Composite image-based tumor cell assisted positioning system and method

Abstract

The invention provides a tumor cell assisted positioning system and method based on a composite image, and relates to the technical field of medical image.The system comprises a data acquisition module used for synchronously acquiring radio frequency echo signals, ultrasonic images and computed tomography data of target biological tissue to generate an original fusion data set; the feature processing module is used for performing spatial registration and feature extraction on the original fusion data set to obtain a multi-dimensional feature set; and the identification imaging module is used for extracting a multi-band radio frequency echo signal intensity level value of the target area based on the multi-dimensional feature set, generating a tissue radio frequency characteristic diagram through logarithmic compression, and performing pixel-level fusion on the tissue radio frequency characteristic diagram and the synchronous ultrasonic image to generate a composite image. According to the method, the three-dimensional coordinates of the tumor boundary are determined through multi-modal data fusion and precise calibration in combination with anatomical structure constraints, and the accuracy of tumor cell positioning is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical imaging technology, in particular to a tumor cell auxiliary positioning system and method based on a composite image. BACKGROUND

[0002] In early breast cancer screening, traditional techniques for tumor cell positioning have some limitations, and the imaging clarity and tissue depth adaptability are insufficient. For example, when using ultrasound technology alone for breast cancer screening, the shape and boundary of superficial breast tumors can usually be clearly displayed. However, when the tumor is located in the deep layer of the breast (such as near the chest wall), the ultrasound wave will be attenuated when passing through the thick breast tissue or chest wall muscle, which may cause the tumor image to appear blurred and the edge to be unclear, increasing the difficulty of accurately positioning the tumor size and infiltration range.

[0003] In addition, the multi-dimensional information fusion capability is weak. For example, when relying solely on magnetic resonance imaging (MRI) for breast cancer tumor positioning, the anatomical structure details of breast soft tissue can be better presented, but the functional information such as blood perfusion of the tumor is not directly reflected. When encountering some early tumors with atypical blood supply characteristics, it may be difficult to accurately distinguish the boundary between the tumor and the surrounding benign proliferative tissue based solely on anatomical structure images, thereby affecting the accuracy of positioning. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a tumor cell auxiliary positioning system and method based on a composite image, which can accurately fuse multi-source image information and improve the accuracy and efficiency of tumor cell positioning.

[0005] To solve the above technical problems, the technical solutions of the present application are as follows: In a first aspect, a tumor cell auxiliary positioning system based on a composite image includes: A data acquisition module for synchronously acquiring radio frequency echo signals, ultrasound images and computed tomography data of a target biological tissue to generate an original fusion data set; A feature processing module for spatial registration and feature extraction of the original fusion data set to obtain a multi-dimensional feature set; An identification imaging module for extracting multi-band radio frequency echo signal intensity level values of a target region based on the multi-dimensional feature set, generating a tissue radio frequency characteristic map through logarithmic compression, and performing pixel-level fusion with a synchronous ultrasound image to generate a composite image; A positioning analysis module for subcellular distribution analysis of the composite image to obtain a set of nuclear positioning positive cell coordinates; The positioning calibration module is configured to select a set of preset biomarkers from the set of coordinates of the nucleus positioning positive cells, construct a spatial topological relationship, quantify the microenvironment heterogeneity of the spatial topological relationship, and generate a positioning calibration parameter based on a regional heterogeneity distribution feature; The positioning output module is configured to determine a three-dimensional coordinate of a tumor boundary based on the positioning calibration parameter, through a spatial gradient change analysis and a radio frequency signal feature attenuation association rule, to obtain a tumor cell positioning result.

[0006] Further, spatial registration and feature extraction are performed on the original fusion data set to obtain a multi-dimensional feature set, including: Based on the original fusion data set, the radio frequency echo signal, the ultrasound image, and the computed tomography data are sequentially subjected to rigid transformation and affine transformation to achieve spatial accurate alignment, and a spatially registered fusion data set is generated; The texture features of the ultrasound image and the computed tomography density distribution features are extracted from the spatially registered fusion data set, and the frequency energy features of the radio frequency echo signal are calculated through fast Fourier transform; The ultrasound image texture features, the computed tomography density distribution features, and the frequency energy features are integrated to obtain a multi-dimensional feature set.

[0007] Further, based on the multi-dimensional feature set, the multi-band radio frequency echo signal intensity level values of the target region are extracted, logarithmic compression is performed to generate a tissue radio frequency characteristic map, and pixel-level fusion is performed with a synchronous ultrasound image to generate a composite image, including: The multi-band radio frequency echo signal intensity level values corresponding to the target region are extracted and separated from the multi-dimensional feature set; The multi-band radio frequency echo signal intensity level values of the target region are subjected to dynamic range logarithmic compression processing to generate a radio frequency signal intensity distribution matrix representing the radio frequency signal intensity distribution of the target region; The radio frequency signal intensity distribution matrix is mapped to a two-dimensional plane corresponding to the spatial position of the ultrasound image to generate a tissue radio frequency characteristic map reflecting the radio frequency characteristics of the target region tissue; The tissue radio frequency characteristic map and the ultrasound image of the corresponding target region synchronously collected are combined according to a predetermined rule to achieve pixel-level superposition fusion to generate a composite image.

[0008] Further, subcellular distribution analysis is performed on the composite image to obtain a set of coordinates of nucleus positioning positive cells, including: The intensity correlation of the cell nucleus marker signal and the target protein marker signal in the composite image is calculated pixel by pixel to obtain a three-dimensional data set representing the co-localization degree of each three-dimensional spatial position point; Based on the three-dimensional data set representing the degree of colocalization, by setting a determination threshold for nuclear localization positivity, all three-dimensional spatial regions in the three-dimensional data set with values greater than the determination threshold are identified as candidate nuclear localization positive cell regions; According to the candidate nuclear localization positive cell regions, three-dimensional spatial connectivity analysis is applied to segment the spatial regions that are connected to each other into independent single candidate cell three-dimensional structures; Based on the independent single candidate cell three-dimensional structures, the centroid coordinates of the cells in the three-dimensional space are obtained by calculating the geometric center; The centroid coordinates of all independent candidate cells in the three-dimensional space are integrated to obtain a set of nuclear localization positive cell coordinates.

[0009] Further, a group of pre-set biomarkers is selected from the set of nuclear localization positive cell coordinates, a spatial topological relationship is constructed, the spatial topological relationship is quantified for microenvironment heterogeneity, and localization calibration parameters are generated based on the regional heterogeneity distribution characteristics, including: Based on the set of nuclear localization positive cell coordinates, a three-dimensional coordinate set of a target biomarker cell subset is generated by screening cell coordinates with pre-set biomarker identifiers; According to the three-dimensional coordinate set of the target biomarker cell subset, a three-dimensional spatial adjacency network is constructed with cell coordinates as nodes and spatial adjacency relationships as edges by calculating the three-dimensional Euclidean distance between each coordinate point and applying an adjacency distance threshold to determine connectivity; Based on the three-dimensional spatial adjacency network, a comprehensive heterogeneity quantification data set is generated by calculating the local cell density gradient value set of all nodes in the network, the node cluster coefficient value set, and the network global spatial entropy value; According to the comprehensive heterogeneity quantification data set, the heterogeneity spatial distribution feature description within the metabolic target area is extracted by mapping to the three-dimensional spatial grid of the metabolic target area and analyzing the numerical distribution within the grid elements; Based on the heterogeneity spatial distribution feature description, a boundary correction vector parameter set is derived by establishing a spatial mapping relationship between the feature and the original boundary of the metabolic target area, and a localization calibration parameter is generated.

[0010] Further, based on the localization calibration parameters, the three-dimensional coordinates of the tumor boundary are determined by spatial gradient change analysis and radio frequency signal feature attenuation association rules to obtain tumor cell localization results, including: Based on the localization calibration parameters and the lesion classification results, the spatial mapping information of the localization calibration parameters is superimposed on the three-dimensional spatial coordinate framework of the original fusion data set to generate the initial spatial position data of the calibrated candidate boundary; Based on the calibrated candidate boundary initial spatial position data, the radio frequency signal attenuation characteristic value of the region near the candidate position is extracted; meanwhile, the image space gradient intensity value of the region near the same position in the ultrasonic image component data is calculated, and matching verification is performed based on a preset spatial correlation rule to generate a candidate boundary spatial position data set; Based on the candidate boundary spatial position data set, the spatial continuity of the boundary points is adjusted by applying a dynamic programming algorithm to generate a boundary point set conforming to the anatomical structure constraint; Based on the boundary point set conforming to the anatomical structure constraint, three-dimensional spatial coordinate information is extracted to obtain a tumor boundary three-dimensional coordinate point set as a tumor cell positioning result.

[0011] Further, based on the candidate boundary spatial position data set, the spatial continuity of the boundary points is adjusted by applying a dynamic programming algorithm to generate a boundary point set conforming to the anatomical structure constraint, including: The tumor boundary three-dimensional coordinate set is input as the candidate boundary spatial position data set to a framework based on a dynamic programming algorithm; For the candidate boundary spatial position data set, a path evaluation rule for measuring the smoothness of the connection path between any two candidate boundary points is generated according to the target organ anatomical structure constraint; Based on the path evaluation rule, the global smoothness final boundary point spatial connection sequence is found in the candidate boundary spatial position data set by a dynamic programming algorithm calculation; Based on the boundary point spatial connection sequence, a final tumor boundary point set conforming to the anatomical structure smoothness constraint is generated.

[0012] In a second aspect, a tumor cell auxiliary positioning method based on a composite image includes: Step 1, synchronously acquiring radio frequency echo signals, ultrasonic images and computed tomography data of a target biological tissue to generate an original fusion data set; Step 2, performing spatial registration and feature extraction on the original fusion data set to obtain a multi-dimensional feature set; Step 3, extracting multi-band radio frequency echo signal intensity level values of a target region based on the multi-dimensional feature set, generating a tissue radio frequency characteristic map through logarithmic compression, and performing pixel-level fusion with a synchronous ultrasonic image to generate a composite image; Step 4, performing subcellular distribution analysis on the composite image to obtain a nucleus positioning positive cell coordinate set; Step 5, selecting a group of preset biomarkers in the nucleus positioning positive cell coordinate set, constructing a spatial topological relationship, quantifying the microenvironment heterogeneity of the spatial topological relationship, and generating positioning calibration parameters based on the regional heterogeneity distribution characteristics; Step 6, based on the positioning calibration parameter, determine the three-dimensional coordinates of the tumor boundary by the spatial gradient change analysis and the radio frequency signal feature attenuation associated rule to obtain the tumor cell positioning result.

[0013] In a third aspect, a computing device includes: one or more processors; a memory device storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the system.

[0014] In a fourth aspect, a computer readable storage medium stores a program, when the program is executed by a processor, the system is implemented.

[0015] The above scheme of the present application at least includes the following beneficial effects: In data fusion and utilization, the radio frequency echo signal, ultrasonic image and computed tomography data are synchronously collected, the multi-modal data are accurately aligned through spatial registration, multi-dimensional features are extracted and integrated, comprehensive and rich information basis is provided for tumor positioning, positioning deviation caused by one-sided data is reduced, in positioning accuracy and depth, a progressive positioning path from macro to micro is constructed, nuclear positioning positive cell coordinates are obtained through subcellular distribution analysis, positioning accuracy is improved to cell level, and positioning accuracy is improved.

[0016] In response to complex environment and calibration capability, by selecting a preset biomarker, constructing spatial topological relationship and quantifying microenvironment heterogeneity, generating positioning calibration parameters, fully considering the influence of tumor microenvironment complexity on positioning, effectively correcting the deviation in the positioning process, making the positioning result more consistent with the actual distribution of tumor cells, enhancing the reliability of positioning in complex tumor microenvironment, in clinical application value, based on the positioning calibration parameter, the spatial gradient change analysis and the radio frequency signal feature attenuation associated rule are determined to determine the three-dimensional coordinates of the tumor boundary, and the boundary is ensured to meet the anatomical structure constraint through the dynamic programming algorithm, which provides a reliable basis for accurate diagnosis, surgical plan formulation, efficacy evaluation and the like of the tumor, and reduces the damage to normal tissues during surgery. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a schematic diagram of a tumor cell auxiliary positioning system based on a composite image provided by an embodiment of the present application.

[0018] Figure 2 is a flowchart of a tumor cell auxiliary positioning method based on a composite image provided by an embodiment of the present application. DETAILED DESCRIPTION

[0019] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.

[0020] As Figure 1 shown, the embodiments of the present application propose a tumor cell auxiliary positioning system based on a composite image, comprising: a data acquisition module for synchronously acquiring radio frequency echo signals, ultrasound images and computed tomography data of a target biological tissue, and generating an original fusion data set; a feature processing module for spatial registration and feature extraction of the original fusion data set to obtain a multi-dimensional feature set; an identification imaging module for extracting multi-band radio frequency echo signal intensity level values of a target region based on the multi-dimensional feature set, generating a tissue radio frequency characteristic map through logarithmic compression, and performing pixel-level fusion with a synchronous ultrasound image to generate a composite image; a positioning analysis module for subcellular distribution analysis of the composite image to obtain a set of nuclear positioning positive cell coordinates; a positioning calibration module for selecting a group of preset biomarkers from the set of nuclear positioning positive cell coordinates, constructing a spatial topological relationship, quantifying the microenvironment heterogeneity of the spatial topological relationship, and generating positioning calibration parameters based on the regional heterogeneity distribution characteristics; a positioning output module for determining three-dimensional coordinates of a tumor boundary based on the positioning calibration parameters through spatial gradient change analysis and radio frequency signal feature attenuation correlation rules to obtain a tumor cell positioning result.

[0021] In the embodiment of the present application, the multiple data of the target biological tissue are synchronously collected and the original fusion data set is generated, which can make the obtained information more comprehensive, cover different dimensional biological tissue characteristics, avoid the information limitation brought by a single data type, and perform spatial registration and feature extraction on the original fusion data set, so that the data from different sources can be accurately corresponded in spatial position, and the key multi-dimensional features are extracted, so that the analysis is more focused on valuable information, the effectiveness and pertinence of data processing are improved, the high metabolic target area is located based on the multi-dimensional feature set, the area with high lesion activity can be accurately locked, the spatial structure and distribution of the target area are clearly presented in the three-dimensional image, which provides intuitive and specific basis for in-depth understanding of the lesion characteristics, the nuclear localization positive cell coordinate set is obtained by performing subcellular distribution analysis on the composite image, the analysis accuracy can be further improved to the cell level, the cell position with specific characteristics can be accurately captured, the positioning is more in line with the actual state of the biological tissue, the spatial topological relationship is constructed by selecting the preset biomarker in the nuclear localization positive cell coordinate set and generating the positioning calibration parameter, the differences in the microenvironment of the biological tissue can be fully considered, the positioning deviation caused by the heterogeneity of the microenvironment is reduced through quantitative analysis and parameter calibration, and the reliability of the positioning result is improved, the three-dimensional coordinates of the tumor boundary are determined based on the positioning calibration parameter, the spatial gradient change and the collaborative application of the radio frequency signal feature attenuation and other information are realized, the tumor boundary can be more accurately outlined, and accurate tumor cell positioning results are obtained, which provides reliable spatial position information for tumor diagnosis, treatment and the like.

[0022] In a preferred embodiment of the present application, the spatial registration and feature extraction are performed on the original fusion data set to obtain a multi-dimensional feature set, which can include: Based on the original fusion data set, the radio frequency echo signal, the ultrasound image and the computed tomography data are sequentially subjected to rigid transformation and affine transformation to realize spatial accurate alignment, and a spatially registered fusion data set is generated; The texture features of the ultrasound image and the computed tomography density distribution features are extracted from the spatially registered fusion data set, and the frequency energy features of the radio frequency echo signal are calculated through fast Fourier transform; The ultrasound image texture features, the computed tomography density distribution features and the frequency energy features are integrated to obtain a multi-dimensional feature set.

[0023] In the embodiment of the present application, the spatial coordinate system of the radio frequency echo signal, the ultrasound image and the computed tomography data in the original fusion data set is extracted, stable anatomical structures (such as the edge of the bone, the bifurcation point of the blood vessel) in the biological tissue are selected as the reference marks, the three-dimensional coordinates of the marks in the three kinds of data are recorded respectively, the deviation of the coordinates of the reference marks in each data from the preset standard coordinates is calculated, the coordinate origin is adjusted through the translation operation, so that the position deviation of the reference marks in the x, y and z axis directions is reduced to the preset range; the coordinate axis direction is adjusted through the rotation operation, so that the spatial orientation of the reference marks is kept consistent, and the process is repeated until the coordinate error of all the reference marks in the three kinds of data is lower than the set threshold, and the rigid transformation is completed.

[0024] On the basis of the rigid transformation, more subtle anatomical feature points (such as the texture boundary of the tissue and the density mutation point) are added as the registration marks, the actual coordinates of the marks in the three kinds of data are measured respectively, the coordinate correspondence table is established, the scaling coefficient that can make the marks coincide in different data to the highest degree is calculated according to the correspondence, the data proportion is adjusted in the x, y and z axis directions according to the coefficient, so that the size of the marks is matched; at the same time, the shearing parameter is calculated, the inclination angle of the data in the plane and the space is adjusted, the deformation caused by the difference of the imaging angle is eliminated, the scaling and shearing parameters are optimized through multiple iterations, so that the average coordinate deviation of all the marks is minimized, and the fusion data set after spatial registration is generated.

[0025] In the ultrasound image after spatial registration, a plurality of non-overlapping square regions are divided, the specific value of the gray value of all the pixels in each region is counted, the sum of all the gray values is calculated and divided by the number of pixels to obtain the average gray value; the square of the difference between each gray value and the average value is calculated, the sum is divided by the number of pixels to obtain the gray variance; the sum of the products of the gray values of adjacent pixels (horizontal, vertical and diagonal directions) in the region is calculated to obtain the correlation index; the sum of the absolute values of the difference between the gray values of adjacent pixels is calculated to obtain the contrast index; the number of times of occurrence of different gray values in the region is counted, the proportion of the number of times of occurrence of each gray value to the total number of pixels is calculated, the sum of the squares of the proportions is calculated to obtain the energy index, and the sum of the negative values of the product of the proportion and the natural logarithm is calculated to obtain the entropy index; the indexes of all the regions are summarized to form the texture features of the ultrasound image.

[0026] In the spatially registered computed tomography data, the three-dimensional range of the target tissue is demarcated, the CT values of the voxels in the range are extracted one by one, the sum of all CT values is counted and divided by the number of voxels to obtain the average density value, the maximum and minimum values in the CT values are screened to determine the density variation interval; the square of the difference between each CT value and the average density value is calculated, and the sum is divided by the number of voxels to obtain the density standard deviation, the density interval is divided into several continuous subintervals, the number of voxels in each subinterval is counted, the proportion of the number of voxels in each subinterval to the total number of voxels is calculated, a density distribution histogram is formed, the CT value (mode) corresponding to the peak value in the histogram is recorded, the skewness and kurtosis of the histogram are recorded, and the computed tomography density distribution characteristics are obtained by comprehensively analyzing these data.

[0027] The radio frequency echo signals of a continuous time period are intercepted from the spatially registered fusion data set, the signal amplitude values at each time point are recorded, the amplitude values are arranged in time sequence to form a time domain signal sequence, when the fast Fourier transform is applied, the time domain signal sequence is first divided into equal-length subsequences, each subsequence is subjected to a transform operation to obtain the corresponding frequency components and the amplitudes of each frequency, the energy (amplitude square) of each frequency is calculated according to the amplitude, the total energy of all frequencies in the preset low frequency band, medium frequency band and high frequency band is counted to obtain the total energy of each frequency band; the proportion of the energy of each frequency band to the total energy is calculated, the frequency point with the highest energy (peak frequency) and the frequency range (bandwidth) from the peak frequency to the frequency at which the energy drops to half are recorded, and these data are summarized to form the frequency energy characteristics of the radio frequency echo signals.

[0028] All specific values of the ultrasound image texture features, the computed tomography density distribution features and the radio frequency echo signal frequency energy characteristics are collected, for each feature, the maximum and minimum values are found, the difference between each feature value and the minimum value is calculated, and then divided by the difference between the maximum value and the minimum value, the feature values are standardized to the interval of 0-1, the standardized feature values are arranged in order according to the sequence of “texture feature-density feature-frequency domain feature” to form a feature matrix containing multiple types of indexes, whether there is missing value in the matrix is checked, the average value of the same type of feature is used to fill in the missing data, and finally a complete multi-dimensional feature set is obtained.

[0029] The step-by-step operation of rigid transformation and affine transformation can realize high-precision matching of radio frequency echo signals, ultrasonic images and computed tomography data in spatial position, so that the same tissue area described by different types of data can be accurately corresponded, the texture features of the ultrasonic image can be extracted, and the uniformity of the internal gray scale distribution of the tissue and the correlation between pixels can be captured in detail. These features can effectively reflect the structural differences of the tissue, such as the difference in texture complexity between tumor tissue and normal tissue, and provide rich structural basis for distinguishing the lesion area. The extraction of the computed tomography density distribution features can quantitatively reflect the density properties and distribution rules of the tissue, and the density of the tissue in different pathological states is different. These features can be used as an important basis for identifying abnormal tissues and improving the accuracy of judging the properties of the tissue. The frequency domain energy features of the radio frequency echo signals obtained by fast Fourier transform can convert the frequency characteristics that are difficult to directly observe in the time domain signals into quantifiable indexes, reflect the differences in the reflection characteristics of the tissue to radio frequency signals, and provide a new dimension for distinguishing different types of tissues and enrich the diversity of features. The integration of the multi-dimensional feature set combines the advantages of different imaging methods, avoids the limitations of single features, and makes the feature set more comprehensively reflect the physiological and pathological characteristics of biological tissues, thereby improving the reliability of the overall analysis.

[0030] In a preferred embodiment of the present application, the multi-band radio frequency echo signal intensity level value of the target area is extracted based on the multi-dimensional feature set, a tissue radio frequency characteristic map is generated by logarithmic compression, and pixel-level fusion is performed with a synchronous ultrasonic image to generate a composite image, which can include: The radio frequency echo signal intensity level values of the target area corresponding to multiple frequency bands are extracted and separated from the multi-dimensional feature set; The multi-band radio frequency echo signal intensity level values of the target area are subjected to dynamic range logarithmic compression processing to generate a radio frequency signal intensity distribution matrix representing the radio frequency signal intensity distribution of the target area; The radio frequency signal intensity distribution matrix is mapped to a two-dimensional plane corresponding to the spatial position of the ultrasonic image to generate a tissue radio frequency characteristic map reflecting the radio frequency characteristics of the tissue of the target area; The tissue radio frequency characteristic map and the ultrasonic image of the corresponding target area collected synchronously are combined according to a predetermined rule to realize pixel-level superposition fusion and generate a composite image.

[0031] In the embodiment of the present application, the data source range of the multi-dimensional feature set containing all the radio frequency echo signal related data of the target region, such as the specific tissue region of the human body and the surrounding region, including signal acquisition time, signal propagation path, original signal amplitude and other features, is determined, then the feature data corresponding to the target region is screened, the spatial coordinate range of the target region is set, such as the X-Y-Z axis coordinate interval determined by the previous positioning, the spatial position information of each signal data is checked one by one from the multi-dimensional feature set, all signal data with spatial coordinates falling within the target region range is retained, and the surrounding signal data with coordinates exceeding the target region is eliminated.

[0032] The multiple frequency band ranges to be separated, such as the preset low frequency band, medium frequency band and high frequency band, are determined, each frequency band corresponds to a fixed frequency interval, such as 1-3MHz for the low frequency band, 3-5MHz for the medium frequency band and 5-8MHz for the high frequency band, the frequency screening threshold is established for each frequency band, the frequency of the screened target region signal data is detected, the actual frequency value of each signal data is calculated one by one, the frequency is calculated by detecting the signal period in the way of "1 / period", the calculated actual frequency value is compared with the frequency threshold of each frequency band, if the actual frequency of a signal falls within the low frequency band threshold range, it is classified as a low frequency band radio frequency echo signal, and similarly, the medium frequency band and high frequency band signals are classified respectively, the separation of signals of different frequency bands is completed, the intensity level value of all signals in each separated frequency band is calculated, the original amplitude data of the signal in each frequency band is measured, then the preset level value conversion standard is used, such as comparing the amplitude data with the reference voltage value according to the logic of "level value=20xlg (signal amplitude / reference amplitude)", the amplitude data is converted into the corresponding intensity level value by calculating the logarithm and multiplying by 20 one by one, ensuring that each signal in each frequency band has a unique intensity level value.

[0033] Determine the dynamic range of the target area multi-band radio frequency echo signal strength level value, collect all the calculated strength level values in all frequency bands, compare the size of each level value one by one, record the maximum and minimum values, and the difference between the two is the dynamic range of the current signal strength. Set the target dynamic range of the logarithmic compression, such as according to the subsequent image display requirements, the preset target dynamic range is 0-60dB, calculate the compression ratio, divide the current dynamic range by the target dynamic range, get the compression coefficient, such as the current dynamic range is 0-120dB, the target is 0-60dB, then the compression coefficient is 2; Perform point-by-point logarithmic compression processing on each strength level value of each frequency band. First, take each original strength level value, such as a signal level value of 80dB, calculate the difference between it and the minimum value of the original dynamic range, then divide the difference by the compression coefficient, and finally add the minimum value of the target dynamic range to get the compressed level value; Repeat the operation to complete the logarithmic compression of all signal strength level values, and ensure that all level values after compression fall within the target dynamic range.

[0034] Generate a radio frequency signal strength distribution matrix. First, determine the number of rows and columns of the matrix, which should be consistent with the number of pixel rows and columns of the subsequent ultrasound image. For example, if the ultrasound image is 512x512 pixels, the matrix is set to 512 rows and 512 columns. Then fill the compressed signal strength level values of each frequency band into the matrix according to the spatial position correspondence principle. According to the original spatial coordinates X-Y axis position of each signal, find the corresponding row number in the matrix, such as the X coordinate corresponding to the row number and the column number, and the Y coordinate corresponding to the column number. Fill the compressed level value of the signal into the corresponding matrix cell. If there is no direct signal data in a certain matrix cell, take the average of the signal level values of its four adjacent positions, add the four adjacent values, and then divide by 4 to get the level value of the cell. Finally, form a complete two-dimensional matrix representing the radio frequency signal strength distribution of the target area.

[0035] The spatial parameters of the synchronous ultrasound image are acquired, including the pixel size of the ultrasound image, such as 0.1 mm x 0.1 mm of the actual space corresponding to each pixel, the spatial coordinate system of the image, such as the upper left corner as the origin, the right as the positive direction of the X axis, and the downward as the positive direction of the Y axis, the correspondence between the matrix of the radio frequency signal intensity distribution and the spatial position of the ultrasound image is established, the row number and column number of the matrix are associated with the Y axis and X axis coordinates of the ultrasound image respectively, such as the first row and first column cell of the matrix corresponding to the pixel position of the upper left corner (X=0, Y=0) of the ultrasound image, the i-th row and j-th column cell of the matrix corresponding to the pixel position of the ultrasound image (X=j x 0.1 mm, Y=i x 0.1 mm), ensuring that the spatial position of each cell of the matrix matches the spatial position of the ultrasound image pixel, setting the mapping rule of the level value and the image gray / color, such as mapping the level value of 0-20 dB after compression to black, 20-40 dB to gray, and 40-60 dB to white, or corresponding to different level values according to the rainbow color scale), generating the tissue radio frequency characteristic map point by point, according to the level value of each cell in the radio frequency signal intensity distribution matrix, the gray value or color value corresponding to the cell is determined by comparing the mapping rule, such as the level value of 30 dB corresponding to the medium gray value 128, and the gray / color value is assigned to the pixel point of the corresponding spatial position of the ultrasound image, all cells are processed one by one in the order of matrix rows and columns, and finally the tissue radio frequency characteristic map completely consistent with the spatial size of the ultrasound image is formed.

[0036] The synchronous acquisition of the target region ultrasound image is acquired, the pixel format of the ultrasound image is determined, such as RGB format or gray scale format and the numerical range of each pixel, such as 0-255 of the gray scale image pixel value, the predetermined rule of pixel-level fusion is set, such as the weighted superposition rule, the radio frequency characteristic map weight is preset as 0.4, and the ultrasound image weight is 0.6, the pixel position alignment verification of the two images is carried out, the pixel number of the tissue radio frequency characteristic map and the ultrasound image is checked one by one, such as 512 x 512 pixels, and 10 feature points are randomly selected, such as the obvious contour points of the edge of the target region, the pixel coordinates of the feature points in the radio frequency characteristic map and the pixel coordinates of the corresponding feature points in the ultrasound image are checked whether they are completely consistent, if there is deviation, such as (100, 100) of a feature point in the radio frequency map and (101, 100) in the ultrasound image, the radio frequency characteristic image pixel is translated, and the radio frequency map is translated left by 1 pixel as a whole, to ensure that all feature point coordinates are aligned.

[0037] For each pixel position in the image, the pixel value of the tissue radio frequency characteristic map at the position is read first, and then the pixel value of the ultrasound image at the position is read, and the fused pixel value is calculated according to the weighting rule; if it is in RGB format, the fused value is calculated according to the same weight rule for the pixel values of the R, G and B channels respectively, the fused pixel values of all pixel positions are arranged in the row and column order of the original image to form a complete composite image, and it is ensured that each pixel contains radio frequency characteristic information and ultrasound image information at the same time.

[0038] Through the process of first screening the target area signal, then separating the frequency band according to the frequency, and finally calculating the level value, it is ensured that the extracted signal only comes from the target area, and the influence of the surrounding noise is avoided; at the same time, the signal is separated according to the frequency band, and the level value is calculated separately, so that the unique information of the radio frequency signal in different frequency bands can be reserved, the excessive intensity difference in the original signal is compressed to a range suitable for image display, and the loss of details caused by the over-bright strong signal and the over-dark weak signal in the original signal is avoided; at the same time, the logarithmic compression can reserve the relative difference of the weak signal, ensure that the subtle radio frequency signal changes in the target area are not covered, and improve the presentation ability of the image to the subtle features of the tissue; through the process of matching the spatial parameters of the ultrasound image, establishing the matrix and pixel correspondence, it can be ensured that the spatial size and coordinate of the tissue radio frequency characteristic map and the ultrasound image are completely consistent, and the accurate correspondence of the two kinds of image information in space is ensured; at the same time, the way of mapping the level value to the gray scale / color can convert the abstract radio frequency signal data into intuitive image information, and reduce the difficulty of signal data interpretation.

[0039] In a preferred embodiment of the present application, the subcellular distribution analysis is performed on the composite image to obtain a set of coordinates of the nucleus localization positive cells, which can include: For the nucleus marker signal and the target protein marker signal in the composite image, the intensity correlation of the two signals is calculated pixel by pixel to obtain a three-dimensional data set representing the co-localization degree of each three-dimensional spatial position point; Based on the three-dimensional data set representing the co-localization degree, by setting a determination threshold of the nucleus localization positive, all three-dimensional spatial regions in the three-dimensional data set with a value greater than the determination threshold are identified as candidate nucleus localization positive cell regions; According to the candidate nucleus localization positive cell regions, three-dimensional spatial connectivity analysis is applied to divide the mutually connected spatial regions into independent single candidate cell three-dimensional structures; Based on the independent single candidate cell three-dimensional structure, the centroid coordinates of the cell in the three-dimensional space are obtained by calculating the geometric center; The centroid coordinates of all independent candidate cells in the three-dimensional space are integrated to obtain a set of coordinates of the nucleus localization positive cells.

[0040] In the embodiment of the present application, the cell nucleus marker signal and the target protein marker signal are extracted from the composite image, and the two signals are processed pixel by pixel, each pixel corresponding to a position point in a three-dimensional space. For each position point, the intensity value of the cell nucleus signal and the intensity value of the target protein signal are read respectively, and the correlation between the two intensity values is calculated. The specific method is to count the change trend of the intensity values of the two signals of a plurality of adjacent position points. If the change trends are consistent, the correlation is high, and vice versa. The correlation is expressed by a specific numerical value, which ranges from 0 to 1. 1 represents complete correlation, and 0 represents no correlation. The correlation values of all position points are arranged according to their three-dimensional space positions to form a three-dimensional data set representing the colocalization degree of each three-dimensional space position point.

[0041] A determination threshold of nuclear localization positivity is determined by analyzing the colocalization degree values of known nuclear localization positive samples and negative samples. For example, the threshold is set to 0.6, that is, the position points with a colocalization degree value greater than 0.6 are considered to belong to the nuclear localization positive region. The three-dimensional data set representing the colocalization degree is traversed, and the value of each position point is checked one by one. If the value is greater than 0.6, the position point is marked as a suspected nuclear localization positive point. The three-dimensional space regions where all the marked suspected nuclear localization positive points are located are integrated to obtain all the three-dimensional space regions in the three-dimensional data set with values greater than the determination threshold. These regions are the candidate nuclear localization positive cell regions.

[0042] For the candidate nuclear localization positive cell regions, a three-dimensional connected domain analysis algorithm is applied. An unmarked position point is randomly selected from the candidate regions as a starting point. Then, it is checked whether the adjacent position points of the starting point in the three-dimensional space (six directions of front and back, left and right, and up and down) also belong to the candidate nuclear localization positive cell regions. If they belong, they are included in the current connected domain. Then, the newly included position points are taken as starting points to continue checking their adjacent position points, and the range of the connected domain is continuously expanded until no new position point can be included. In this way, an independent connected domain is obtained, representing a single candidate cell three-dimensional structure. The above process is repeated to process all unmarked position points, and the entire candidate nuclear localization positive cell region is divided into a plurality of independent single candidate cell three-dimensional structures.

[0043] For each independent single candidate cell three-dimensional structure, the coordinate information of all the three-dimensional space position points contained therein is collected. The coordinates of each position point consist of three values of x, y, and z. The average value of the x coordinates of these position points is calculated to obtain the center position of the cell three-dimensional structure in the x-axis direction. Similarly, the average values of the y coordinates and the z coordinates are calculated to obtain the center positions in the y-axis and z-axis directions, respectively. The coordinates (x average, y average, z average) composed of these three average values are the geometric center of the single candidate cell three-dimensional structure, that is, the centroid coordinates of the cell in the three-dimensional space.

[0044] The centroid coordinates of all independent candidate cells in three-dimensional space are collected one by one, arranged in a certain order (such as in the order of increasing x coordinates), and checked for duplicates or errors, and if there are, they are corrected or removed, and the final set of all centroid coordinates is the set of coordinates of the nucleus localization positive cells.

[0045] The pixel-by-pixel calculation of the intensity correlation of the two signals to obtain a set of three-dimensional data representing the degree of colocalization can accurately capture the spatial correlation of the nucleus and the target protein at the subcellular level, and the correlation value of each position point can objectively reflect the colocalization degree of the point, which helps to improve the accuracy of nuclear localization analysis. Based on the determination threshold, the region of the candidate nucleus localization positive cells can be quickly screened from a large amount of three-dimensional data by setting a reasonable threshold, reducing the interference of irrelevant regions, focusing the analysis on the target region, improving the efficiency and pertinence of the identification of nucleus localization positive cells, and applying three-dimensional connected domain analysis algorithm to segment the three-dimensional structure of independent single candidate cells, which can accurately divide the complex interconnected regions into independent cell structures, avoiding the problem of multiple cell structures being confused with each other, so that each candidate cell can be analyzed individually, ensuring the accuracy of single cell analysis. The centroid coordinates of the cells are obtained by calculating the geometric center, which can represent the position of a single cell in three-dimensional space with a simple coordinate value, accurately reflecting the spatial distribution characteristics of the cells, making the description of the cell position more intuitive. Integrating all centroid coordinates into the set of nucleus localization positive cell coordinates, the spatial position information of all nucleus localization positive cells is summarized, forming a systematic and complete coordinate data, which is helpful for in-depth study of the spatial distribution and functional relationship of nucleus localization positive cells.

[0046] In a preferred embodiment of the present application, a set of pre-set biomarkers is selected from the set of nucleus localization positive cell coordinates, a spatial topological relationship is constructed, the microenvironment heterogeneity of the spatial topological relationship is quantified, and a localization calibration parameter is generated based on the regional heterogeneity distribution characteristics, which can include: Based on the set of nucleus localization positive cell coordinates, a three-dimensional coordinate set of a target biomarker cell subset is generated by screening cell coordinates with pre-set biomarker identifiers; According to the three-dimensional coordinate set of the target biomarker cell subset, a three-dimensional spatial adjacency network is constructed with cell coordinates as nodes and spatial adjacency relationship as edges by calculating the three-dimensional Euclidean distance between coordinate points and applying an adjacency distance threshold to determine connectivity; Based on the three-dimensional spatial adjacency network, a comprehensive heterogeneity quantification dataset is generated by calculating the local cell density gradient value set of all nodes in the network, the node cluster coefficient value set and the network global spatial entropy value; According to the comprehensive heterogeneity quantitative data set, the spatial distribution characteristics of the metabolic target area are extracted by mapping to the three-dimensional space grid of the metabolic target area and analyzing the numerical distribution in the grid unit. Based on the spatial distribution characteristics of the heterogeneity, the boundary correction vector parameter set is derived by establishing the spatial mapping relationship between the characteristics and the original boundary of the metabolic target area, and the positioning calibration parameters are generated.

[0047] In the embodiment of the application, the coordinate information of each cell is extracted from the set of nucleus positioning positive cell coordinates one by one, and it is checked whether each cell carries the identification of the preset biomarker. The cell coordinates carrying the identification of the preset biomarker are screened out and sorted into a new set. For example, if the preset biomarker is "CD44+", all cell coordinates marked as "CD44+" are selected from the nucleus positioning positive cell coordinates, arranged in the order of their positions in the three-dimensional space, and a three-dimensional coordinate set of the target biomarker cell subset is formed, each coordinate containing specific numerical values of x, y and z dimensions.

[0048] For each coordinate point in the three-dimensional coordinate set of the target biomarker cell subset, the three-dimensional Euclidean distance between it and all other coordinate points in the set is calculated. When calculating, the difference of two coordinate points on the x, y and z axes is calculated respectively, each difference is squared and added, and then the square root is taken to obtain the straight-line distance between the two points. A neighborhood distance threshold (such as 50 microns) is set. If the three-dimensional Euclidean distance between two coordinate points is less than or equal to the threshold, it is determined that the two points have connectivity. Each cell coordinate is taken as a node in the network, and the nodes with connectivity are connected by edges to form a three-dimensional spatial neighborhood network with spatial neighborhood relationship as the edge, which directly presents the spatial connection between cells.

[0049] In the three-dimensional space adjacency network, for each node, the number of cells in the sphere with the node as the center and the radius of 100 microns is counted, and then divided by the volume of the sphere to obtain the local cell density of the node. By comparing the local cell densities of adjacent nodes, the density change rate, i.e. the local cell density gradient value, is calculated. The gradient values of all nodes are summarized as a local cell density gradient value set, and the clustering coefficient of each node is calculated. First, find all adjacent nodes of the node, count the actual number of edges between these adjacent nodes, and then divide by the maximum number of edges that may exist (i.e. the number of any two combinations of adjacent nodes) to obtain the clustering coefficient of the node. The clustering coefficients of all nodes form a node clustering coefficient value set. When calculating the global spatial entropy value of the network, the three-dimensional space is divided into several equal cubic grids, the number of nodes in each grid is counted, and the global spatial entropy value is obtained according to the distribution probability of the number of nodes in each grid and the calculation method of entropy. The local cell density gradient value set, the node clustering coefficient value set and the network global spatial entropy value are integrated together to form a comprehensive heterogeneity quantification data set.

[0050] Map each value in the comprehensive heterogeneity quantification data set to the three-dimensional space grid of the metabolic target area, ensure that each grid unit corresponds to a corresponding heterogeneity quantification value, analyze the distribution of the values in each grid unit, for example, count the average value, maximum value, minimum value and proportion of values in different intervals, etc. According to these statistical results, describe the spatial distribution characteristics of heterogeneity in the metabolic target area, such as which regions have high heterogeneity, which regions have low heterogeneity, and whether the change trend of heterogeneity is increasing or decreasing from the center to the edge, etc. Form a heterogeneity spatial distribution characteristic description.

[0051] Based on the heterogeneity spatial distribution characteristic description, establish the spatial mapping relationship between these characteristics and the original boundary of the metabolic target area, and determine the corresponding original boundary position of different heterogeneity characteristics. According to the deviation of heterogeneity distribution and the original boundary, deduce the boundary correction vector parameter set, each correction vector contains the distance and direction that need to be adjusted in x, y and z directions, for example, if the heterogeneity characteristic of a certain region shows that the original boundary is inside, the correction vector in this direction is positive, indicating that the boundary is adjusted outward by a certain distance. Integrate these boundary correction vector parameter sets to generate positioning calibration parameters for precise calibration of tumor boundaries.

[0052] The three-dimensional coordinate set generated by screening the cell coordinates with the preset biomarker identifier can accurately focus on the specific cell population related to the research target, exclude the interference of irrelevant cells, lay the foundation for constructing accurate spatial relationships, calculate the three-dimensional Euclidean distance, and construct a three-dimensional spatial adjacency network combined with an adjacency distance threshold, which can objectively reflect the actual spatial connection between cells, convert abstract cell coordinates into intuitive network structures, facilitate clear observation of the spatial distribution pattern and interaction relationship of cells, and provide an effective tool for analyzing the organization mode of the cell population, calculate the local cell density gradient, cluster coefficient and global spatial entropy to generate a comprehensive heterogeneity quantification data set, which quantifies the heterogeneity of cell distribution in multiple dimensions from local to global, comprehensively captures the density changes, aggregation degree and overall disorder of cells in spatial distribution, maps to a three-dimensional spatial grid and extracts heterogeneity spatial distribution features, tightly combines the quantified heterogeneity data with the spatial position of the metabolic target area, clearly shows the specific distribution rule of heterogeneity in the target area, provides detailed spatial information for understanding the complexity of the microenvironment of the metabolic target area, helps to find potential regional differences, and derives a boundary correction vector based on the heterogeneity characteristics to generate positioning calibration parameters, which can accurately correct the original boundary according to the actual heterogeneity of the microenvironment, make the positioning result more consistent with the true state of the biological tissue, reduce the positioning deviation caused by ignoring the microenvironment difference, and improve the accuracy and reliability of tumor boundary positioning.

[0053] In a preferred embodiment of the present application, based on the positioning calibration parameters, the three-dimensional coordinates of the tumor boundary are determined by spatial gradient change analysis and radio frequency signal feature attenuation association rules to obtain the tumor cell positioning result, which can include: Based on the positioning calibration parameters and the lesion classification result, the spatial mapping information of the positioning calibration parameters is superimposed on the three-dimensional spatial coordinate framework of the original fusion data set to generate the initial spatial position data of the calibrated candidate boundary; Based on the initial spatial position data of the calibrated candidate boundary, the radio frequency signal attenuation feature values of the region near the candidate position are extracted, and the image spatial gradient intensity values of the region near the same position in the ultrasonic image component data are calculated, and matching verification is performed based on the preset spatial coordination association rule to generate the spatial position data set of the candidate boundary; Based on the candidate boundary spatial position data set, the spatial continuity of the boundary points is adjusted by applying a dynamic programming algorithm to generate a boundary point set that meets the anatomical structure constraint, specifically including: taking the tumor boundary three-dimensional coordinate set as the candidate boundary spatial position data set, inputting into the framework based on the dynamic programming algorithm; for the candidate boundary spatial position data set, according to the target organ anatomical structure constraint, a path evaluation rule for measuring the smoothness of the connection path between any two candidate boundary points is generated; based on the path evaluation rule, the global smoothness final boundary point spatial connection sequence is found in the candidate boundary spatial position data set by the dynamic programming algorithm; based on the boundary point spatial connection sequence, the final tumor boundary point set that meets the anatomical structure smoothness constraint is generated. Based on the boundary point set that meets the anatomical structure constraint, three-dimensional spatial coordinate information is extracted to obtain a tumor boundary three-dimensional coordinate point set as a tumor cell positioning result.

[0054] In the embodiment of the present application, the positioning calibration parameters and the lesion classification results are obtained. The positioning calibration parameters include device error compensation, patient position offset correction and other data, such as the spatial coordinate deviation value between different imaging devices (such as ultrasound and radio frequency devices), and the position correction amount caused by patient breathing and limb movement. The lesion classification result clearly shows the approximate type of the tumor, the possible growth range and other information, for example, whether it is a benign or malignant tumor, and the common boundary morphology characteristics of the tumor of this type according to the past case statistics. Then, the three-dimensional spatial coordinate framework of the original fusion data set is processed. The original fusion data set is a set formed by integrating various modal data such as ultrasound image data and radio frequency signal data. The three-dimensional spatial coordinate framework is established according to the scanning range and accuracy of the imaging device. Each data point has corresponding x, y and z axis coordinate values for identifying the specific position of the point in the three-dimensional space of the human body.

[0055] Then, the superposition of the spatial mapping information is performed. The spatial mapping information in the positioning calibration parameters, that is, the coordinate conversion rule after error correction, is applied to the three-dimensional spatial coordinate framework of the original fusion data set. For example, if the positioning calibration parameters show that there is a 2mm offset in the x-axis of the ultrasound device, the x-coordinate of all data points in the original framework is increased by 2mm to complete the calibration and generate the calibrated candidate boundary initial spatial position data. These data points preliminarily outline the possible boundary range of the tumor.

[0056] Based on the generated calibrated candidate boundary initial spatial position data, each candidate position point is determined, and then a specific nearby area, such as a sphere with a radius of 5 mm, is demarcated around each candidate position point. In this nearby area, the radio frequency signal attenuation characteristic value is extracted. Specifically, the intensity data of all radio frequency signals in the area is collected, and the attenuation amount of the signal from transmission to reception is calculated. For example, if the transmission signal intensity is 100 units and the received signal intensity is 30 units, then the attenuation amount of the point is 70 units. The attenuation amounts of all points in the area are then averaged or taken as the median to obtain the radio frequency signal attenuation characteristic value of the area.

[0057] At the same time, the same spatial position range as the above-mentioned candidate position nearby area is found in the ultrasound image component data, and the image spatial gradient intensity value is calculated in this range. The gray values of different pixel points in the ultrasound image are different, and the gradient intensity reflects the speed of change of the gray value. When calculating, the gray value of each pixel point in the area is first obtained, and then the gray difference of adjacent pixel points is compared. For example, for two adjacent pixels in the horizontal direction, the gray values are 80 and 120 respectively, and their difference is 40. The average gradient value is calculated as the image spatial gradient intensity value by synthesizing the gray difference of all adjacent pixels in the area.

[0058] Then, a preset spatial correlation rule is called, which specifies the matching relationship that should be met between the radio frequency signal attenuation characteristic value and the ultrasound image spatial gradient intensity value at the tumor boundary. For example, when the radio frequency signal attenuation characteristic value is in the range of 60-80 units, the ultrasound image spatial gradient intensity value should be in the range of 30-50 units. The two characteristic values of the area calculated above are compared with the rule. If the matching condition is met, the candidate position point is included in the candidate boundary spatial position data set; if not, the point is excluded. Finally, the candidate boundary spatial position data set is generated.

[0059] A dynamic programming algorithm is applied to adjust the spatial continuity of the boundary points. The dynamic programming algorithm processes each candidate boundary point in a certain order (such as from the top to the bottom of the tumor, or from left to right). For two adjacent boundary points, the spatial distance and angle between them are calculated. If the distance is too large or the angle change is too abrupt, it means that the continuity of the two points is poor and there may be errors. The algorithm adjusts according to the preset continuity standard. For example, the maximum allowed distance between adjacent points is set to 10 mm, and the maximum allowed angle change is 30 degrees. If the distance between two adjacent points is 15 mm, which exceeds the standard, a new point is inserted between the two points to make the distance between adjacent points meet the requirements. If the angle change is too large, the coordinates of one of the points are adjusted to make the angle within a reasonable range.

[0060] In the adjustment process, anatomical constraint conditions are introduced at the same time, which are determined according to human anatomy knowledge, such as the tumor boundary cannot pass through specific anatomical structures such as bones, blood vessels and the like, and should meet the morphological characteristics of the normal tissues of the part, for example, the tumor near the lung cannot enter the normal inflation area of the lung, and the tumor near the liver should be coordinated with the outline of the liver, and the dynamic programming algorithm checks whether each adjusted boundary point meets the anatomical constraint, and if not, the position of the boundary point is further adjusted until the curve or surface formed by all the boundary points is continuous in space and meets the anatomical constraint, and finally a boundary point set meeting the anatomical constraint is generated.

[0061] Based on the generated boundary point set meeting the anatomical constraint, the three-dimensional spatial coordinate information of each boundary point is extracted one by one, that is, the specific numerical value of each point corresponding to the x, y and z axes, and these three-dimensional coordinate points are arranged in order to form a complete set, which is a tumor boundary three-dimensional coordinate point set. The point set can accurately depict the boundary shape and position of the tumor in three-dimensional space, so as to obtain the tumor cell positioning result and provide accurate position information for diagnosis and treatment.

[0062] By calibrating the original data through the positioning calibration parameter, the initial spatial position data is more accurate, and by combining the synergistic matching verification of the radio frequency signal attenuation characteristics and the ultrasonic image spatial gradient intensity, the points meeting the tumor boundary characteristics are further screened, the accuracy of the tumor boundary positioning is improved, the foundation for accurately judging the tumor range is laid, the dynamic programming algorithm is used to adjust the spatial continuity of the boundary points, the smooth transition of the tumor boundary in space is ensured, and abrupt transitions or breaks are avoided. At the same time, the anatomical constraint is introduced, so that the tumor boundary meets the normal anatomical structure characteristics of the human body, the boundary crossing unreasonable tissue regions is avoided, the rationality and reliability of the positioning result are enhanced, the possibility of misdiagnosis and missed diagnosis is reduced, and the tumor boundary three-dimensional coordinate point set can clearly and accurately show the specific position and morphology of the tumor in the body. A more accurate treatment plan can be developed according to the information.

[0063] As shown in Figure 2 The embodiment of the present application also provides a tumor cell auxiliary positioning method based on a composite image, which comprises the following steps: Step 1, synchronously collecting radio frequency echo signals, ultrasonic images and computed tomography data of a target biological tissue to generate an original fusion data set; Step 2, performing spatial registration and feature extraction on the original fusion data set to obtain a multi-dimensional feature set; Step 3, extracting multi-band radio frequency echo signal intensity level values of a target region based on the multi-dimensional feature set, generating a tissue radio frequency characteristic map through logarithmic compression, and performing pixel-level fusion with a synchronous ultrasonic image to generate a composite image. Step 4, subcellular distribution analysis is performed on the composite image to obtain a set of nuclear localization positive cell coordinates; Step 5, a set of preset biomarkers is selected from the set of nuclear localization positive cell coordinates, a spatial topological relationship is constructed, microenvironment heterogeneity is quantified on the spatial topological relationship, and positioning calibration parameters are generated based on the regional heterogeneity distribution characteristics; Step 6, based on the positioning calibration parameters, a three-dimensional coordinate of the tumor boundary is determined through spatial gradient change analysis and radio frequency signal feature attenuation correlation rules, to obtain a tumor cell positioning result.

[0064] It should be noted that the method is a method corresponding to the above-mentioned system, and all the implementation manners in the above-mentioned system embodiment are applicable to this embodiment and can achieve the same technical effects.

[0065] Embodiments of the application also provide a computing device, comprising a processor and a memory storing a computer program, wherein the computer program is executed by the processor to perform the method as described above. All the implementation manners in the above-mentioned method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0066] Embodiments of the application also provide a computer readable storage medium storing instructions, which, when executed on a computer, cause the computer to perform the method as described above. All the implementation manners in the above-mentioned method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0067] The above is the preferred embodiment of the application, and it should be noted that for those skilled in the art, without departing from the principles of the application, a number of improvements and refinements can be made, which should also be considered within the scope of protection of the application.

Claims

1. A tumor cell assisted positioning system based on composite images, characterized in that, The method comprises the following steps: a data acquisition module for synchronously collecting radio frequency echo signals, ultrasound images and computed tomography data of a target biological tissue to generate an original fusion data set; a feature processing module for spatial registration and feature extraction of the original fusion data set to obtain a multi-dimensional feature set; an identification imaging module for extracting multi-band radio frequency echo signal intensity level values of a target region based on the multi-dimensional feature set, generating a tissue radio frequency characteristic map through logarithmic compression, and performing pixel-level fusion with a synchronous ultrasound image to generate a composite image; a positioning analysis module for subcellular distribution analysis of the composite image to obtain a set of nuclear positioning positive cell coordinates; a positioning calibration module for selecting a set of pre-set biomarkers from the set of nuclear positioning positive cell coordinates, constructing a spatial topological relationship, quantifying the microenvironment heterogeneity of the spatial topological relationship, and generating positioning calibration parameters based on the regional heterogeneity distribution characteristics; a positioning output module for determining a three-dimensional coordinate of a tumor boundary by spatial gradient change analysis and radio frequency signal feature attenuation correlation rules based on the positioning calibration parameters to obtain a tumor cell positioning result.

2. The composite image based tumor cell assisted positioning system of claim 1, wherein, The spatial registration and feature extraction of the original fusion data set to obtain a multi-dimensional feature set comprises: based on the original fusion data set, the radio frequency echo signals, ultrasound images and computed tomography data are sequentially subjected to rigid transformation and affine transformation to achieve spatial accurate alignment, and a spatially registered fusion data set is generated; the texture features of the ultrasound image and the computed tomography density distribution features are extracted from the spatially registered fusion data set, and the frequency energy features of the radio frequency echo signals are calculated by fast Fourier transform; the ultrasound image texture features, computed tomography density distribution features and frequency energy features are integrated to obtain a multi-dimensional feature set.

3. The composite image based tumor cell assisted positioning system of claim 2, wherein, Based on the multi-dimensional feature set, the multi-band radio frequency echo signal intensity level values of the target region are extracted, the tissue radio frequency characteristic map is generated through logarithmic compression, and the pixel-level fusion is performed with the synchronous ultrasound image to generate the composite image, which comprises: extracting and separating the radio frequency echo signal intensity level values of multiple frequency bands corresponding to the target region from the multi-dimensional feature set; performing dynamic range logarithmic compression processing on the multi-band radio frequency echo signal intensity level values of the target region to generate a radio frequency signal intensity distribution matrix representing the radio frequency signal intensity distribution of the target region; mapping the radio frequency signal intensity distribution matrix to a two-dimensional plane corresponding to the spatial position of the ultrasound image to generate a tissue radio frequency characteristic map reflecting the radio frequency characteristics of the target region tissue; combining the tissue radio frequency characteristic map with the ultrasound image of the corresponding target region collected synchronously according to a predetermined rule to realize pixel-level superposition fusion and generate a composite image.

4. The composite image based tumor cell assisted positioning system of claim 3, wherein, The subcellular distribution analysis of the composite image to obtain a set of nuclear positioning positive cell coordinates comprises: calculating the intensity correlation of the cell nucleus marker signal and the target protein marker signal in the composite image by pixel to obtain a three-dimensional data set representing the co-localization degree of each three-dimensional spatial position point; Based on the three-dimensional data set representing the degree of colocalization, a determination threshold for nuclear localization positivity is set, all three-dimensional spatial regions in the three-dimensional data set with values greater than the determination threshold are identified as candidate nuclear localization positive cell regions; According to the candidate nuclear localization positive cell regions, three-dimensional spatial connectivity analysis is applied to segment the spatial regions connected to each other into independent single candidate cell three-dimensional structures; Based on the independent single candidate cell three-dimensional structures, the geometric center is calculated to obtain the centroid coordinates of the cells in the three-dimensional space; The centroid coordinates of all independent candidate cells in the three-dimensional space are integrated to obtain a set of nuclear localization positive cell coordinates.

5. The composite image based tumor cell assisted positioning system of claim 4, wherein, In the set of nuclear localization positive cell coordinates, a group of preset biomarkers is selected, a spatial topological relationship is constructed, the spatial topological relationship is quantified for microenvironment heterogeneity, and a localization calibration parameter is generated based on the regional heterogeneity distribution characteristics, including: Based on the set of nuclear localization positive cell coordinates, a three-dimensional coordinate set of a target biomarker cell subset is generated by screening cell coordinates with preset biomarker identifiers; According to the three-dimensional coordinate set of the target biomarker cell subset, a three-dimensional spatial adjacency network is constructed with cell coordinates as nodes and spatial adjacency relationships as edges by calculating the three-dimensional Euclidean distance between each coordinate point and applying an adjacency distance threshold to determine connectivity; Based on the three-dimensional spatial adjacency network, a comprehensive heterogeneity quantification data set is generated by calculating the local cell density gradient value set, the node cluster coefficient value set, and the network global spatial entropy value of all nodes in the network; According to the comprehensive heterogeneity quantification data set, the heterogeneity spatial distribution feature description in the metabolic target area is extracted by mapping to the three-dimensional spatial grid of the metabolic target area and analyzing the numerical distribution in the grid cells; Based on the heterogeneity spatial distribution feature description, a boundary correction vector parameter set is derived by establishing a spatial mapping relationship between the feature and the original boundary of the metabolic target area, and a localization calibration parameter is generated.

6. The composite image based tumor cell assisted positioning system of claim 5, wherein, Based on the localization calibration parameter, the three-dimensional coordinates of the tumor boundary are determined by spatial gradient change analysis and radio frequency signal feature attenuation association rule, to obtain the tumor cell localization result, including: Based on the localization calibration parameter and the lesion classification result, the spatial mapping information of the localization calibration parameter is superimposed on the three-dimensional spatial coordinate framework of the original fusion data set to generate the initial spatial position data of the calibrated candidate boundary; Based on the initial spatial position data of the calibrated candidate boundary, the radio frequency signal attenuation feature values of the region near the candidate position are extracted; at the same time, the image spatial gradient intensity values of the region near the same position in the ultrasound image component data are calculated, and matching verification is performed based on the preset spatial cooperative association rule to generate a candidate boundary spatial position data set; Based on the candidate boundary spatial position data set, the spatial continuity of the boundary points is adjusted by applying a dynamic programming algorithm to generate a boundary point set that conforms to the anatomical structure constraint; Based on the boundary point set that conforms to the anatomical structure constraint, the three-dimensional spatial coordinate information is extracted to obtain the three-dimensional coordinate point set of the tumor boundary as the tumor cell localization result.

7. The composite image based tumor cell assisted positioning system of claim 6, wherein, Based on the candidate boundary spatial position data set, the spatial continuity of the boundary points is adjusted by applying a dynamic programming algorithm to generate a boundary point set that meets the anatomical structure constraints, including: Inputting the tumor boundary three-dimensional coordinate set as the candidate boundary spatial position data set into the framework based on the dynamic programming algorithm; For the candidate boundary spatial position data set, according to the target organ anatomical structure constraint, a path evaluation rule for measuring the smoothness of the connection path between any two candidate boundary points is generated; Based on the path evaluation rule, the global smoothness final boundary point spatial connection sequence is found in the candidate boundary spatial position data set by the dynamic programming algorithm; Based on the boundary point spatial connection sequence, the final tumor boundary point set that meets the anatomical structure smoothness constraint is generated.

8. A method for tumor cell assisted positioning based on a composite image, the method implementing the system of any one of claims 1 to 7, characterized in that, It includes: Step 1, synchronously collecting radio frequency echo signals, ultrasonic images and computed tomography data of the target biological tissue to generate an original fusion data set; Step 2, spatial registration and feature extraction are performed on the original fusion data set to obtain a multi-dimensional feature set; Step 3, based on the multi-dimensional feature set, the multi-frequency band radio frequency echo signal intensity level value of the target region is extracted, and the tissue radio frequency characteristic map is generated after logarithmic compression, and then pixel-level fusion is performed with the synchronous ultrasonic image to generate a composite image; Step 4, subcellular distribution analysis is performed on the composite image to obtain a set of nuclear localization positive cell coordinates; Step 5, a group of pre-set biomarkers is selected from the set of nuclear localization positive cell coordinates, the spatial topological relationship is constructed, the microenvironment heterogeneity is quantified, and the positioning calibration parameter is generated based on the regional heterogeneity distribution characteristics; Step 6, based on the positioning calibration parameter, the tumor boundary three-dimensional coordinates are determined through the spatial gradient change analysis and the radio frequency signal feature attenuation correlation rule to obtain the tumor cell positioning result.

9. A computing device, comprising: It includes: One or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, so that the one or more processors implement the system as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program which is executed by the processor to implement the system as claimed in any one of claims 1 to 7.

Images