Method for analyzing abnormal three-dimensional blood flow distribution in pulmonary ventilation-perfusion development area

Abstract

The invention discloses a lung ventilation-perfusion development area three-dimensional blood flow distribution abnormity analysis method, and relates to the technical field of lung function quantitative evaluation, and the method comprises the steps: obtaining lung CT image data of a patient to be analyzed, carrying out the structural segmentation of a plurality of sections of the lung CT image data, and obtaining a ventilation phase grayscale image and a blood flow perfusion phase grayscale image; based on the lung CT image data, accurately extracting a non-black lung region through threshold segmentation, respectively calculating a ratio of a ventilation CT value to a blood flow CT value in the region, and generating lung region segmentation data; and on the basis of the three-dimensional V / Q ratio map, identifying an abnormal sub-region in which the local ventilation and blood flow ratio deviates from the physiological range, and performing structural analysis on the abnormal sub-region to obtain a blood flow distribution abnormality analysis report. According to the method, irregular sub-regions are divided by executing the K-means clustering algorithm on the gray value in each section, adaptive segmentation of the functional heterogeneity structure in the lung is achieved, and the local ratio calculation reasonability is improved.

Description

Technical Field

[0001] This invention relates to the field of quantitative assessment of lung function, and in particular to a method for analyzing abnormal three-dimensional blood flow distribution in the lung ventilation-perfusion imaging area. Background Technology

[0002] The method for analyzing abnormal blood flow distribution in the three-dimensional ventilation-perfusion imaging area belongs to the field of quantitative analysis technology of medical images. Current conventional methods are mostly based on two-dimensional cross-sectional images to calculate the ventilation-perfusion ratio, relying on manually delineated areas or global mean estimation. This method can achieve basic lung function assessment and is suitable for preliminary clinical screening and qualitative judgment.

[0003] However, conventional methods lack the ability to integrate three-dimensional continuity in space, making it difficult to achieve accurate spatial localization of abnormal areas. At the same time, they do not perform structured cluster analysis on local functional heterogeneity at the quantitative level, resulting in a weak correlation between volume assessment and functional distribution, which affects the accurate identification and clinical interpretation of abnormal areas. Summary of the Invention

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

[0005] Therefore, this invention provides a method for analyzing three-dimensional abnormal blood flow distribution in the pulmonary ventilation-perfusion imaging region to solve the problem of inaccurate localization and quantification of abnormal pulmonary blood flow regions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region, comprising, Acquire lung CT image data of the patient to be analyzed, and perform structural segmentation of the lung CT image data in multiple sections to obtain ventilation phase grayscale image and blood perfusion phase grayscale image; Based on the lung CT image data, non-black lung regions are accurately extracted through threshold segmentation, and the ratio of ventilation CT value to blood flow CT value within each region is calculated to generate lung region segmentation data. The lung region segmentation results are input into the U-Net model for lung segmentation processing to obtain the gray values ​​in the ventilation phase image and the gray values ​​in the blood perfusion phase image. K-means clustering algorithm is performed on the gray values ​​in each section to divide the left and right lungs into multiple irregular sub-regions based on spatial proximity and functional gray similarity. Calculate the average ventilation gray value and the average blood perfusion gray value of all pixels within each irregular sub-region to determine the local ventilation-perfusion ratio of the sub-region. The local ventilation-perfusion ratio values ​​of all irregular sub-regions in each section are relocated and integrated in three-dimensional space using spatial coordinates to generate a three-dimensional V / Q ratio map. Based on the three-dimensional V / Q ratio map, abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range are identified. Through structured analysis of the abnormal sub-regions, an abnormal blood flow distribution analysis report is obtained.

[0007] As a preferred embodiment of the method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging region described in this invention, wherein: The method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging area is characterized by: acquiring lung CT image data of the patient to be analyzed, with the following specific steps. Based on the DICOM protocol interface, the system receives the ventilation phase SPECT data and perfusion phase SPECT data acquired by the patient simultaneously, and analyzes them through normalization processing to obtain lung CT images. By segmenting and quantizing lung CT images and performing masking processing, the ventilation and perfusion phase images are cropped at the pixel level using the masking, generating lung CT image data that preserves the lung tissue region in both the ventilation phase grayscale image and the blood flow perfusion phase grayscale image.

[0008] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region is characterized by: performing structural segmentation of the lung CT image data into multiple sections to obtain ventilation phase grayscale images and blood flow perfusion phase grayscale images. The specific steps are as follows. Based on synchronously acquired SPECT ventilation and perfusion combined lung CT image data, a precisely matched three-dimensional joint dataset was obtained by normalizing the spatial dimension. Based on the registered 3D joint dataset, the ventilation and perfusion phase images are cropped at the pixel level using lung parenchyma segmentation masks to generate grayscale images of the ventilation and perfusion phases.

[0009] The method for analyzing abnormal three-dimensional blood flow distribution in the ventilation-perfusion imaging region of the lungs is characterized by: based on the lung CT image data, accurately extracting non-black lung regions through threshold segmentation, calculating the ratio of ventilation CT value to blood flow CT value within each region, and generating lung region segmentation data. The specific steps are as follows. Based on lung region segmentation data, threshold segmentation is used to perform preliminary segmentation of CT image data to obtain a mask containing only lung tissue, resulting in SPECT grayscale images of ventilation and perfusion phases. Pixel mask extraction is performed on the synchronously acquired ventilation and perfusion phase SPECT grayscale images based on the mask. Lung region segmentation data is generated using the lung pixel data of the lung parenchyma region.

[0010] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region is characterized by: inputting the lung region segmentation results into a U-Net model for lung segmentation processing to obtain the grayscale values ​​in the ventilation phase image and the grayscale values ​​in the blood flow perfusion phase image. The specific steps are as follows. Based on the registered CT image data, upsampling is performed through the decoder of the U-Nt model, and multi-scale feature fusion of CT image data is combined to extract the grayscale images of ventilation phase and blood perfusion phase pixel by pixel to obtain the grayscale distribution map of pulmonary ventilation function and the grayscale distribution map of pulmonary blood perfusion function. Based on the grayscale distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function, the U-Net model network architecture is used to perform high-precision contour recognition of lung lobe boundaries and generate lung lobe segmentation masks. Based on lung lobe segmentation masking, a mask overlay method is applied to the grayscale images of the ventilation phase and the blood perfusion phase respectively to obtain the grayscale values ​​in the ventilation phase image and the grayscale values ​​in the blood perfusion phase image.

[0011] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region is characterized by: performing K-means clustering algorithm on the gray values ​​within each section, and dividing the left and right lungs into multiple irregular sub-regions based on spatial proximity and functional gray-level similarity. The specific steps are as follows. Based on grayscale images of the ventilation and perfusion phases, and combined with the corresponding spatial coordinate information, a multidimensional feature vector for each pixel is constructed. By calculating the Euclidean distance between the multidimensional feature vectors of each pixel, an iterative optimization method is used to cluster the left and right lungs respectively to obtain an initial lung sub-region distribution map. Based on the distribution map of lung sub-regions, the position of each cluster center is dynamically updated to the mean of the feature vectors of all pixels within its region. Through repeated iterations, multiple irregular sub-regions are generated for each of the left and right lungs.

[0012] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region is characterized by: calculating the average ventilation gray value and the average blood flow perfusion gray value of all pixels within each irregular sub-region to determine the local ventilation-perfusion ratio of the sub-region. The specific steps are as follows. Based on the lung region segmentation results, lung pixels in the ventilation phase grayscale image and the blood perfusion phase grayscale image are registered to obtain ventilation and perfusion grayscale values ​​with strict spatial correspondence. Based on the gray values ​​of ventilation and perfusion, a weighted fusion method is used to fuse the gray values ​​of pixels in each sub-region with the ventilation CT image and calculate the pixel values ​​to obtain the local ventilation-perfusion ratio of the sub-region.

[0013] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area is characterized by: relocating and integrating the local ventilation-perfusion ratios of all irregular sub-regions in each section using spatial coordinates to generate a three-dimensional V / Q ratio map. The specific steps are as follows: Based on the lung region segmentation results of the three planes (coronal, sagittal and axial) after registration, the central coordinates of each sub-region and the corresponding local ventilation-perfusion ratio in each plane are spatially encoded to obtain the precise location information of each sub-region in pixel space. Based on the three-dimensional location information and ratio data, K-means clustering analysis was used to perform spatial clustering analysis on the sub-regions between adjacent sections to obtain the ventilation and perfusion information of discrete regions covering the complete lung volume. The ventilation and perfusion status of discrete regions are distributed to the ventilation-perfusion ratio corresponding to each pixel in the V / Q ratio of the CT image using a color mapping method, generating a three-dimensional V / Q ratio map.

[0014] The method for analyzing abnormal three-dimensional blood flow distribution in the lung ventilation-perfusion imaging area is characterized by: identifying abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range based on the three-dimensional V / Q ratio map. The specific steps are as follows. Based on the three-dimensional V / Q ratio map, all pixels are initially screened to obtain preliminary abnormal pixel sub-regions. Based on the initial abnormal pixel sub-regions, the pixel decoder is used to aggregate spatially adjacent abnormal pixels to generate independent three-dimensional abnormal pixel sub-regions. The U-Net model, a deep learning system, is used for automated analysis of medical images. It performs quantitative analysis on each three-dimensional abnormal pixel sub-region to identify abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range.

[0015] The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area is characterized by obtaining a blood flow distribution anomaly analysis report through structured analysis of abnormal sub-regions. The specific steps are as follows: Based on the three-dimensional V / Q ratio map, morphological connections are made to pixel regions that continuously deviate from the physiological range to obtain a set of three-dimensional abnormal sub-regions. Spatial localization and volume quantification of threshold ventilation-perfusion ratio were performed on the three-dimensional abnormal sub-region set to obtain an abnormal blood flow distribution analysis report.

[0016] The beneficial effects of this invention are as follows: by performing K-means clustering algorithm on the gray values ​​in each section to divide irregular sub-regions, adaptive segmentation of functional heterogeneous structures in the lungs is achieved, improving the rationality of local ratio calculation; then, by relocating and integrating in three-dimensional space to generate a three-dimensional V / Q ratio map, the spatial distribution of abnormal areas is visualized and volume quantification is supported, ultimately achieving the dual beneficial effects of accurate localization and clinically interpretable analysis. Attached Figure Description

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

[0018] Figure 1 A flowchart for the method of analyzing abnormal three-dimensional blood flow distribution in the lung ventilation-perfusion imaging area.

[0019] Figure 2 This is a flowchart of lung CT image data processing and segmentation.

[0020] Figure 3 A flowchart for generating a 3D V / Q ratio graph.

[0021] Figure 4 A flowchart for generating abnormal blood flow distribution analysis reports. Detailed Implementation

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

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

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

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region, including the following steps: S1. Obtain lung CT image data of the patient to be analyzed, perform structural segmentation of the lung CT image data in multiple sections, and obtain ventilation phase grayscale image and blood perfusion phase grayscale image.

[0026] Based on the DICOM protocol interface, the system receives the ventilation phase SPECT data and perfusion phase SPECT data acquired by the patient simultaneously, and analyzes them through normalization processing to obtain lung CT images. Furthermore, the DICOM protocol interface is used to acquire ventilation-phase SPECT data and perfusion-phase SPECT data from medical imaging equipment, ensuring that the two sets of data are acquired synchronously and aligned in spatial coordinates. Then, normalization processing is performed on the ventilation-phase SPECT data and perfusion-phase SPECT data respectively to unify their data range to the same scale. The normalization processing uses linear scaling to map the minimum value to zero and the maximum value to one. The normalized ventilation-phase SPECT data and perfusion-phase SPECT data are input into the image fusion algorithm. The image fusion uses pixel-weighted summation, and the weighting coefficient is calculated from the pixel value. The fusion result generates a lung CT image.

[0027] It should be noted that in this invention, pixel weighted summation uses the reciprocal of spatial distance as the weighting basis to perform weighted accumulation on pixel grayscale values ​​or ventilation-perfusion ratio values, ensuring that spatial proximity dominates the calculation results. This ensures that local ventilation-perfusion ratio values, pixel ventilation-perfusion ratio mapping values, and representative values ​​of abnormal areas all reflect the weighted average characteristics of spatial distribution, ultimately supporting the generation of three-dimensional V / Q ratio maps and the output of blood flow distribution abnormality analysis reports. In this invention, pixel value calculation achieves proportional transparent overlay of CT images and masks, making the lung tissue contour visible in a semi-transparent form on the CT image. Value range clipping is performed on each pixel of the fused image, limiting the pixel value to the range of 0 to 255, ensuring that the output image conforms to the standard display format. The final fused image is the lung CT image.

[0028] By segmenting and quantizing lung CT images and performing masking processing, the ventilation and perfusion phase images are cropped at the pixel level using the masking to generate lung CT image data that preserves the lung tissue region in both the ventilation phase grayscale image and the blood perfusion phase grayscale image. Furthermore, by segmenting and quantizing the lung CT images and performing masking processing, and by identifying the lung tissue regions in the lung CT images through image segmentation, a mask image corresponding to the lung tissue regions is generated. In the mask image, the pixel values ​​of the lung tissue regions are valid values, and the pixel values ​​of the non-lung tissue regions are invalid values. Subsequently, the mask image is applied to the ventilation phase image and the perfusion phase image respectively, and a logical AND operation is performed pixel by pixel, so that only the pixels corresponding to the positions of the valid values ​​in the mask image are retained in the ventilation phase image and the perfusion phase image, and the pixels at other positions are set to invalid values. The final output ventilation phase grayscale image and perfusion phase grayscale image both contain only the lung tissue regions, and the non-lung tissue regions are completely removed, achieving accurate preservation of the lung tissue regions.

[0029] It should be noted that in this invention, masking processing is used to perform image segmentation on lung CT images. Image segmentation is based on the distribution characteristics of pixel gray values ​​in lung CT images (such as normal air-filled lung parenchyma, pulmonary vessels and bronchial walls, extrathoracic tissues, etc.) to identify the set of pixels belonging to the lung tissue region. Pixels that conform to the density characteristics of lung tissue are selected by setting a gray value range to form a preliminary outline of the lung tissue region. The masking processing results are used for subsequent lung segmentation, gray value extraction and sub-region division to ensure that all calculations and analyses are limited to the anatomical range of lung tissue, thereby improving the spatial accuracy and physiological relevance of ventilation-perfusion ratio calculation.

[0030] It should be noted that the grayscale range is set based on the differences in physical density of different tissues in lung CT images. By observing the typical grayscale performance of lung tissue, air, soft tissue and background in CT images, and combining clinical anatomy and imaging standards, the effective grayscale range for segmentation and calculation is determined. Based on synchronously acquired SPECT ventilation and perfusion combined lung CT image data, a precisely matched three-dimensional joint dataset was obtained by normalizing the spatial dimension. Furthermore, spatial resampling was performed on the SPECT ventilation phase and perfusion combined lung CT image data to ensure that the physical spacing of each pixel is consistent in the three spatial axes. Normalization was used to redistribute the pixel values. Then, the SPECT ventilation phase image and the perfusion combined lung CT image were aligned in the same spatial coordinate system. By adjusting the position and angle deviations, mutual information was used as the similarity criterion during the registration process. The transformation parameters were iteratively optimized until convergence, and finally, a three-dimensional joint dataset with accurate matching of spatial dimensions was output.

[0031] Based on the registered 3D joint dataset, the ventilation and perfusion phase images are cropped at the pixel level using the lung parenchyma segmentation mask to generate grayscale images of the ventilation phase and the blood perfusion phase. Furthermore, the lung parenchyma segmentation mask is spatially aligned pixel by pixel with the ventilation phase image, retaining the ventilation phase image pixels corresponding to the non-zero regions in the lung parenchyma segmentation mask, while the remaining regions have a value of zero, thus forming a ventilation phase grayscale image; the lung parenchyma segmentation mask is spatially aligned pixel by pixel with the perfusion phase image, retaining the perfusion phase image pixels corresponding to the non-zero regions in the lung parenchyma segmentation mask, while the remaining regions have a value of zero, thus forming a blood flow perfusion phase grayscale image.

[0032] S2. Based on lung CT image data, threshold segmentation is used to accurately extract non-black lung regions and calculate the ratio of ventilation CT value to blood flow CT value within each region to generate lung region segmentation data.

[0033] Based on lung region segmentation data, threshold segmentation is used to perform preliminary segmentation of CT image data to obtain a mask containing only lung tissue, resulting in SPECT grayscale images of ventilation and perfusion phases. Furthermore, by reading the grayscale value of each pixel in the CT image data and using the corresponding grayscale value range of lung tissue in the CT image, pixels whose grayscale values ​​fall within the range are retained, while pixels whose grayscale values ​​fall outside the range are set to zero, generating a mask containing only lung tissue. This mask is then applied to both the ventilated and perfusion-phase SPECT grayscale images. The same spatial location mask is applied to both images, retaining only the grayscale values ​​corresponding to the lung tissue region identified by the mask, while setting the grayscale values ​​of other locations to zero, resulting in the ventilated and perfusion-phase SPECT grayscale images defined by lung tissue.

[0034] Pixel mask extraction is performed on the synchronously acquired ventilation and perfusion phase SPECT grayscale images based on the mask. Lung region segmentation data is generated using lung pixel data of the lung parenchyma region. Furthermore, when extracting pixel masks from the synchronously acquired ventilation and perfusion phase SPECT grayscale images, the corresponding positions in the ventilation and perfusion phase SPECT grayscale images are determined pixel by pixel based on the area indicated by the mask. For example, at a pixel coordinate (x, y, z), if the CT mask value is 1, it indicates that the location is within the anatomical lung tissue. The radioactivity count at this point is then extracted from both the ventilation and perfusion phase SPECT images (e.g., ventilation count is 850, perfusion count is 1200) to calculate the local ventilation-perfusion ratio. If the mask value is 0 (e.g., located in the rib or mediastinal region), the radioactivity count is calculated. Even if a signal is present at a location in the SPECT image, it is not considered part of the lung parenchyma and is excluded, ensuring that all functional analyses are strictly limited to the actual lung tissue area, thus improving the physiological accuracy and clinical reliability of the V / Q ratio. If a pixel location is marked as a lung parenchyma region in the mask, the grayscale value of that pixel location in both the ventilated and perfusion phase SPECT grayscale images is retained; otherwise, the grayscale value of the pixel location is set to invalid. Image data containing only lung parenchyma region pixels is extracted from the ventilated and perfusion phase SPECT grayscale images. The extracted lung parenchyma region pixel data is used as lung region segmentation data.

[0035] S3. Input the lung region segmentation results into the U-Net model for lung segmentation processing to obtain the gray values ​​in the ventilation phase image and the gray values ​​in the blood perfusion phase image.

[0036] Based on the registered CT image data, the U-Net model decoder is used for upsampling, and combined with multi-scale feature fusion of CT image data, the grayscale images of ventilation phase and blood perfusion phase are extracted pixel by pixel to obtain the grayscale distribution map of pulmonary ventilation function and the grayscale distribution map of pulmonary blood perfusion function. Furthermore, starting from the low-resolution feature map output by the encoder, transposed convolution is performed layer by layer to gradually restore the spatial resolution. Simultaneously, during the upsampling stage at each layer, multi-scale features of CT image data extracted from the corresponding layer of the decoder in the U-Net model are stitched to the current decoder feature map, achieving the fusion of spatial details and semantic information. In the final output layer of the decoder, grayscale images of the ventilation phase and blood perfusion phase are extracted pixel by pixel. The extraction process directly maps the grayscale value corresponding to each spatial position output by the decoder to form grayscale distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function. The spatial dimensions of the two distribution maps are consistent with the registered CT image data, thus obtaining grayscale distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function.

[0037] It should be noted that, in this invention, the training of the U-Net model typically includes the following steps: First, a paired input-label dataset is prepared, using lung CT images as input and corresponding manually or semi-automatically labeled lung parenchyma masks as labels; the data is preprocessed (e.g., normalization, cropping, data augmentation, etc.) to improve generalization ability; a U-Net network structure is constructed, where the encoder path extracts multi-scale features through convolution and downsampling, and the decoder path gradually recovers spatial details through upsampling and skip connections; during training, a pixel-level loss function is used to measure the difference between the predicted mask and the real mask, and the network parameters are iteratively updated through backpropagation and an optimizer to accurately segment the lung tissue using the U-Net model; the entire processing maintains spatial alignment between input and output, ensuring that the grayscale distribution map of ventilation function and the grayscale distribution map of blood perfusion function strictly correspond at the pixel-level spatial position, providing anatomical constraints and spatial consistency guarantees for subsequent sub-region division and ventilation-perfusion ratio calculation; the grayscale value corresponding to each spatial position output by the decoder is obtained by receiving registered CT image data as input through the encoder path of the U-Net model, and the CT image data... For multi-channel images, each channel corresponds to grayscale information of different cross-sections or functional phases. The encoder extracts spatial features of the image step by step through successive convolutional layers and downsampling layers. The size of the output feature map is halved and the number of channels is doubled at each level, forming a multi-scale feature pyramid from shallow to deep. At the output of each encoder level, the spatial feature map retains the corresponding spatial resolution and local anatomical details. The decoder path starts from the deepest feature map and restores the spatial resolution step by step through transposed convolution or upsampling. After each upsampling, the feature map size is doubled and the number of channels is halved. The decoder concatenates the current layer feature map with the corresponding layer and spatial size feature map of the encoder along the channel dimension to achieve skip connections, thus fusing deep semantic information with shallow spatial details. The decoder output layer compresses the number of channels of the spatial feature map to be consistent with the number of channels of the input image through a 1×1 convolution kernel, resulting in a spatial feature map whose spatial size is completely aligned with the original input. The pixel value at each spatial location in the spatial feature map is the gray value output by the decoder at the spatial location. The gray value distribution maps output correspond to the gray distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function, respectively.

[0038] Based on the grayscale distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function, the U-Net model network architecture is used to perform high-precision contour recognition of lung lobe boundaries and generate lung lobe segmentation masks. Furthermore, two functional grayscale distribution images are used as input and fed into the network architecture of the U-Net model. The encoder part of the U-Net model extracts the spatial features of the lung lobe boundary layer by layer and reduces the resolution. The decoder part gradually restores the spatial dimension through upsampling and fuses the feature maps of the corresponding encoder layers to generate a lung lobe segmentation mask with the same size as the input image. The lung lobe segmentation mask covers the entire lung lobe area, and the value of each pixel position reflects the probability that the pixel position belongs to the lung lobe region. High-precision contour recognition of the lung lobe boundary is completed through pixel-by-pixel classification to generate a lung lobe segmentation mask covering the entire lung lobe area.

[0039] It should be noted that in this invention, the decoder consists of multiple decoding layers, each containing an upsampling and a convolution. Upsampling employs bilinear interpolation or transposed convolution to double the spatial size of the current layer's feature map, restoring the spatial resolution lost due to encoder downsampling. After each upsampling, the decoder extracts feature maps of the same spatial size from the corresponding encoder layer and fuses the convolutional and upsampled feature maps through channel-dimensional concatenation, achieving multi-scale feature fusion. The concatenated feature map then passes through two consecutive 3×3 convolutional layers, followed by a non-linear activation function to integrate contextual semantic information and local spatial details, while eliminating the checkerboard effect or blurred boundaries of upsampling. The number of channels in the feature map output by each decoding layer is halved at each level, while the spatial size is doubled at each level, gradually approaching the spatial resolution of the original lung CT image data. The decoder output layer compresses the number of channels to match the number of segmentation categories through 1×1 convolution, generating a probability value for each pixel belonging to a lung lobe structure. The probability map maintains voxel-level spatial alignment with the original lung CT image data.

[0040] Based on the lung lobe segmentation mask, the mask superposition method is applied to the grayscale images of the ventilation phase and the blood perfusion phase respectively to obtain the grayscale values ​​in the ventilation phase image and the grayscale values ​​in the blood perfusion phase image. Furthermore, the spatial locations corresponding to the non-zero regions in the lung lobe segmentation mask are extracted into the grayscale image of the ventilation phase, forming a set of grayscale values ​​in the ventilation phase image; the spatial locations corresponding to the non-zero regions in the lung lobe segmentation mask are extracted into the grayscale image of the blood perfusion phase, forming a set of grayscale values ​​in the blood perfusion phase image; grayscale value extraction is achieved through pixel-by-pixel position matching, obtaining the grayscale values ​​in the ventilation phase image and the grayscale values ​​in the blood perfusion phase image.

[0041] S4. Perform K-means clustering algorithm on the gray values ​​in each section, and divide the left and right lungs into multiple irregular sub-regions based on spatial proximity and functional gray similarity.

[0042] Based on grayscale images of the ventilation and perfusion phases, and combined with the corresponding spatial coordinate information, a multidimensional feature vector for each pixel is constructed. Furthermore, the gray values ​​of each pixel are extracted from the ventilation phase grayscale image, the gray values ​​of corresponding pixels are extracted from the blood perfusion phase grayscale image, and the X, Y, and Z coordinates of the pixels in three-dimensional space are extracted from the spatial coordinate information. The five values ​​of the ventilation phase grayscale image are arranged in a fixed order to form a five-dimensional vector, which is the multi-dimensional feature vector of each pixel.

[0043] By calculating the Euclidean distance between the multidimensional feature vectors of each pixel, an iterative optimization method is used to cluster the left and right lungs respectively to obtain an initial lung sub-region distribution map. Furthermore, for each pixel in the left and right lung regions, its corresponding multidimensional feature vector is extracted, and the Euclidean distance between any two pixels is calculated. Based on the Euclidean distance as a similarity measure, an iterative optimization method is used to cluster all pixels. In each iteration, the category to which the pixel belongs is adjusted to minimize the intra-class distance and maximize the inter-class distance. The cluster centers are updated iteratively and the pixels are redistributed until the displacement of all cluster centers in the multidimensional feature space is less than the minimized intra-class distance, thus generating the initial lung sub-region distribution map corresponding to the left and right lungs.

[0044] It should be noted that in this invention, the iterative optimization method gradually enhances the spatial proximity and functional grayscale similarity of each pixel. When the displacement of all pixel centers in the multidimensional feature space is less than the fixed convergence criterion in two consecutive iterations, the iteration stops. The final output pixel clustering results constitute the initial lung sub-region distribution map, with each sub-region corresponding to a cluster. The pixels within the sub-region have a high degree of consistency in spatial location and ventilation / perfusion grayscale. The iterative optimization method achieves functional spatial division of the lung structure by minimizing the intra-cluster Euclidean distance and maximizing the inter-cluster separation.

[0045] Based on the lung sub-region distribution map, the position of each cluster center is dynamically updated to the mean of the feature vectors of all pixels in its region, and multiple irregular sub-regions are generated for each of the left and right lungs through repeated iterations. Furthermore, the pixel set corresponding to each cluster is extracted from the lung sub-region distribution map. The arithmetic mean of the feature vectors of all pixels in each cluster is calculated, and the mean is used as the new center position of the cluster. Based on the updated center positions of each cluster, each pixel is redistributed to the region to which the nearest cluster center belongs. The pixel redistribution and cluster center position update steps are repeated until the center positions of each cluster no longer change in continuous iterations. The division result consisting of multiple irregular sub-regions in each lung is obtained. Each irregular sub-region corresponds to a stable and convergent cluster center, and the mean of the feature vectors of all pixels in each irregular sub-region is equal to the center position of its cluster, thus generating multiple irregular sub-regions in each lung.

[0046] S5. Calculate the average ventilation gray value and the average blood perfusion gray value of all pixels within each irregular sub-region to determine the local ventilation-perfusion ratio of the sub-region.

[0047] Based on the lung region segmentation results, lung pixels in the ventilation phase grayscale image and the blood perfusion phase grayscale image are registered to obtain ventilation and perfusion grayscale values ​​with strict spatial correspondence. Furthermore, using the lung region segmentation results, the pixel sets belonging to the lung region in the ventilation phase grayscale image and the blood perfusion phase grayscale image are extracted respectively. A registration method based on mutual information is used to align the spatial positions of the lung region pixels in the two images. By maximizing the mutual information of the joint distribution of the lung region grayscale values ​​in the two images, the spatial transformation parameters are optimized. The blood perfusion phase grayscale image is resampled according to the optimized spatial transformation parameters, so that each lung pixel in the resampled blood perfusion phase grayscale image completely coincides with the corresponding lung pixel in the ventilation phase grayscale image in spatial coordinates, thus obtaining ventilation and perfusion grayscale values ​​with strict spatial correspondence.

[0048] Based on the gray values ​​of ventilation and perfusion, a weighted fusion method is used to fuse the gray values ​​of pixels in each sub-region with the ventilation CT image and calculate the pixel values ​​to obtain the local ventilation-perfusion ratio of the sub-region. Furthermore, for each sub-region, the registered grayscale value is extracted and compared with the grayscale value of the corresponding pixel in the ventilation CT image. The two sets of grayscale values ​​are then weighted and fused according to a predetermined weighting coefficient. The weighting coefficient is based on the anatomical location and physiological characteristics of the sub-region. Through statistical analysis of clinical experience data, this can be further elaborated into collecting a large amount of clinical imaging and functional examination data (such as SPECT / CT, pulmonary function tests, arterial blood gas analysis, etc.) to quantify and statistically analyze the ventilation capacity, blood perfusion level, ventilation-perfusion ratio (V / Q) and their variation patterns of different lung sub-regions. The weighted and fused pixel grayscale value is used to characterize the local ventilation and blood flow status of the pixel location. The spatial average of the weighted and fused grayscale values ​​of all pixels in the sub-region is the local ventilation-perfusion ratio of the sub-region.

[0049] It should be noted that in this invention, weighted fusion uses the reciprocal of the spatial Euclidean distance between a pixel and the geometric center of the sub-region as a weight to perform pixel-weighted summation on the gray values ​​of the ventilation phase and the perfusion phase, respectively, to obtain the weighted sum of the gray values ​​of ventilation and the weighted sum of the gray values ​​of perfusion. The difference between the two yields the local ventilation-perfusion ratio of the sub-region. This ratio reflects the spatially weighted average matching relationship between ventilation and perfusion functions within the sub-region, ensuring that pixels in the central region contribute more to the results and that the contribution of pixels at the edges decreases, thereby improving the spatial representativeness and physiological rationality of the local ventilation-perfusion ratio.

[0050] S6. The local ventilation-perfusion ratio of all irregular sub-regions in each section is relocated and integrated in three-dimensional space using spatial coordinates to generate a three-dimensional V / Q ratio map.

[0051] Based on the lung region segmentation results of the three planes (coronal, sagittal and axial) after registration, the central coordinates of each sub-region and the corresponding local ventilation-perfusion ratio in each plane are spatially encoded to obtain the precise location information of each sub-region in pixel space. Furthermore, based on the lung region segmentation results of the three registered coronal, sagittal, and axial planes, the center coordinates of each sub-region within each plane are extracted. The center coordinate values ​​are obtained by calculating the geometric centroid of the pixels contained in each sub-region in its respective plane coordinate system. The center coordinates of the three planes are mapped to a unified pixel space coordinate system according to the spatial registration relationship, realizing the three-dimensional spatial alignment of the center coordinates of the three planes. The local ventilation-perfusion ratio corresponding to each sub-region is bound to its aligned center coordinates in pixel space, forming a joint encoded data structure containing spatial location and physiological parameters. The precise location information of each sub-region in pixel space is obtained by completely recording the data through a color-coded data structure.

[0052] It should be noted that in this invention, color coding is based on the ventilation-perfusion ratio value corresponding to each pixel in the three-dimensional V / Q ratio map, which is mapped to a color value. Different colors are assigned sequentially according to the ventilation-perfusion ratio value from low to high. Low ratio areas are mapped to blue, medium ratio areas are mapped to green or yellow, and high ratio areas are mapped to red. Each pixel finds the corresponding color value in the scale based on the ventilation-perfusion ratio value and assigns it to the pixel. This allows different ventilation and perfusion matching states in the three-dimensional V / Q ratio map to be presented intuitively through color differences, facilitating visual identification of the ventilation-perfusion balance zone, ventilation dominance zone, and perfusion dominance zone. The color coding process maintains a one-to-one correspondence between values ​​and colors, ensuring the spatial continuity of the ventilation-perfusion ratio distribution and the smoothness of color transitions in three-dimensional space.

[0053] Based on the three-dimensional location information and ratio data, K-means clustering analysis was used to perform spatial clustering analysis on the sub-regions between adjacent sections to obtain the ventilation and perfusion status of discrete regions covering the complete lung volume. The ventilation and perfusion status of discrete regions were distributed to the ventilation-perfusion ratio corresponding to each pixel in the V / Q ratio of the CT image through a color mapping method, generating a three-dimensional V / Q ratio map. Furthermore, by acquiring three-dimensional location information and ratio data, the three-dimensional location information is composed of the spatial coordinates of pixels in the CT image, and the ratio data is the ventilation-perfusion ratio value corresponding to each pixel. The three-dimensional location information and ratio data are combined to form a feature vector, and each feature vector contains spatial coordinates and ventilation-perfusion ratio values. K-means clustering analysis is performed on sub-regions between adjacent slices. During the clustering process, the feature vectors are used as input, and clusters are divided according to spatial proximity and ventilation-perfusion ratio. The number of clusters is determined by the natural partitioning of the lung anatomy. The cluster centers are updated iteratively until convergence. The clustering results form discrete regions covering the complete lung volume, and each discrete region has a unified ventilation and perfusion status label. A color mapping method is used to map the ventilation and perfusion status label of each discrete region to a specific color value. The color value is linearly distributed according to the ventilation-perfusion ratio value range. Each pixel in the V / Q ratio of the CT image is assigned a color value according to the color value. The color value of a pixel is determined by the ventilation and perfusion status label of its discrete region, and finally a three-dimensional V / Q ratio map is generated.

[0054] It should be noted that in this invention, K-means clustering analysis uses a multidimensional feature vector composed of three-dimensional location information and ratio data as input to perform clustering on sub-regions between adjacent sections. The number of clusters is set, and the center point of each cluster is randomly initialized. The Euclidean distance between the feature vector of each pixel and the center point of each cluster is calculated, and the pixel is assigned to the nearest cluster. After one round of assignment, the center point of each cluster is recalculated as the mean of the feature vectors of all pixels in the cluster. The assignment and center update steps are repeated until the change in the cluster center is less than the convergence criterion, thus obtaining a discrete region covering the complete lung volume. The pixels within each discrete region have a high degree of consistency in spatial location and ventilation-perfusion ratio, which is used to characterize a unified ventilation and perfusion situation, providing a structured partitioning basis for color mapping and the generation of three-dimensional V / Q ratio maps.

[0055] S7. Based on the three-dimensional V / Q ratio map, abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range are identified. Through structured analysis of the abnormal sub-regions, a blood flow distribution abnormality analysis report is obtained.

[0056] Based on the three-dimensional V / Q ratio map, all pixels are initially screened to obtain preliminary abnormal pixel sub-regions. Furthermore, in the three-dimensional V / Q ratio map, based on the continuity characteristics of the pixels in spatial and numerical distribution, by screening for values ​​that deviate from the normal range of ventilation-perfusion ratio (V / Q) by more than 1.0 ± 0.5, or SPECT signal intensity that is significantly lower than that of the adjacent area, spatial regions with abnormal performance are manually selected on the visualization interface based on the pixel set, and the selected regions are the preliminary abnormal pixel sub-regions.

[0057] It should be noted that in this invention, the ventilation-perfusion ratio is set on the three-dimensional V / Q ratio atlas visualization interface. The ventilation-perfusion ratio of each pixel in the atlas is compared according to the physiological reference interval. Pixels that exceed the spatial range are automatically marked as candidate abnormal pixels, ensuring that the initial screening results of abnormal areas meet both physiological standards and imaging experience.

[0058] Based on the initial abnormal pixel sub-regions, the pixel decoder is used to aggregate spatially adjacent abnormal pixels to generate independent three-dimensional abnormal pixel sub-regions. Furthermore, taking each pixel marked as abnormal in the initial abnormal pixel sub-region as the starting point, the pixel decoder traverses its six directly adjacent pixel positions in three-dimensional space. If an adjacent pixel also belongs to the initial abnormal pixel sub-region, it is included in the currently constructed independent three-dimensional abnormal pixel sub-region. The current independent three-dimensional abnormal pixel sub-region is no longer expanded in space. The next abnormal pixel that has not been assigned to any independent three-dimensional abnormal pixel sub-region is selected as the new starting point, and the abnormal pixel sub-region traversal and aggregation process is repeated until all pixels in the initial abnormal pixel sub-region are assigned to a certain independent three-dimensional abnormal pixel sub-region, thus generating an independent three-dimensional abnormal pixel sub-region.

[0059] The U-Net model, a deep learning system, is used for automated analysis of medical images. It performs quantitative analysis on each three-dimensional abnormal pixel sub-region to identify abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range. Furthermore, using the deep learning U-Net model, which outputs a segmentation probability map for each pixel, a set of pixels that are continuous in three-dimensional space and have a probability higher than the normal physiological distribution range is extracted based on the segmentation probability map to form a three-dimensional abnormal pixel sub-region. For each three-dimensional abnormal pixel sub-region, the statistical distribution characteristics of the ventilation-perfusion ratio corresponding to all pixels within it are calculated. The statistical distribution characteristics are compared with the physiological range point by point to mark the pixels whose ventilation-perfusion ratio falls outside the physiological range. Based on the region formed by the pixels, the abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range are obtained.

[0060] It should be noted that in this invention, the statistical distribution characteristics of the ventilation-perfusion ratio corresponding to all pixels within each three-dimensional abnormal pixel sub-region are calculated. The calculation process is as follows: the ventilation-perfusion ratio of all pixels within each three-dimensional abnormal pixel sub-region is extracted to form a one-dimensional numerical sequence. The arithmetic mean of the one-dimensional numerical sequence is calculated. The sum of the ventilation-perfusion ratios of all pixels is divided by the total number of pixels to obtain the average ventilation-perfusion ratio of the three-dimensional abnormal pixel sub-region.

[0061] Based on the three-dimensional V / Q ratio map, morphological connections are made to pixel regions that continuously deviate from the physiological range to obtain a set of three-dimensional abnormal sub-regions; Furthermore, by identifying the spatial locations where all pixel values ​​deviate from the physiological range in the three-dimensional V / Q ratio map, pixels that are spatially adjacent and both deviate from the physiological range are merged into independent connected regions. Each independent connected region is a three-dimensional abnormal sub-region, thus obtaining a set of three-dimensional abnormal sub-regions.

[0062] Spatial localization and volume quantification of threshold ventilation-perfusion ratio were performed on the three-dimensional abnormal sub-region set to obtain an abnormal blood flow distribution analysis report.

[0063] This embodiment also provides a computer device applicable to the method for analyzing abnormal three-dimensional blood flow distribution in the lung ventilation-perfusion imaging area, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the method for analyzing abnormal three-dimensional blood flow distribution in the lung ventilation-perfusion imaging area as proposed in the above embodiment.

[0064] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0065] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0066] In summary, this invention achieves adaptive segmentation of pulmonary functional heterogeneity by dividing irregular sub-regions using the K-means clustering algorithm on the gray values ​​within each cross-section, thereby improving the rationality of local ratio calculations. Furthermore, it generates a three-dimensional V / Q ratio map through three-dimensional spatial relocation and integration, making the spatial distribution of abnormal regions visible and supporting volume quantification. Ultimately, it achieves the dual benefits of accurate localization and clinically interpretable analysis.

[0067] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging region, characterized in that: include, Acquire lung CT image data of the patient to be analyzed, and perform structural segmentation of the lung CT image data in multiple sections to obtain ventilation phase grayscale image and blood perfusion phase grayscale image; Based on the lung CT image data, non-black lung regions are accurately extracted through threshold segmentation, and the ratio of ventilation CT value to blood flow CT value within each region is calculated to generate lung region segmentation data. The lung region segmentation results are input into the U-Net model for lung segmentation processing to obtain the gray values ​​in the ventilation phase image and the gray values ​​in the blood perfusion phase image. K-means clustering algorithm is performed on the gray values ​​in each section to divide the left and right lungs into multiple irregular sub-regions based on spatial proximity and functional gray similarity. Calculate the average ventilation gray value and the average blood perfusion gray value of all pixels within each irregular sub-region to determine the local ventilation-perfusion ratio of the sub-region. The local ventilation-perfusion ratio values ​​of all irregular sub-regions in each section are relocated and integrated in three-dimensional space using spatial coordinates to generate a three-dimensional V / Q ratio map. Based on the three-dimensional V / Q ratio map, abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range are identified. Through structured analysis of the abnormal sub-regions, an abnormal blood flow distribution analysis report is obtained.

2. The method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging area as described in claim 1, characterized in that: The specific steps for obtaining lung CT image data of the patient to be analyzed are as follows: Based on the DICOM protocol interface, the system receives the ventilation phase SPECT data and perfusion phase SPECT data acquired by the patient simultaneously, and analyzes them through normalization processing to obtain lung CT images. The lung CT images are segmented and quantized, and then masked. The ventilation and perfusion phase images are cropped at the pixel level using the mask to generate lung CT image data that preserves the lung tissue region in both the ventilation and perfusion phase grayscale images.

3. The method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging area as described in claim 2, characterized in that: The lung CT image data is segmented into multiple planes to obtain ventilation phase grayscale images and blood perfusion phase grayscale images. The specific steps are as follows. Based on synchronously acquired SPECT ventilation and perfusion combined lung CT image data, a precisely matched three-dimensional joint dataset was obtained by normalizing the spatial dimension. Based on the registered 3D joint dataset, the ventilation and perfusion phase images are cropped at the pixel level using lung parenchyma segmentation masks to generate grayscale images of the ventilation and perfusion phases.

4. The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as described in claim 3, characterized in that: Based on the lung CT image data, threshold segmentation is used to accurately extract non-black lung regions, and the ratio of ventilation CT value to blood flow CT value within each region is calculated to generate lung region segmentation data. The specific steps are as follows. Based on lung region segmentation data, threshold segmentation is used to perform preliminary segmentation of CT image data to obtain a mask containing only lung tissue, resulting in SPECT grayscale images of ventilation and perfusion phases. Pixel mask extraction is performed on the synchronously acquired ventilation and perfusion phase SPECT grayscale images based on the mask. Lung region segmentation data is generated using the lung pixel data of the lung parenchyma region.

5. The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as described in claim 4, characterized in that: The lung region segmentation results are input into the U-Net model for lung segmentation processing to obtain the gray values ​​in the ventilation phase image and the blood perfusion phase image. The specific steps are as follows. Based on the registered CT image data, upsampling is performed through the decoder of the U-Nt model, and multi-scale feature fusion of CT image data is combined to extract the grayscale images of ventilation phase and blood perfusion phase pixel by pixel to obtain the grayscale distribution map of pulmonary ventilation function and the grayscale distribution map of pulmonary blood perfusion function. Based on the grayscale distribution maps of intrapulmonary ventilation function and intrapulmonary blood perfusion function, the U-Net model network architecture is used to perform high-precision contour recognition of lung lobe boundaries and generate lung lobe segmentation masks. Based on lung lobe segmentation masking, a mask overlay method is applied to the grayscale images of the ventilation phase and the blood perfusion phase respectively to obtain the grayscale values ​​in the ventilation phase image and the grayscale values ​​in the blood perfusion phase image.

6. The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as described in claim 5, characterized in that: For each cross-section, a K-means clustering algorithm is performed on the gray values. Based on spatial proximity and functional gray-level similarity, the left and right lungs are each divided into multiple irregular sub-regions. The specific steps are as follows. Based on grayscale images of the ventilation and perfusion phases, and combined with the corresponding spatial coordinate information, a multidimensional feature vector for each pixel is constructed. By calculating the Euclidean distance between the multidimensional feature vectors of each pixel, an iterative optimization method is used to cluster the left and right lungs respectively to obtain an initial lung sub-region distribution map. Based on the distribution map of lung sub-regions, the position of each cluster center is dynamically updated to the mean of the feature vectors of all pixels within its region. Through repeated iterations, multiple irregular sub-regions are generated for each of the left and right lungs.

7. The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as described in claim 6, characterized in that: Calculate the average ventilation grayscale value and average blood perfusion grayscale value of all pixels within each irregular sub-region to determine the local ventilation-perfusion ratio of the sub-region. The specific steps are as follows. Based on the lung region segmentation results, lung pixels in the ventilation phase grayscale image and the blood perfusion phase grayscale image are registered to obtain ventilation and perfusion grayscale values ​​with strict spatial correspondence. Based on the gray values ​​of ventilation and perfusion, a weighted fusion method is used to fuse the gray values ​​of pixels in each sub-region with the ventilation CT image and calculate the pixel values ​​to obtain the local ventilation-perfusion ratio of the sub-region.

8. The method for analyzing three-dimensional blood flow distribution anomalies in the lung ventilation-perfusion imaging area as described in claim 7, characterized in that: The local ventilation-perfusion ratios of all irregular sub-regions in each section are relocated and integrated in three-dimensional space using spatial coordinates to generate a three-dimensional V / Q ratio map. The specific steps are as follows. Based on the lung region segmentation results of the three planes (coronal, sagittal and axial) after registration, the central coordinates of each sub-region and the corresponding local ventilation-perfusion ratio in each plane are spatially encoded to obtain the precise location information of each sub-region in pixel space. Based on the three-dimensional location information and ratio data, K-means clustering analysis was used to perform spatial clustering analysis on the sub-regions between adjacent sections to obtain the ventilation and perfusion information of discrete regions covering the complete lung volume. The ventilation and perfusion status of discrete regions are distributed to the ventilation-perfusion ratio corresponding to each pixel in the V / Q ratio of the CT image using a color mapping method, generating a three-dimensional V / Q ratio map.

9. The method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging area as described in claim 8, characterized in that: Based on the aforementioned three-dimensional V / Q ratio map, abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range are identified. The specific steps are as follows. Based on the three-dimensional V / Q ratio map, all pixels are initially screened to obtain preliminary abnormal pixel sub-regions. Based on the initial abnormal pixel sub-regions, the pixel decoder is used to aggregate spatially adjacent abnormal pixels to generate independent three-dimensional abnormal pixel sub-regions. The U-Net model, a deep learning system, is used for automated analysis of medical images. It performs quantitative analysis on each three-dimensional abnormal pixel sub-region to identify abnormal sub-regions where the local ventilation-perfusion ratio deviates from the physiological range.

10. The method for analyzing three-dimensional blood flow distribution abnormalities in the lung ventilation-perfusion imaging area as described in claim 9, characterized in that: A blood flow distribution anomaly analysis report is obtained through structured analysis of abnormal sub-regions. The specific steps are as follows. Based on the three-dimensional V / Q ratio map, morphological connections are made to pixel regions that continuously deviate from the physiological range to obtain a set of three-dimensional abnormal sub-regions. Spatial localization and volume quantification of threshold ventilation-perfusion ratio were performed on the three-dimensional abnormal sub-region set to obtain an abnormal blood flow distribution analysis report.

Images