi-channel image sequence. In a three-dimensional voxel coordinate system, spectral unmixing of multi-channel image sequences is performed, and the unmixed result vector is constrained by three-dimensional spatial regularization and continuous spatial constraints to obtain voxel-level fluorescence intensity vectors. Based on the component nuclear dye signal of the voxel-level fluorescence intensity vector and the fluorescence reference microbeads, rigid and non-rigid registration methods are used to perform unified spatial alignment on multi-channel images acquired in different cycles. Combined with three-dimensional deconvolution processing, high-resolution reconstructed voxel values ​​are obtained, and cell segmentation is performed in a unified coordinate system to obtain a set of cell voxels. The proximity relationships and co-localization indices of cells with different phenotypes in three-dimensional space were calculated for cytochrome sets to obtain multiplex immunofluorescence detection results.

2. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 1, characterized in that: The specific steps for preparing tissue samples and adding fluorescent reference beads to obtain serial sections are as follows: The tissue to be tested was fixed and dehydrated to prepare a tissue sample, and fluorescent reference beads were added to the tissue sample. Tissue samples containing fluorescent reference beads were sectioned to obtain serial sections.

3. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 2, characterized in that: The process involves acquiring signals from blank slides to establish an autofluorescence baseline, performing monochromatic spectral calibration on each fluorescent dye, and obtaining a complete dataset of spectral mixing matrices and baseline parameters. The specific steps are as follows: Signals were acquired from blank slides, and fluorescence calculations were performed on the blank slide signals to obtain the autofluorescence baseline. Monochromatic excitation and acquisition were performed on each fluorescent dye individually, and spectral mixing calculations were performed using the monochromatic signals of each fluorescent dye to obtain the spectral mixing matrix; By combining the autofluorescence baseline with the spectral mixing matrix, a complete dataset of the spectral mixing matrix and baseline parameters is obtained.

4. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 3, characterized in that: The blank section refers to a tissue section that has not been treated with any fluorescent dye.

5. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 4, characterized in that: The tissue samples are sequentially preprocessed in cycles, and the signals emitted by fluorescent reference microbeads are used for intensity normalization and bleaching time drift correction to obtain a corrected multi-channel image sequence. The specific steps are as follows: Tissue samples were subjected to cyclic pretreatment to obtain cyclic immunofluorescence acquisition sequences; The pretreatment includes immunofluorescence staining, imaging, and bleaching; The signal emitted by the fluorescent reference microbeads was collected, and the intensity normalization coefficient was calculated. The intensity normalization coefficient was used to perform numerical normalization on the cyclic immunofluorescence acquisition sequence channel by channel to obtain the intensity-normalized multi-channel image sequence. The bleaching time drift correction parameters were calculated using the signal emitted by the fluorescent reference microbeads. The intensity-normalized multi-channel image sequence is corrected by using drift correction parameters to obtain the corrected multi-channel image sequence.

6. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 5, characterized in that: The process involves spectral demixing of a multi-channel image sequence in a three-dimensional voxel coordinate system, followed by constraint processing of the demixed result vector using three-dimensional spatial regularization and continuity constraints to obtain voxel-level fluorescence intensity vectors. The specific steps are as follows: The corrected multi-channel image sequence is mapped to a three-dimensional voxel coordinate system; Spectral unmixing calculations are performed using multi-channel images mapped to a three-dimensional voxel coordinate system and spectral mixing matrices to obtain an initial unmixing result vector. The initial unmixed result vector is constrained by three-dimensional space regularization constraints, and the unmixed result vector after constraint processing is constrained by continuous space constraints to obtain the unmixed result vector. The unmixed result vector is used to calculate the voxel-level intensity to obtain the voxel-level fluorescence intensity vector.

7. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 6, characterized in that: The process involves using the components of the voxel-level fluorescence intensity vector and fluorescence reference beads, employing rigid and non-rigid registration methods to perform unified spatial alignment on multi-channel images acquired in different cycles, combined with three-dimensional deconvolution processing to obtain high-resolution reconstructed voxel values, and then performing cell segmentation in a unified coordinate system to obtain a set of cell voxels. The specific steps are as follows: The nuclear dye signal and the signal emitted by the fluorescent reference microbeads were extracted from the calibrated multichannel image sequence and used as registration references; Rigid and non-rigid registration are performed on multi-channel images acquired in different cycles using the established registration reference to obtain unified spatial alignment data; Three-dimensional deconvolution processing is performed using uniform spatially aligned data to generate high-resolution reconstructed voxel values. High-resolution reconstructed voxel values ​​are used for cell segmentation in a unified coordinate system to extract cell voxel sets.

8. The multiplex immunofluorescence detection method based on spatial reconstruction as described in claim 7, characterized in that: The specific steps for calculating the proximity relationships and co-localization indices of cells with different phenotypes in three-dimensional space using cytochrome sets to obtain multiplex immunofluorescence detection results are as follows: Phenotypic determination of cell voxel sets is performed by combining voxel-level fluorescence intensity vectors to obtain cell sets with different phenotypes; Proximity relationships are calculated using the positional relationships of cell assemblies with different phenotypes in three-dimensional space; Colocalization indices were calculated using the signal distribution of cell assemblies with different phenotypes in three-dimensional space. Multiplex immunofluorescence detection results are generated based on voxel-level fluorescence intensity vectors, proximity relationships, and colocalization indices. The results of the multiplex immunofluorescence assay include voxel-level fluorescence signals, cell-level quantitative parameters, and spatial microenvironment distribution characteristics.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the spatial reconstruction-based multiplex immunofluorescence detection method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the spatial reconstruction-based multiplex immunofluorescence detection method according to any one of claims 1 to 8.

Images