A three-dimensional reconstruction method for pathological image features of proliferative trichomythoma

Abstract

The present invention proposes a three-dimensional reconstruction method for pathological image features of proliferative tricholoma, which relates to the field of three-dimensional reconstruction technology. The method includes: obtaining a pathological slice image and its depth information to generate a pathological slice depth image containing the depth information; emitting laser signals to detect the pathological tissue and extract high-characteristic regions of the pathological tissue; emitting gamma rays to detect the pathological tissue and generate gamma-ray absorption key regions that highlight the tumor region of the pathological tissue; and combining the extracted high-characteristic regions of the pathological tissue with the gamma-ray absorption key regions to supplement the three-dimensional information of the pathological slice depth image to obtain a reconstructed three-dimensional morphology of the pathological image of proliferative tricholoma. The method of the present invention provides detailed anatomical information for subsequent treatment plan formulation.

Description

Technical Field

[0001] The invention proposes a three-dimensional reconstruction method for pathological image features of proliferative trichomoniasis, and relates to the technical field of ultrasonic image processing. Background Art

[0002] In the medical field, accurate diagnosis of proliferative tricholoma is of vital importance, and pathological section examination is one of the important ways to diagnose proliferative tricholoma. In this process, doctors obtain morphological information of tissues by observing pathological sections under a microscope. However, conventional pathological section images are only two-dimensional plane images, and the information they provide is severely limited. Two-dimensional images cannot show the structure and characteristics of pathological tissues in the depth direction, making it difficult for doctors to fully and three-dimensionally grasp the true situation inside the tissue. Key information such as the infiltration range of the tumor at different depths and the positional relationship with the surrounding normal tissues in three-dimensional space are difficult to obtain clearly from two-dimensional section images, which greatly affects the accuracy and comprehensiveness of the diagnosis.

[0003] Previous technologies also had shortcomings in detecting key characteristic areas of pathological tissue. Ordinary optical observation methods make it difficult to accurately locate high-characteristic areas with special biological or morphological significance in pathological tissue. These high-characteristic areas often contain key clues to important characteristics such as tumor growth, differentiation, and invasion. Due to the lack of effective extraction methods, doctors can easily miss this key information, resulting in inaccurate tumor assessments. In addition, existing technologies also face challenges when highlighting tumor areas in pathological tissue. Some traditional imaging technologies have difficulty clearly outlining the boundaries and scope of deep tumors or tumor areas with low contrast with surrounding tissues. This leads to large errors when doctors judge the actual size, shape, and degree of invasion of the tumor, which in turn affects the accurate formulation of subsequent treatment plans. Therefore, traditional pathological diagnosis technology has obvious shortcomings in obtaining comprehensive information about pathological tissue, accurately extracting key characteristic areas, and clearly highlighting tumor areas. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a three-dimensional reconstruction method for the pathological image features of proliferative trichomema, including: obtaining a pathological section image and the depth information of the pathological section image, and generating a pathological section depth image containing the depth information; emitting a laser signal, detecting the pathological tissue, and extracting the high-feature area of ​​the pathological tissue; emitting gamma rays, detecting the pathological tissue, and generating a gamma-ray absorption key area that highlights the tumor area of ​​the pathological tissue; combining the high-feature area and the gamma-ray absorption key area extracted from the pathological tissue, supplementing the three-dimensional information of the pathological section depth image, and obtaining a reconstructed three-dimensional morphology of the pathological image of proliferative trichomema.

[0005] In a preferred embodiment, when collecting laser signals, pathological tissue absorbs laser energy to generate a transmission signal. Transmission data at different angles is acquired through a detector. Transmission data at different transmission angles is collected to obtain a transmission laser image. The transmission data is projected onto a two-dimensional plane using a mapping function, and the characteristic map of the transmission laser image is calculated. The frequency of occurrence of each characteristic value is counted, and high-feature areas are extracted.

[0006] In a preferred embodiment, the transmission data collected by the detector at the transmission angle θ is set as P(θ, r), r is the radial coordinate, and the transmission laser image I is obtained. PA The formula is:

[0007] ;

[0008] Among them, h (rr ， ) is the mapping function, r ， Represents the radial mapping coordinates of the coordinate (x, y) in the image at the transmission angle θ;

[0009] Calculate transmitted laser image I PA The feature map is used to count the frequency of each feature value, and all the threshold values ​​T are set to calculate the difference value. :

[0010] ;

[0011] in, It is the proportion of the two types of features after segmentation according to the threshold T; is the average eigenvalue of the two types of features, is the global eigenvalue;

[0012] Traverse all thresholds T and calculate , select The largest T is used as the segmentation threshold, and the areas with feature values ​​greater than the segmentation threshold are identified as high-feature areas and extracted.

[0013] In a preferred embodiment, the projection data formed by the gamma ray recorded by the detector is assumed to be p m ,The projection data is converted into tissue absorption distribution images using an iterative algorithm:

[0014] ;

[0015] in, represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the kth iteration; A represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the k+1th iteration, m,i is the system response matrix, describing the contribution of voxel i to detector unit m, R m,iis the normalization factor, k is the number of iterations, and the tissue absorption distribution image is gradually optimized through multiple iterations.

[0016] In a preferred embodiment, the tissue absorption distribution image is subjected to three-dimensional Gaussian convolution denoising, and the denoised absorption value corresponding to voxel i in the tissue absorption distribution image is calculated. :

[0017] ;

[0018] Where Ω is the spatial neighborhood, G is the Gaussian kernel function, ji represents the spatial position difference between voxel j and voxel i, and ρ is the smoothing parameter that controls the degree of smoothing;

[0019] Based on the threshold segmentation method, the absorption threshold T is selected H , extract the key areas of gamma-ray absorption:

[0020] ;

[0021] all The area where the value is 1 constitutes the key area of ​​gamma-ray absorption.

[0022] In a preferred embodiment, DPA q Represents the image feature value of the qth neighboring voxel in the high feature area, using SPT q Represents the image feature value of the qth neighboring voxel of the gamma-ray absorption key area, and calculates the correlation coefficient r between the high feature area and the absorption key area DS :

[0023] ;

[0024] Among them, Q is the total number of neighborhood voxels, q is the qth neighborhood voxel, is the average of the image feature values ​​of all neighboring voxels in the high feature area, is the average value of the image feature values ​​of all neighboring voxels in the absorption focus area;

[0025] Based on the correlation coefficient r DS Generate a feature-absorption correlation plot:

[0026] ;

[0027] Among them, I G is the characteristic-absorption correlation value, r H is the correlation threshold; the region with the extracted feature-absorption correlation value of 100 is the region of the feature-absorption correlation diagram where the high feature region and the gamma-ray absorption key region are closely correlated.

[0028] In a preferred embodiment, the image characteristic values ​​and absorption characteristic values ​​of the characteristic-absorption correlation map are extracted and superimposed on the region corresponding to the characteristic-absorption correlation map in the depth image of the pathological slice to obtain a reconstructed three-dimensional morphological image of the pathological image of the proliferative trichomeoma. F :

[0029] ;

[0030] Among them, E0(X,Y,Z) is the grayscale eigenvalue at the depth image (X,Y,Z) of the pathological section, DPA(X,Y,Z) is the image eigenvalue at the feature-absorption correlation map (X,Y,Z), and SPT(X,Y,Z) is the absorption eigenvalue at the feature-absorption correlation map (X,Y,Z). is the image feature weight coefficient, is the absorption feature weight coefficient.

[0031] Compared with the prior art, the present invention has the following beneficial technical effects:

[0032] 1. By acquiring pathological slice images and their depth information to generate pathological slice depth images, the traditional two-dimensional pathological images can be expanded to three-dimensional space, allowing doctors to not only observe the planar structure of the tissue, but also understand its characteristics in the depth direction. This helps to more comprehensively grasp the morphology, size, positional relationship and other information of the pathological tissue, and reduce missed judgments due to the limitations of two-dimensional observation.

[0033] 2. Laser signals are emitted to detect pathological tissue and extract high-characteristic regions, highlighting portions of the pathological tissue with unique biological or morphological characteristics. These high-characteristic regions are associated with tumor growth, differentiation, invasion, and other characteristics, helping doctors more accurately identify lesions and providing a more precise basis for diagnosis.

[0034] 3. Using gamma rays to detect pathological tissues generates gamma ray absorption focus areas, specifically highlighting tumor areas within pathological tissues. Gamma rays have high energy and penetrating power, capable of penetrating tissue to a certain depth. By detecting their absorption, the boundaries and extent of the tumor can be clearly outlined. This allows for more accurate positioning and assessment of deep tumors or those with low contrast with surrounding tissue.

[0035] 4. Combining the high-feature areas extracted by laser and the absorption focus areas generated by gamma rays, and supplementing them with the 3D information of the depth image of the pathological section, the reconstructed 3D morphology integrates multiple information, allowing doctors to observe the pathological characteristics of proliferative trichomythoma more intuitively and comprehensively, helping doctors make quick and accurate judgments.

[0036] 5. Accurate 3D reconstructed images can provide detailed anatomical information for subsequent treatment planning. Doctors can gain a deeper understanding of the pathogenesis and development of leiomyoma by analyzing a large number of 3D pathological images based on the tumor's 3D morphology, location, and relationship with surrounding tissues. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 This is a flow chart of the multimodal ultrasound image fusion method based on artificial intelligence of the present invention;

[0039] Figure 2 This is a digital pathological image of rhizotheca;

[0040] Figure 3 This is a flow chart of extracting original features of an ultrasonic image according to the present invention;

[0041] Figure 4 Schematic diagram of tissue absorption distribution image. DETAILED DESCRIPTION

[0042] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0043] First, a pathological slice image and depth information of the pathological slice image are acquired to generate a pathological slice depth image containing the depth information.

[0044] In this example, the tissue sample was surgically removed skin and subcutaneous tissue. The tumor, located in the dermis and subcutaneous tissue layers, was a nodular mass protruding from the skin surface, preferably measuring 2.0 cm × 1.5 cm × 1.0 cm, 1.2 cm × 1.0 cm × 1.0 cm, or 1.8 cm × 1.5 cm × 1 cm. This sample was obtained from a 44-year-old male patient with a 1 cm diameter nodule located on the scalp.

[0045] The pathological tissue of the patient with the above-mentioned proliferative trichomoniasis was selected, and according to the conventional pathological section preparation process, the tissue was cut into slices of appropriate thickness, generally about 3-5 microns, to facilitate subsequent observation and imaging.

[0046] Obtain high-resolution pathological slice images and depth information of the pathological slice images, where the depth information is a distance map captured by a depth structured light camera.

[0047] Specifically, a pathology slice scanner is used to scan the pathology slices to generate high-resolution digital pathology images, the pathology slices are photographed through a microscope, a high-resolution camera is used to obtain two-dimensional images of the pathology slices, and a digital pathology image processing tool library is used to read and operate large-scale pathology slice images.

[0048] In a preferred embodiment, data preprocessing is performed on the acquired pathological slice images.

[0049] Normalization processing: Normalize the input pathological slice images to unify the data range.

[0050] Feature extraction: Use convolutional neural networks (such as ResNet50 or SEU-Net) to extract features of pathological slide images.

[0051] Downsampling: A sparse sampling algorithm is used to downsample pathology slide images to reduce data volume and improve processing efficiency.

[0052] like Figure 2 Shown are digital pathology images of rhizotheliomas after data preprocessing. Among the pathological features of some external rhizotheliomas, proliferation of interstitial fibrous connective tissue is a prominent finding. This proliferation is often particularly pronounced in the central region of the tumor, exhibiting a unique evolutionary process. As the tumor progresses, the fibrous connective tissue in the central region continues to proliferate, becoming dense and disordered. This irregular, pseudo-invasive appearance can easily interfere with diagnosis, requiring careful identification and comprehensive assessment of multiple pathological features to accurately distinguish external rhizotheliomas from other neoplastic lesions with similar appearances.

[0053] Optical coherence tomography (OCT) technology is used based on the principle of light interference to obtain depth information of different positions of pathological sections by measuring the reflection and scattering of light in pathological tissues, such as Figure 3 shown.

[0054] After obtaining the pathological slice image and depth information, the generator receives the pathological slice image and depth information as input and generates a fused image; the discriminator judges whether the fused image is true or false. The generator and the discriminator are continuously trained against each other. The generator learns to generate more realistic fused images, making it difficult for the discriminator to distinguish, and finally obtains the ideal fusion result, that is, a pathological slice depth image containing depth information is generated.

[0055] Secondly, a laser signal is emitted to the pathological tissue to detect the pathological tissue and extract the high-feature areas of the pathological tissue; gamma rays are emitted to the pathological tissue to detect the pathological tissue and generate gamma-ray absorption key areas that highlight the tumor areas of the pathological tissue.

[0056] The pathological characteristics of proliferative tricholomas are proliferation of outer root sheath cells of hair follicles, the distribution of tumor cells in clusters or cords, and a unique matrix composition. In the captured laser images, densely packed tumor cell areas and areas with unique matrix components appear as highly characteristic regions in the transmitted laser images. Key gamma-ray absorption areas reflect areas of active tumor cell metabolism or tracer enrichment. Therefore, it is necessary to subsequently emit laser signals to detect the pathological tissue, extract the highly characteristic regions of the pathological tissue, and then emit gamma rays again to detect the pathological tissue, generating key gamma-ray absorption areas that highlight the tumor areas in the pathological tissue.

[0057] When emitting laser signals, detecting pathological tissues and collecting laser transmission signals, the pathological tissues absorb the laser energy to generate transmission signals, and the transmission data at different angles are obtained through the detector.

[0058] Collect transmission data at different transmission angles to obtain a transmission laser image. Use a mapping function to map the transmission data to a two-dimensional plane, calculate the characteristic map of the transmission laser image, count the frequency of each characteristic value, and extract high-feature areas. The specific method is as follows:

[0059] Assume that the transmission data collected by the detector at the transmission angle θ is P(θ, r), r is the radial coordinate, and the transmission laser image I is obtained PA The formula is:

[0060]

[0061] Among them, r ， Represents the radial mapping coordinates of the image coordinate (x, y) at the transmission angle θ.

[0062] h(rr ， ) is a mapping function that improves the image resolution and controls the degree of filtering by adjusting the filtering parameter β.

[0063] Extract high feature areas based on threshold segmentation:

[0064] Calculate transmitted laser image I PAThe feature map is used to count the frequency of each eigenvalue.

[0065] Traverse all set thresholds T and calculate the difference value :

[0066] ;

[0067] in, It is the proportion of the two types of features after segmentation according to the threshold T; is the average eigenvalue of the two types of features, is the global eigenvalue.

[0068] Calculate all thresholds T , select The maximum T is used as the segmentation threshold. The segmentation method corresponding to this segmentation threshold can make the transmitted laser image I PA The difference between the two types of features is the largest.

[0069] Using the segmentation threshold corresponding to this maximum difference value as the segmentation criterion, regions with feature values ​​greater than the segmentation threshold T are identified as high-feature regions and extracted. This is because under this segmentation, the difference between high-feature regions and other regions is the most significant, which can better highlight the high-feature parts in the transmitted laser image.

[0070] Next, gamma rays are emitted to detect pathological tissue.

[0071] The detector records the projection data formed by gamma rays and uses an iterative algorithm to convert the projection data into a tissue absorption distribution image. PT Extract key areas of gamma-ray absorption that highlight tumor regions in pathological tissue.

[0072] Specifically, the projection data formed by the gamma rays recorded by the detector is iteratively converted into a tissue absorption distribution image, such as Figure 4 As shown in FIG, high-frequency noise in the tissue absorption distribution image is suppressed by three-dimensional Gaussian convolution to obtain a denoised image, thereby obtaining a high-quality image reflecting the pathological tissue tumor area.

[0073] Assume that the projection data formed by the gamma ray recorded by the detector is p m ,The projection data is converted into tissue absorption distribution images using an iterative algorithm:

[0074] ;

[0075] in, represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the kth iteration; It represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the k+1th iteration.

[0076] A m,i is the system response matrix, describing the contribution of voxel i to detector unit m, A m,i is the system response matrix, which describes the contribution of voxel i to detector unit m and reflects the probability that the ray emitted from voxel i is detected by detector unit m; R m,i is the normalization factor, which is used to normalize the relevant calculations to ensure the rationality and numerical stability of the calculation process; k is the number of iterations, and the tissue absorption distribution image is gradually optimized through multiple iterations to make the image more accurately reflect the true situation of the tissue.

[0077] p m The detector records the projection data formed by gamma rays, and m is the detector single element index, which represents the projection signal intensity received by the mth detector unit.

[0078] The iterative algorithm uses the projection data p of the detector m , combined with the system response matrix A m,i and normalization factor R m,i , continuously updating tissue absorption distribution image I PT The absorption value of each voxel i in the image is gradually reconstructed from the projection data to obtain the tissue absorption distribution image I PT .

[0079] Tissue absorption distribution image I PT Perform three-dimensional Gaussian convolution denoising and calculate the denoised absorption value corresponding to voxel i in the tissue absorption distribution image :

[0080] ;

[0081] Where Ω is the spatial neighborhood, G is the Gaussian kernel function, ji represents the spatial position difference between voxel j and voxel i, and ρ is the smoothing parameter that controls the degree of smoothing. The larger ρ is, the wider the Gaussian kernel is and the more obvious the smoothing effect is.

[0082] Based on the threshold segmentation method, the absorption threshold T is selected H , extract the key areas of gamma-ray absorption:

[0083] ;

[0084] Absorption threshold T H It should be determined based on clinical criteria or statistical methods. The area where the value is 1 constitutes the key area of ​​gamma-ray absorption.

[0085] Finally, the extracted high-feature areas and gamma-ray absorption key areas of the pathological tissue are combined with the three-dimensional information of the depth image of the pathological section to obtain the three-dimensional morphology of the reconstructed hyperplastic trichomyoma pathological image.

[0086] The extracted high-feature areas and gamma-ray absorption key areas are mapped to the spatial positions of pathological sections, marking key structures such as the trichomonoblastoma cell proliferation area and metabolically active area.

[0087] Based on the spatial position relationship of multiple slices, a 3D reconstruction algorithm is used to q The image feature value of the qth neighboring voxel in the high feature area is represented by SPT q Represents the image feature value of the qth neighborhood voxel in the gamma-ray absorption focus area.

[0088] Calculate the correlation coefficient r between high feature areas and absorption key areas DS :

[0089] ;

[0090] Among them, Q is the total number of neighborhood voxels, q is the qth neighborhood voxel, is the average of the image feature values ​​of all neighboring voxels in the high feature area, is the average value of the image feature values ​​of all neighboring voxels in the absorption focus area.

[0091] Based on the correlation coefficient r DS Generate a feature-absorption correlation plot:

[0092] ;

[0093] Among them, I G is the characteristic-absorption correlation value, r H is the correlation threshold; extract the area where the feature-absorption correlation value is 100, that is The corresponding area is the area of ​​the feature-absorption correlation diagram where the high feature area and the gamma ray absorption key area are closely correlated. The image feature value and absorption feature value of the feature-absorption correlation diagram are extracted and superimposed on the area corresponding to the feature-absorption correlation diagram in the depth image of the pathological section (I G = 100), and obtain the reconstructed three-dimensional morphological image of the proliferative trichomeoma pathological image I F :

[0094] ;

[0095] Among them, E0(X,Y,Z) is the grayscale eigenvalue at the depth image (X,Y,Z) of the pathological section, DPA(X,Y,Z) is the image eigenvalue at the feature-absorption correlation map (X,Y,Z), and SPT(X,Y,Z) is the absorption eigenvalue at the feature-absorption correlation map (X,Y,Z). is the image feature weight coefficient, is the absorption feature weight coefficient.

[0096] It should be noted that the feature-absorption correlation map is a partial area map in the pathological section depth image, that is, the area of ​​focus in the pathological section depth image. Therefore, the feature-absorption correlation map (X, Y, Z) and the pathological section depth image (X, Y, Z) correspond to the same position.

[0097] Deep imaging of pathological sections provides morphological information at the cellular and tissue levels, but lacks metabolic and functional information. Laser signals reflect the optical absorption properties of tissue and are sensitive to information such as vascularity and blood oxygen levels. Gamma-ray absorption data can reveal the metabolic activity of tissue. Combining these three data sets allows for comprehensive assessment from multiple perspectives, including morphology, function, and metabolism, to reduce misdiagnosis and missed diagnoses.

[0098] The quantitative evaluation index results of each reconstruction method are organized into Table 1 for comparison, which intuitively shows the differences in reconstruction accuracy among different methods.

[0099] Table 1 Differences in reconstruction accuracy among different methods

[0100]

[0101] As shown in Table 1, the reconstruction method of the present invention combines laser signals with gamma-ray detection to precisely capture pathological tissue features with high resolution (0.5 μm), a low measurement error rate (2%) for tumor regions and characteristic structures, and a small deviation from the true value (1 μm).

[0102] Traditional two-dimensional image reconstruction method: Relying only on planar image reconstruction, it is difficult to accurately present three-dimensional structures, with low resolution (50μm), high error rate (20%), and large deviation (20μm).

[0103] 3D reconstruction method based on pure optical imaging: Limited by the resolution of the optical system, it has insufficient ability to distinguish subtle features, with a resolution of 1μm, an error rate of 5%, and a deviation of 5μm.

[0104] Three-dimensional reconstruction method based on low-dose CT: The resolution of subtle soft tissue features is poor, with a resolution of 3μm, an error rate of 8%, and a deviation of 8μm.

[0105] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0106] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0107] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0108] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. The database involved in the embodiments provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited to this. The processor involved in the embodiments provided in this application may be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, etc., but is not limited to this.

[0109] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0110] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A three-dimensional reconstruction method for pathological image features of proliferative trichomythoma, characterized in that: include: Acquire a pathological slice image and depth information of the pathological slice image, wherein the generator receives the pathological slice image and the depth information as input and generates a fused image; The discriminator determines whether the fused image is real or fake. The generator and the discriminator are continuously trained against each other. The generator learns to generate more realistic fused images, making it difficult for the discriminator to distinguish. Ultimately, the ideal fusion result is obtained, and a pathological slice depth image containing depth information is generated. Emit laser signals, detect pathological tissue, and extract high-feature areas of pathological tissue; Emit gamma rays to detect pathological tissue and generate gamma ray absorption focus areas that highlight tumor areas in pathological tissue; The extracted high-feature areas and gamma-ray absorption key areas are mapped to the spatial positions of the pathological sections to mark the areas of cell proliferation and metabolic activity of the tricholoma. The high-feature areas and gamma-ray absorption key areas of the pathological tissue are combined with the three-dimensional information of the depth image of the pathological section to obtain the reconstructed three-dimensional morphology of the pathological image of the proliferative tricholoma. Based on the spatial position relationship of multiple slices, a 3D reconstruction algorithm is used to q The image feature value of the qth neighboring voxel in the high feature area is represented by SPT q Represents the image feature value of the qth neighboring voxel of the gamma-ray absorption key area, and calculates the correlation coefficient r between the high feature area and the absorption key area DS : ； Among them, Q is the total number of neighborhood voxels, q is the qth neighborhood voxel, is the average of the image feature values ​​of all neighboring voxels in the high feature area, is the average value of the image characteristic values ​​of all neighboring voxels in the absorption focus area; based on the correlation coefficient r DS Generate a feature-absorption correlation plot: ； Among them, I G is the characteristic-absorption correlation value, r H is the correlation threshold; the region with a feature-absorption correlation value of 100 is the region of the feature-absorption correlation diagram where the high feature region and the gamma-ray absorption key region are closely correlated; the image eigenvalue and absorption eigenvalue of the feature-absorption correlation diagram are extracted and superimposed on the region corresponding to the feature-absorption correlation diagram in the depth image of the pathological slice to obtain the reconstructed three-dimensional morphological image I of the pathological image of the proliferative trichomythoma F : ； Among them, E0(X,Y,Z) is the grayscale eigenvalue at the depth image (X,Y,Z) of the pathological section, DPA(X,Y,Z) is the image eigenvalue at the feature-absorption correlation map (X,Y,Z), and SPT(X,Y,Z) is the absorption eigenvalue at the feature-absorption correlation map (X,Y,Z). is the image feature weight coefficient, is the absorption feature weight coefficient.

2. The three-dimensional reconstruction method of pathological image features of proliferative trichomyoma according to claim 1, characterized in that: During laser signal acquisition, pathological tissue absorbs laser energy to generate transmission signals. Transmission data at different angles is acquired through the detector. Transmission data at different transmission angles is collected to obtain a transmission laser image. The transmission data is projected onto a two-dimensional plane using a mapping function, and the characteristic map of the transmission laser image is calculated. The frequency of occurrence of each characteristic value is counted, and high-feature areas are extracted.

3. The three-dimensional reconstruction method of pathological image features of proliferative trichomyoma according to claim 2, characterized in that: Assume the detector is at an angle The transmission data collected below is , r is the radial coordinate, and the transmitted laser image I is obtained PA The formula is: ； in, is the mapping function, Indicates that the coordinate (x, y) in the image is at the transmission angle The radial mapping coordinates under; Calculate transmitted laser image I PA The feature map is used to count the frequency of each feature value, and all the threshold values ​​T are set to calculate the difference value. : ； in, 、 It is the proportion of the two types of features after segmentation according to the threshold T; 、 is the average eigenvalue of the two types of features, is the global eigenvalue; Traverse all thresholds T and calculate , select The largest T is used as the segmentation threshold, and the areas with feature values ​​greater than the segmentation threshold are identified as high-feature areas and extracted.

4. The three-dimensional reconstruction method of pathological image features of proliferative trichomyoma according to claim 1, characterized in that: Assume that the projection data formed by the gamma ray recorded by the detector is p m , an iterative algorithm is used to convert the projection data into a tissue absorption distribution image: ； in, represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the kth iteration; A represents the absorption value of the tissue absorption distribution image corresponding to voxel i at the k+1th iteration, m,i is the system response matrix, describing the contribution of voxel i to detector unit m, R m,i is the normalization factor, k is the number of iterations, and the tissue absorption distribution image is gradually optimized through multiple iterations.

5. The three-dimensional reconstruction method of pathological image features of proliferative trichomyoma according to claim 4, characterized in that: Perform three-dimensional Gaussian convolution denoising on the tissue absorption distribution image and calculate the denoised absorption value corresponding to voxel i in the tissue absorption distribution image : ； in, is the spatial neighborhood, G is the Gaussian kernel function, where ji represents the spatial position difference between voxel j and voxel i, is the smoothing parameter that controls the degree of smoothing; Based on the threshold segmentation method, the absorption threshold T is selected H , extract the key areas of gamma-ray absorption: ； all The area where the value is 1 constitutes the key area of ​​gamma-ray absorption.

Images