A method, device and program product for automatic classification of pulmonary blood perfusion based on dual-energy CT

Abstract

This application relates to the field of intelligent medicine, and particularly to an automatic classification method, device and program product for pulmonary blood perfusion based on dual-energy CT. It includes obtaining the dual-energy CT scan data of the person to be tested, where the dual-energy CT scan data includes CT images and PBV images; performing lung segmentation on the CT images to obtain the segmented CT images; registering the segmented CT images with the PBV images to obtain the registered functional images; extracting functions based on the registered functional images to obtain function information; and dividing the pulmonary functional regions through the function information to obtain the division result. This application combines dual-energy CT images and deep learning technology to provide a fast, accurate and non-invasive pulmonary perfusion assessment tool for clinicians, which is particularly suitable for the diagnosis and differentiation of pulmonary vascular diseases and has good clinical value.

Description

Technical Field

[0001] The present application relates to the field of intelligent healthcare, and particularly to an automatic classification method, device, program product, and computer-readable storage medium for pulmonary blood perfusion based on dual-energy CT. Background Art

[0002] Pulmonary blood perfusion is one of the key functions of the lungs, and blood perfusion assessment plays a crucial role in the diagnosis, treatment, and prognosis evaluation of respiratory diseases. Traditional pulmonary blood perfusion assessment methods, such as pulmonary perfusion tests, although they can provide important physiological information, often cannot provide detailed spatial distribution information on the internal structure and function of the lungs. In addition, some patients may be unable to complete conventional pulmonary perfusion tests, or the test results may be affected by various factors.

[0003] In the diagnosis of pulmonary vascular diseases, the blood perfusion of acute pulmonary embolism and chronic thromboembolic pulmonary hypertension is closely related to the prognosis of patients. The secondary pulmonary blood perfusion of acute pulmonary embolism thromboembolism is highly correlated with the clinical risk of acute pulmonary embolism. The lower the vascular perfusion, the higher the risk of acute right heart failure and short-term death in patients. The corresponding quantitative evaluation of pulmonary thrombus load and pulmonary blood perfusion will be beneficial to the clinical risk assessment and treatment strategy selection of acute pulmonary embolism. In patients with chronic thromboembolic pulmonary hypertension, the pulmonary artery emboli do not match the vascular perfusion. The blood perfusion is closely related to pulmonary vascular remodeling, increased pulmonary vascular resistance, and reduced right heart function. Evaluating the degree of impairment of pulmonary blood perfusion can be used to non-invasively evaluate hemodynamics and right heart function status. Therefore, the pulmonary vascular perfusion of patients with chronic thromboembolic pulmonary hypertension is an important supplement to hemodynamic assessment. In chronic obstructive pulmonary disease and interstitial lung disease, pulmonary vascular damage, abnormal blood perfusion is closely related to the disease severity and patient prognosis. Quantitative analysis of chronic pulmonary blood perfusion during respiration is of great value for guiding treatment and monitoring the disease. Currently, blood perfusion assessment mainly uses radionuclide ventilation-perfusion imaging of the lungs. This technique requires two radionuclide imaging scans, has a low spatial resolution, poor quantitative analysis performance, and cannot accurately display diseased blood vessels. Summary of the Invention

[0004] In recent years, the development of medical imaging technology has provided new possibilities for pulmonary blood perfusion assessment. Especially the emergence of dual-energy CT technology enables the acquisition of high-resolution anatomical structure information while also providing functional information. In view of the above problems, the present invention proposes an automatic classification method for pulmonary blood perfusion based on dual-energy CT, which specifically includes:

[0005] Obtain the dual-energy CT scan data of the lungs of the person to be tested, where the dual-energy CT scan data includes CT images and PBV images;

[0006] Perform lung segmentation on the CT images to obtain the segmented CT images;

[0007] Register the segmented CT images with the PBV images to obtain registered functional images;

[0008] Extract functional information based on the registered functional images;

[0009] Divide the lung functional regions through the functional information to obtain the division result.

[0010] The region division includes perfusion defect regions, perfusion reduction regions, and healthy regions;

[0011] Optionally, the division is to input the functional information and functional images into a classifier for division to obtain the division result;

[0012] Optionally, the functional information includes local blood flow, blood volume, and mean transit time;

[0013] Optionally, the process of division by the classifier is as follows:

[0014] Obtain functional images and functional information;

[0015] Perform feature transformation on the functional information to obtain functional features;

[0016] Extract spatial distribution features based on the functional images;

[0017] Input the functional features and spatial distribution features into a classifier for division to obtain the division result;

[0018] Optionally, the process of division by the classifier further includes region elimination. After region elimination, the resulting region after removal is obtained, and an evaluation is performed based on the resulting region after removal; the region elimination is to perform morphological processing on the regions in the division result to remove isolated small regions to obtain the resulting region after removal;

[0019] Optionally, the classifier includes one or more of the following: random forest, decision tree, support vector machine, extreme learning machine, perceptron, convolutional neural network, residual network, dilated convolutional neural network.

[0020] The division is to compare the local blood flow with a preset threshold to obtain the division result;

[0021] Optionally, the preset threshold of the local blood flow in the perfusion defect region is less than that in the perfusion reduction region, and the preset threshold of the local blood flow in the perfusion reduction region is less than that in the healthy region;

[0022] Optionally, when the local blood flow is less than 10 ml / 100 ml / min, it is determined as a perfusion defect region, and when the local blood flow is greater than or equal to 30 ml / 100 ml / min, it is determined as a healthy region;

[0023] Optionally, the partitioning is first to obtain a partitioning region by comparing the local blood flow with a preset threshold, and then to input the partitioning region and the functional information into a classifier for fine partitioning of the region to obtain a partitioning result.

[0024] The registration is performed by an algorithm combining rigidity and non-rigidity for registering the segmented CT image and the PBV image; first, a rough registration result is obtained by a rigid algorithm, and then a non-rigid algorithm is used to perform local shape constraint on the rough registration result to obtain a registered functional image;

[0025] Optionally, the registration further includes fine adjustment. After local shape constraint is performed by a non-rigid algorithm, precise alignment of the local region is performed by a block matching algorithm to obtain a registered functional image;

[0026] Optionally, the rigid algorithm obtains a rough registration result by calculating the rigid transformation parameters between the segmented CT image and the PBV image and performing rough registration based on the rigid transformation parameters;

[0027] Optionally, the rigid algorithm further includes parameter optimization. The rigid transformation parameters are optimized by gradient descent to obtain optimized rigid transformation parameters, and rough registration is performed based on the optimized rigid transformation parameters to obtain a rough registration result;

[0028] Optionally, the non-rigid algorithm performs local deformation constraint through a B-spline deformation model;

[0029] Optionally, the regions for fine adjustment include: lung lobe boundaries, vascular structures;

[0030] Optionally, the registered functional image is quantitatively evaluated by a structural similarity index and a registered mutual information value. When the quantitative evaluation result is unqualified, registration or manual adjustment is performed again until the quantitative evaluation result is qualified.

[0031] The segmentation is performed by a pre-trained segmentation model to obtain a segmentation result, and the segmentation result includes the left lung and the right lung;

[0032] Optionally, the segmentation model includes one or more of the following: U-Net, FCN, SegNet, DeepLabV3+, ENet;

[0033] Optionally, the segmentation model includes an encoder and a decoder. The encoder consists of N residual blocks, where N is a natural number greater than 1. Input data extracts features through the N residual blocks connected in series in the encoder to obtain multi-scale features. The features of the Nth residual block are fed into the decoder, and the multi-scale features of the first N - 1 blocks are fed into the decoder through skip connections to fuse features of different scales to obtain decoded features. Then, an attention mechanism is used to assign weights to the decoded features to obtain important regions, and the important regions are segmented to obtain the segmentation result.

[0034] The method further includes data preprocessing. After preprocessing the CT images, processed CT images are obtained. The processed images are segmented to obtain segmented CT images.

[0035] Optionally, the data preprocessing includes pixel normalization, size unification, image denoising, content highlighting, and data augmentation.

[0036] Optionally, the content highlighting highlights the lung tissue through window width and window level.

[0037] Optionally, the image denoising removes image noise through Gaussian filtering.

[0038] Optionally, the segmentation further includes data postprocessing. After postprocessing the segmented CT images, postprocessed CT images are obtained, and the postprocessed CT images are registered with the PBV images. The postprocessing is to perform morphological processing on the segmented CT images to obtain postprocessed CT images.

[0039] Optionally, the morphological processing includes opening and closing operations, removing isolated regions, and filling holes.

[0040] Optionally, the postprocessing further includes edge optimization. After edge optimization of the CT images obtained by morphological processing, postprocessed CT images are obtained.

[0041] Optionally, the edge optimization is completed through conditional random fields.

[0042] The purpose of the present invention is to provide a method for evaluating pulmonary blood perfusion based on dual-energy CT, including:

[0043] Obtaining the dual-energy CT scan data of the lungs of the subject to be measured, where the dual-energy CT scan data includes CT images and PBV images.

[0044] Obtaining a classification result based on the above-mentioned automatic classification method for pulmonary blood perfusion based on dual-energy CT.

[0045] Performing pulmonary blood perfusion evaluation based on the classification result to obtain an evaluation result.

[0046] The object of the present invention is to provide a computer product, including a computer program or instruction, which is executed by a processor to implement the above-mentioned automatic classification method of pulmonary blood perfusion based on dual-energy CT or to execute the above-mentioned evaluation method of pulmonary blood perfusion based on dual-energy CT.

[0047] The object of the present invention is to provide a computer device, including a memory, a processor, and a computer program or instruction stored on the memory, which is executed by the processor to implement the above-mentioned automatic classification method of pulmonary blood perfusion based on dual-energy CT or to execute the above-mentioned evaluation method of pulmonary blood perfusion based on dual-energy CT.

[0048] The object of the present invention is to provide a computer-readable storage medium, on which a computer program or instruction is stored, which is executed by a processor to implement the above-mentioned automatic classification method of pulmonary blood perfusion based on dual-energy CT or to execute the above-mentioned evaluation method of pulmonary blood perfusion based on dual-energy CT.

[0049] Advantages of the present invention:

[0050] 1. Innovative diagnostic methods and processes. The present invention proposes a method for evaluating pulmonary perfusion based on dual-energy CT and deep learning models, providing a new, fast, and non-invasive diagnostic method for the diagnosis of lung diseases in clinical practice. This method specifically addresses the problems of "same disease with different signs" and "different diseases with the same signs" existing in pulmonary vascular diseases, and adopts an innovative step-by-step diagnostic process. First, the CT images are segmented by a deep learning model to identify the lung regions. Then, the segmented CT images are accurately registered with the PBV images, combining structural information and functional information. Finally, through pixel-level analysis of the PBV images, perfusion defects, reduced perfusion, and healthy regions are identified. This step-by-step refined diagnostic process not only improves the accuracy of diagnosis but also effectively differentiates easily confused disease types, greatly enhancing the clinical diagnostic efficiency.

[0051] 2. Comprehensive utilization of structural and functional information. The present invention makes full use of the advantages of dual-energy CT technology to simultaneously obtain CT images reflecting the lung structure and PBV images reflecting the lung function. Through deep learning models and image registration techniques, the accurate fusion of structural information and functional information is achieved. This comprehensive analysis method can provide more comprehensive and accurate results for pulmonary perfusion evaluation. For example, CT images can clearly show the anatomical structure of the lungs, including airways, blood vessels, and lung parenchyma, etc., while PBV images can reflect the dynamic changes in pulmonary blood perfusion. The combination of the two types of information enables doctors to simultaneously evaluate the morphological changes and functional status of the lungs, providing a solid foundation for accurate diagnosis and individualized treatment.

[0052] 3. Advanced image processing and artificial intelligence technologies. The present invention adopts a number of advanced image processing and artificial intelligence technologies to improve the accuracy and efficiency of diagnosis. First, a deep learning model (such as an improved pre-trained U-Net network) is used to perform high-precision lung segmentation on CT images, providing an accurate anatomical basis for subsequent analysis. Secondly, a non-rigid registration algorithm (such as a B-spline deformation model combined with a mutual information similarity metric) is used to achieve accurate registration of CT images and PBV images, ensuring the spatial consistency of structural and functional information. In addition, the present invention also introduces an edge detection algorithm to extract the imaging features of abnormal perfusion areas, and uses machine learning methods (such as random forests) to intelligently classify functional areas. The combined use of these technologies not only improves the accuracy of diagnosis, but also greatly reduces the time and subjective errors of manual operations.

[0053] 4. Comprehensive feature extraction and analysis. Considering the high similarity of pulmonary vascular disease images, the present invention adopts whole lung imaging and multi-scale feature extraction strategies. Specifically, the whole lung image is segmented into ROIs to obtain multiple sub-regions, and feature extraction is performed on each sub-region separately. This method not only takes into account global information, but also captures local subtle changes. During the feature extraction process, various types of imaging genomics features are used, including first-order statistical features, shape-based features, texture features, etc. Through feature fusion and dimensionality reduction techniques, a high-dimensional and high-precision feature vector is finally obtained. This comprehensive feature analysis strategy greatly improves the model's ability to recognize tiny lesions, and provides strong support for accurately distinguishing different types of interstitial lung diseases. At the same time, this method also lays the foundation for future longitudinal studies and prognostic assessments, and has the potential to discover new imaging biomarkers. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. 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 work.

[0055] Figure 1 A schematic flow chart of an automatic classification method for pulmonary blood perfusion based on dual-energy CT provided in an embodiment of the present invention;

[0056] Figure 2 A schematic diagram of an automatic classification system for pulmonary blood perfusion based on dual-energy CT provided in an embodiment of the present invention;

[0057] Figure 3 A schematic diagram of an automatic classification device for pulmonary blood perfusion based on dual-energy CT provided in an embodiment of the present invention;

[0058] Figure 4 The UNet deep learning network structure provided by the embodiments of the present invention;

[0059] Figure 5 The scheme data flow provided by the embodiments of the present invention. Specific embodiments

[0060] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0061] In some processes described in the specification, claims and above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as S101, S102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0062] Figure 1 A schematic diagram of a method for automatically classifying pulmonary blood perfusion based on dual-energy CT provided by the embodiments of the present invention specifically includes:

[0063] S101: Obtain the dual-energy CT scan data of the lungs of the person to be tested, and the dual-energy CT scan data includes CT images and PBV images;

[0064] In one embodiment, the method further includes data preprocessing. After preprocessing the CT images, the processed CT images are obtained, and after segmenting the processed images, the segmented CT images are obtained;

[0065] Optionally, the data preprocessing includes pixel normalization, size unification, image denoising, content highlighting, and data enhancement.

[0066] In one embodiment, the content is highlighted by window width and window level to highlight the lung tissue.

[0067] Optionally, the image denoising removes image noise through Gaussian filtering;

[0068] Optionally, the segmentation further includes data post-processing. After post-processing the segmented CT images, the post-processed CT images are obtained, and the post-processed CT images are registered with the PBV images; the post-processing is to perform morphological processing on the segmented CT images to obtain the post-processed CT images;

[0069] Optionally, the morphological processing includes opening and closing operations, removing isolated regions, and filling holes;

[0070] Optionally, the post-processing further includes edge optimization, and the edge of the CT image after morphological processing is optimized to obtain the post-processed CT image;

[0071] Optionally, the edge optimization is completed by a conditional random field.

[0072] In a specific embodiment, dual-energy CT (DE-CT) pulmonary enhancement scan data of a patient is obtained, including a chest CT image and a pulmonary blood volume (PBV) image. Among them, the CT image provides the structural information of the lung, and the PBV image provides the pulmonary perfusion function information. The patient starts the DE-CT scan protocol in a breath-holding state. The scanning range covers the entire lung, from the apex to the base of the lung. All patients are intravenously injected with an iodinated contrast agent at a rate of 350 - 370 mg iodine / ml, and the injection rate is 4 - 5 ml / second (patients with a body weight ≤ 80 kg use 80 ml; patients with a body weight > 80 kg use 100 ml). For a dual-source dual-energy CT (DS-DE-CT) device, the scanning parameters are set as follows: the parameters are 80 / 150 kV with a tin filter, and automatic exposure control. The reference tube current is 180 mAs, the pitch is 1.1:1, the scanning gantry rotation time is 0.28 seconds, and the detector configuration is 192×0.6. For a single-source dual-energy CT device (SS-DECT), it is performed under the condition of rapid voltage switching between 80 / 140 kV, with a fixed tube current of 280 - 360 mA, a rotation time of 0.5 seconds, a pitch of 1.375:1, and a detector configuration of 64×0.625. For all DECTs, cross-sectional virtual monoenergetic (40 keV) and material decomposition iodine (MDI) images are generated.

[0073] In a specific embodiment, data acquisition;

[0074] a) Selection of scanning equipment: Data acquisition is the starting point of the entire pulmonary perfusion assessment process, and its specific implementation details are as follows: An advanced dual-energy CT scanner, such as DS-DE-CT, is selected. These devices can simultaneously acquire CT data at two different energy levels.

[0075] b) Contrast agent injection:

[0076]

[0077] c) Scanning parameter setting:

[0078]

[0079]

[0080] d) Image reconstruction method:

[0081] Using iterative reconstruction algorithms Such as ADMIRE or ASiR-V Reconstruction slice thickness 1 mm Reconstruction interval 0.8 mm Field of view (FOV) 350 mm Image size 512×512

[0082] e) Data storage:

[0083] Use a dual-energy CT scanning device to scan the patient to obtain CT images and PBV images including the entire lungs. The CT images mainly provide detailed structural information of the lungs, including anatomical structures such as lung lobes, trachea, and blood vessels. The PBV images provide pulmonary blood perfusion information. The original data is saved in DICOM format, and a patient database is established, including establishing a patient information database, including age, gender, clinical diagnosis, etc.

[0084] In a specific embodiment, data preprocessing: perform standardization processing on the input CT images, including pixel value normalization and image size adjustment. Use window width and window level techniques to highlight lung tissues and apply Gaussian filtering to reduce noise.

[0085] Data preprocessing:

[0086] Image standardization: linearly map CT values to the interval [0, 1];

[0087] Window width and window level adjustment: the lung window is set to a window width of 1500 HU and a window level of -600 HU;

[0088] Image size adjustment: resample all images to a size of 512×512×Z, where Z is the number of slices adapted to the network input;

[0089] Data augmentation: randomly apply operations such as rotation (±15°), scaling (0.9 - 1.1 times), translation (±10%), and flipping.

[0090] S102: Perform lung segmentation on the CT image to obtain the segmented CT image;

[0091] In an embodiment, the segmentation is performed through a pre-trained segmentation model to obtain a segmentation result, and the segmentation result includes the left lung and the right lung.

[0092] Optionally, the segmentation model includes one or more of the following: U-Net, FCN, SegNet, DeepLabV3+, ENet.

[0093] In one embodiment, the segmentation model includes an encoder and a decoder. The encoder consists of N residual blocks, where N is a natural number greater than 1. Input data extracts features through the N residual blocks connected in series in the encoder to obtain multi-scale features. The features of the Nth residual block are fed into the decoder, and the multi-scale features of the first N - 1 blocks are fed into the decoder through skip connections to fuse features of different scales to obtain decoded features. Then, an attention mechanism is used to assign weights to the decoded features to obtain important regions, and the important regions are segmented to obtain the segmentation result.

[0094] In a specific embodiment, the acquired CT images are input into a pre-trained deep learning model for segmenting the CT images to obtain accurate segmentation results of the left and right lungs. The present invention uses an improved U-Net network structure as the deep learning model. This network consists of an encoder, a decoder, and skip connections, and can effectively extract multi-scale features to achieve accurate lung segmentation. The network is trained using a CT image dataset annotated by radiologists in this hospital, and data augmentation techniques are used to expand the training samples, including operations such as random rotation, scaling, translation, and flipping. During the training process, the Dice loss function and the Adam optimizer are used. The initial learning rate is set to 0.001, and a step decay strategy is adopted, with the learning rate reduced by 50% every 80 epochs.

[0095] Network structure design: An improved U-Net architecture is adopted, including an encoder and a decoder. The encoder uses residual blocks to extract multi-scale features, and the decoder fuses feature information of different levels through skip connections. An attention mechanism is added at the end of the network to strengthen the feature expression of key regions.

[0096] Model training: A large number of annotated CT image datasets are used for training. Data augmentation techniques such as random rotation, scaling, and flipping are adopted to increase the generalization ability of the model. The loss function combines the Dice loss and the cross-entropy loss to balance the class imbalance problem. The Adam optimizer is used, and a learning rate decay strategy is adopted.

[0097] Post-processing optimization: Morphological operations such as opening and closing are performed on the segmentation results output by the network to remove small isolated regions and fill holes. Conditional random fields (CRFs) are used for edge optimization to improve the segmentation accuracy. Finally, connected component analysis is applied to ensure the integrity of the left and right lungs.

[0098] In a specific embodiment, the acquired CT images are input into a pre-trained deep learning model to achieve automatic segmentation of the left and right lungs. This embodiment uses an improved pre-trained U-Net network structure, as Figure 4 shown. This network consists of an encoder, a decoder, and skip connections, and can effectively extract multi-scale features to achieve accurate lung segmentation.

[0099] Network structure description:

[0100] Encoder: 4 downsampling blocks, each block contains two 3×3 convolutional layers and one 2×2 max pooling layer;

[0101] Decoder: 4 upsampling blocks, each block contains one 2×2 transposed convolutional layer and two 3×3 convolutional layers;

[0102] Skip connection: Concatenate the feature maps of the corresponding layers of the encoder with the feature maps of the decoder;

[0103] Attention mechanism: Add a spatial attention module in each decoder block to enhance the feature representation of key regions;

[0104] Deep supervision: Add auxiliary losses on feature maps of different scales to accelerate training convergence;

[0105] Activation function: ReLU;

[0106] Final output layer: 1×1 convolutional layer, using the Sigmoid activation function;

[0107] Loss function: Combine multiple loss functions to improve segmentation accuracy:

[0108] Dice loss: Ldice = 1 - (2|X∩Y| + ε) / (|X| + |Y| + ε);

[0109] Cross-entropy loss: CE = -Σ[y log(p) + (1 - y)log(1 - p)];

[0110] Boundary loss:

[0111] Total loss: Ltotal = λ1Ldice + λ2CE + λ3Lb, where λ1, λ2, λ3 are weight coefficients;

[0112] Training strategy:

[0113] Dataset: Use 500 cases of CT datasets with expert annotations, 400 cases for training and 100 cases for validation;

[0114] Optimizer: Adam optimizer, with the initial learning rate set to 0.001;

[0115] Learning rate adjustment: Adopt the cosine annealing strategy, with the minimum learning rate of 1e-6;

[0116] Batch size: 32 (adjusted according to GPU memory);

[0117] Number of training epochs: 200 epochs, and the early stopping strategy is based on the Dice coefficient of the validation set;

[0118] Post - processing:

[0119] Connected - component analysis: Remove isolated regions with a volume less than 50 mm 3 ;

[0120] Morphological operation: Apply an opening operation with a spherical structuring element of radius 2 voxels to remove small protrusions;

[0121] Hole filling: Fill small holes in the lung parenchyma;

[0122] Separation of the left and right lungs: Use the watershed algorithm to separate the adhered left and right lungs;

[0123] Model evaluation:

[0124] Evaluation metrics: Dice coefficient, Hausdorff distance, average surface distance;

[0125] Cross - validation: Perform 5 - fold cross - validation to ensure the stability and generalization ability of the model;

[0126] Comparison with manual annotation: Calculate the consistency between the model segmentation results and the manual segmentation results of two radiologists;

[0127] After training is completed, use an independent test set of 100 cases to evaluate the model performance. The model achieved a Dice coefficient of more than 96.75% on the test set, demonstrating its good segmentation performance.

[0128] S103: Register the segmented CT image with the PBV image to obtain the registered functional image;

[0129] In one embodiment, the registration is performed by an algorithm combining rigid and non - rigid algorithms for registering the segmented CT image and the PBV image; first, perform a rough registration through a rigid algorithm to obtain a rough registration result, and then perform local shape constraint on the rough registration result through a non - rigid algorithm to obtain the registered functional image.

[0130] Optionally, the registration further includes fine - tuning. After performing local shape constraint through a non - rigid algorithm, perform precise alignment on the local area through a block - matching algorithm to obtain the registered functional image;

[0131] Optionally, the rigid algorithm calculates the rigid transformation parameters between the segmented CT image and the PBV image and performs rough registration based on the rigid transformation parameters to obtain a rough registration result;

[0132] Optionally, the rigid algorithm further includes parameter optimization. Optimize the rigid transformation parameters through gradient descent to obtain optimized rigid transformation parameters, and perform rough registration based on the optimized rigid transformation parameters to obtain a rough registration result;

[0133] Optionally, the non-rigid algorithm performs local deformation constraint through a B-spline deformation model.

[0134] In one embodiment, the regions for fine adjustment include: lobe boundaries, vascular structures.

[0135] In one embodiment, the registered functional image is quantitatively evaluated by the structural similarity index and the registered mutual information value. When the quantitative evaluation result is unqualified, re-registration or manual adjustment is performed until the quantitative evaluation result is qualified.

[0136] In a specific embodiment, the segmented CT image is registered with the PBV image to obtain the registered functional image. The registration process combines rigid and non-rigid registration algorithms, specifically using a B-spline deformation model and mutual information as the similarity metric for the registration method. The main steps of the registration algorithm include:

[0137] a) Initial alignment: First, perform a rough rigid registration to correct the overall position and orientation differences between the CT and PBV images. Mutual Information is used as the similarity metric, and the gradient descent method is used to optimize the rigid transformation parameters, including translation, rotation, and scaling. To improve efficiency, a multi-resolution strategy is adopted, gradually transitioning from low resolution to high resolution. This step can effectively eliminate large-scale deviations caused by scanning position differences and respiratory motion.

[0138] b) Non-rigid deformation: Based on the initial alignment, use a B-spline deformation model for non-rigid registration to compensate for local deformation differences. A control point grid is evenly distributed on the image, and non-rigid deformation is achieved by optimizing the positions of the control points. Normalized Cross Correlation is used as the local similarity metric, combined with image gradient information, to construct an objective function. The L-BFGS algorithm is used to iteratively optimize the control point positions and gradually refine the deformation field. To prevent excessive deformation, an elastic energy constraint term is introduced.

[0139] c) Fine adjustment and quality control: Perform local fine adjustment and registration quality assessment. Use the block matching algorithm to perform precise alignment in the local area, paying special attention to key regions such as lobe boundaries and vascular structures. Apply discrete optimization techniques, such as graph cut algorithms, to further optimize the smoothness and consistency of the deformation field. The registration quality is quantitatively evaluated by the structural similarity index (SSIM) and the registered mutual information value. For regions with substandard quality, mark and prompt that manual intervention may be required. Finally, output a high-quality registration result and the corresponding deformation field, providing a reliable basis for subsequent functional information mapping.

[0140] In a specific embodiment, the segmented lung mask is applied to the CT image and the PBV image to extract the lung region. Then, a non-rigid registration algorithm is used to register the CT image and the PBV image. In this embodiment, a B-spline deformation model and mutual information are used as a registration method with similarity measure. The main steps of the registration algorithm include:

[0141] a) Initialization:

[0142] First, an affine transformation is used to align the overall position and orientation of the two images. The affine transformation includes translation, rotation, and scaling. The affine transformation parameters are optimized by minimizing the mutual information between the two images.

[0143] b) Non-rigid deformation:

[0144] Deformation model: The B-spline free-form deformation model is adopted;

[0145] Control point grid: A 10×10×10 control point grid is evenly distributed on the image;

[0146] Similarity measure: Local normalized cross-correlation (LNCC) is used as the similarity measure;

[0147] Regularization term: A bending energy term is added as regularization to prevent excessive deformation;

[0148] c) B-spline deformation model:

[0149] A three-dimensional B-spline function is used to establish a deformation field model. A control point grid is evenly distributed on the image, and non-rigid deformation is achieved by optimizing the positions of the control points. The spacing of the control points changes with the resolution level and gradually decreases from coarse to fine.

[0150] d) Optimization:

[0151] The gradient descent method is used to optimize the mutual information similarity measure. At each resolution level, the positions of the B-spline control points are iteratively optimized until the convergence condition or the maximum number of iterations is reached.

[0152] e) Interpolation:

[0153] Cubic spline interpolation is used to calculate the deformed image to ensure the smoothness of the image deformation.

[0154] The registration quality is evaluated using the mutual information value and the structural similarity index (SSIM) of the registered image. In this embodiment, the registered CT and PBV images reach an SSIM value of more than 0.85, indicating good registration quality.

[0155] S104: Feature extraction is performed on the registered functional image to obtain functional information;

[0156] In one embodiment, the functional information includes regional blood flow, blood volume, and mean transit time.

[0157] In one embodiment, the functional information further includes perfusion abnormal regions, which are obtained by detecting perfusion regions of the registered functional images through an edge detection algorithm.

[0158] In a specific embodiment, based on the registered functional images, the functional information of the lung region is extracted. The main extracted functional parameters include:

[0159] Regional blood flow (unit: ml / 100ml / min);

[0160] Blood volume (unit: ml / 100ml);

[0161] These parameters are obtained by analyzing and calculating the pixel values of the PBV images, and reflect the blood perfusion status of different regions of the lungs.

[0162] In a specific embodiment, based on the registration result, the functional information corresponding to the lung region is extracted from the PBV images. The main extracted functional parameters include: blood flow, blood volume, mean transit time, etc. These parameters are obtained by analyzing and calculating the pixel values of the PBV images, and reflect the blood perfusion status of different regions of the lungs. The specific calculation methods are as follows:

[0163] a) Blood flow:

[0164] Calculated using the maximum slope method. At each voxel position, analyze the rising segment of the PBV time-density curve and calculate its maximum slope. The regional blood flow is proportional to the maximum slope.

[0165] b) Blood volume:

[0166] Calculated by analyzing the ratio of the area of the PBV time-density curve to the area of the arterial input function (AIF) curve. First, automatically select the main pulmonary artery as the AIF, and then use numerical integration methods to calculate the area under the curve.

[0167] c) Mean transit time:

[0168] According to the central volume principle, the mean transit time is equal to the blood volume divided by the blood flow.

[0169] S105: Obtain a partitioning result by partitioning the lung functional regions through the functional information.

[0170] In one embodiment, the region partitioning includes perfusion defect regions, perfusion reduction regions, and healthy regions.

[0171] In one embodiment, the division is to input functional information and functional images into a classifier for division to obtain a division result.

[0172] In one embodiment, the process of division by the classifier is as follows:

[0173] Obtain functional images and functional information;

[0174] Perform feature transformation on the functional information to obtain functional features;

[0175] Extract spatial distribution features based on the functional images;

[0176] Input the functional features and spatial distribution features into a classifier for division to obtain a division result.

[0177] In one embodiment, the perfusion abnormal region includes perfusion defect and perfusion reduction.

[0178] In one embodiment, during the process of division by classification, the functional information includes one or more of the following: regional blood flow, blood volume, mean transit time, perfusion abnormal region. The perfusion abnormal region is obtained by performing feature extraction through an edge detection algorithm to obtain the imaging features of the perfusion abnormal region (one of the functional features).

[0179] In one embodiment, the process of division by the classifier further includes region rejection. After region rejection, the result of the region after removal is obtained, and evaluation is performed based on the result of the region after removal; the region rejection is to perform morphological processing on the regions in the division result to remove isolated small regions to obtain the result of the region after removal.

[0180] In one embodiment, the classifier includes one or more of the following: random forest, decision tree, support vector machine, extreme learning machine, perceptron, convolutional neural network, residual network, dilated convolutional neural network.

[0181] In another embodiment, the division is to obtain a division result by comparing the regional blood flow with a preset threshold;

[0182] Optionally, the preset threshold of the regional blood flow in the perfusion defect region is less than that in the perfusion reduction region, and the preset threshold of the regional blood flow in the perfusion reduction region is less than that in the healthy region;

[0183] In one embodiment, when the regional blood flow is less than 10 ml / 100 ml / min, it is determined as the perfusion defect region, and when the regional blood flow is greater than or equal to 30 ml / 100 ml / min, it is determined as the healthy region;

[0184] In another embodiment, the partitioning is first to obtain a partitioned region by comparing the local blood flow with a preset threshold, and then the partitioned region and functional information are input into a classifier for fine partitioning of the region to obtain a partitioning result.

[0185] In a specific embodiment, the extracted functional information is analyzed and partitioned, and the lung region is partitioned into a perfusion defect area, a reduced perfusion area, and a healthy area. The partitioning method uses a method combining an adaptive threshold and machine learning:

[0186] a) First, based on the statistical data of the normal population, an initial threshold is set;

[0187] Perfusion defect: local blood flow < 10 ml / 100 ml / min;

[0188] Reduced perfusion: 10 ml / 100 ml / min ≤ local blood flow < 30 ml / 100 ml / min;

[0189] Healthy area: local blood flow ≥ 30 ml / 100 ml / min;

[0190] b) Then, a random forest model (RF) classifier is used to optimize the preliminary partitioning result. The feature inputs of the RF classifier include local blood flow, blood volume, and mean transit time, as well as the spatial distribution characteristics of these parameters;

[0191] c) Finally, morphological post-processing operations are applied to remove isolated small regions to obtain the final functional region partitioning result.

[0192] The specific steps include:

[0193] Apply an opening operation (a spherical structuring element with a radius of 2 voxels) to the binary mask of each category;

[0194] Apply a closing operation (a spherical structuring element with a radius of 3 voxels) to the processed mask;

[0195] Remove connected regions smaller than 50 voxels;

[0196] The partitioning result is visualized by color coding, as Figure 5 shown, where red represents the perfusion defect area, blue represents the reduced perfusion area, and green represents the healthy area.

[0197] In a specific embodiment, in order to further improve the diagnostic accuracy, especially for the differentiation of pulmonary vascular diseases, this embodiment further includes the following steps:

[0198] ROI segmentation:

[0199] Perform region of interest (ROI) segmentation on the whole lung imaging to obtain the segmented whole lung imaging containing 36 sub-regions. The specific segmentation method is as follows:

[0200] Divide the lungs into the left and right lungs. Each lung is equally divided into three regions: upper, middle, and lower, in the vertical direction. Each vertical region is equally divided into five regions: front, back, inner, outer, and center, in the horizontal direction. The center region is further subdivided into two regions: near the hilum and far from the hilum. In this way, each lung is divided into 18 sub-regions, and the whole lung has a total of 36 sub-regions. Perform image registration, functional information extraction, and region division on the segmented region of interest.

[0201] In one embodiment, when dividing the lung functional regions, the lungs include one or more of the following: left lung, right lung, left and right lungs.

[0202] In one embodiment, the overall process of the present invention is as Figure 5 shown. Obtain the dual-energy CT scan data of the lungs of the subject to be tested, including lung CT images and lung PBV images. Segment the lung CT images to obtain the lung regions (left and right lungs or left lung or right lung). Register the segmented CT images with the PBV to obtain the registered functional images (registered PBV images). Extract the functional information from the registered PBV images. Perform lung region division on the registered PBV images through the functional information to obtain the division result.

[0203] All steps of the automatic classification method of lung blood perfusion based on dual-energy CT in the present invention are executed by a computer for auxiliary diagnosis.

[0204] The disclosed embodiments of the present invention also provide a computer program product or system, including a computer program, which implements the steps of the above automatic classification method of lung blood perfusion based on dual-energy CT when executed by a processor.

[0205] Figure 2 A schematic diagram of an automatic classification system of lung blood perfusion based on dual-energy CT provided by an embodiment of the present invention specifically includes:

[0206] Acquisition module: Acquire the dual-energy CT scan data of the lungs of the subject to be tested, and the dual-energy CT scan data includes CT images and PBV images;

[0207] Segmentation module: Perform lung segmentation on the CT images to obtain the segmented CT images;

[0208] Registration module: Register the segmented CT images with the PBV images to obtain the registered functional images;

[0209] Function module: Perform function extraction based on the registered functional images to obtain functional information;

[0210] Partitioning module: The pulmonary function regions are partitioned based on the function information to obtain a partitioning result.

[0211] Figure 3 A schematic diagram of an automatic classification device for pulmonary blood perfusion based on dual - energy CT provided by an embodiment of the present invention specifically includes:

[0212] A memory and a processor; the memory is used for storing program instructions; the processor is used for calling the program instructions, and when the program instructions are executed, any one of the above - mentioned automatic classification methods for pulmonary blood perfusion based on dual - energy CT.

[0213] An embodiment of the present invention also provides a computer - readable storage medium. The computer - readable storage medium stores a computer program, and when the computer program is executed by a processor, any one of the above - mentioned automatic classification methods for pulmonary blood perfusion based on dual - energy CT.

[0214] An embodiment of the present invention also provides a method for evaluating pulmonary blood perfusion based on dual - energy CT, including:

[0215] Obtaining dual - energy CT scan data of the lungs of a person to be tested, where the dual - energy CT scan data includes CT images and PBV images;

[0216] Obtaining a partitioning result based on the above - mentioned automatic classification method for pulmonary blood perfusion based on dual - energy CT;

[0217] Evaluating the pulmonary blood perfusion based on the partitioning result to obtain an evaluation result.

[0218] In a specific embodiment, based on the partitioning result of the functional regions, a pulmonary perfusion evaluation report is generated to assist a doctor in making a pulmonary perfusion judgment. The report content includes:

[0219] a) Left and right lung volumes;

[0220] b) The volumes of each functional region (perfusion defect, perfusion reduction, healthy region) and their percentages of the whole lung;

[0221] c) The average blood flow and blood volume of each functional region;

[0222] d) 2D visualization images of the functional region distribution;

[0223] e) Comparison with normal reference values;

[0224] f) Longitudinal comparison (if there are previous examination data).

[0225] In another embodiment, based on the functional region division result and the feature analysis result, a lung perfusion assessment report is generated to assist doctors in making lung perfusion judgments. The report content includes:

[0226] a) The volumes of the left and right lungs;

[0227] b) The volumes of each functional region (perfusion defect, perfusion reduction, healthy region) and their percentages of the whole lung;

[0228] c) The average blood flow, blood volume, and mean transit time of each functional region;

[0229] d) 3D visualization images of the functional region distribution;

[0230] e) Comparison with normal reference values;

[0231] f) Longitudinal comparison (if there are previous examination data);

[0232] g) The disease classification result based on feature analysis and its confidence level;

[0233] h) Treatment suggestions: According to the classification result, specific treatment plans are recommended for any lung disease (including typical and atypical);

[0234] The report adopts a structured format, including numerical results, charts, and 2D visualization displays. Doctors are allowed to zoom in and slice to view the functional region distribution.

[0235] An embodiment of the present invention provides a lung blood perfusion assessment device based on dual-energy CT, including a memory, a processor, and a computer program or instruction stored on the memory. The computer program or instruction is executed by the processor to implement the above-mentioned lung blood perfusion assessment method based on dual-energy CT.

[0236] An embodiment of the present invention provides a computer program product, including a computer program or instruction. The computer program or instruction is executed by the processor to implement the above-mentioned lung blood perfusion assessment method based on dual-energy CT.

[0237] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program or instruction is stored. The computer program or instruction is executed by the processor to implement the above-mentioned lung blood perfusion assessment method based on dual-energy CT.

[0238] The verification results of this verification embodiment show that allocating fixed weights for indications can improve the performance of this method compared to the default settings. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein. In several embodiments provided by this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The storage medium can include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk, or optical disc, etc.

[0239] Those of ordinary skill in the art can understand that all or part of the steps in implementing the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The above-mentioned medium storage can be read-only memory, magnetic disk, or optical disc, etc.

[0240] The above has introduced in detail a computer device provided by the present invention. For those of ordinary skill in the art, according to the idea of the embodiments of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. An automatic classification method for pulmonary blood perfusion based on dual-energy CT, characterized in that, Including: Obtain the dual-energy CT scan data of the lungs of the subject to be tested, where the dual-energy CT scan data includes CT images and PBV images; Perform lung segmentation on the CT images to obtain segmented CT images; Register the segmented CT images with the PBV images to obtain registered functional images; Extract functions based on the registered functional images to obtain functional information; Perform lung function area division through the functional information to obtain a division result; the functional information includes local blood flow, and the division is obtained by comparing the local blood flow with a preset threshold to obtain the division result.

2. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, wherein The area division includes a perfusion defect area, a perfusion reduction area, and a healthy area.

3. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, wherein The division is replaced by inputting the functional information and functional images into a classifier for division to obtain a division result.

4. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 3, wherein The functional information includes local blood flow, blood volume, and mean transit time.

5. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 3, characterized in that, The process of division by the classifier is as follows: Obtain functional images and functional information; Perform feature transformation on the functional information to obtain functional features; Extract spatial distribution features based on the functional images; Input the functional features and spatial distribution features into a classifier for division to obtain a division result.

6. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 3, wherein, The process of division by the classifier also includes region elimination. After region elimination, a region result after removal is obtained, and an evaluation is performed based on the region result after removal; The region elimination is to perform morphological processing on the regions in the division result to remove isolated small regions to obtain a region result after removal.

7. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 3, wherein The classifier includes one or more of the following: random forest, decision tree, support vector machine, extreme learning machine, perceptron vector machine, convolutional neural network, residual network, dilated convolutional neural network.

8. The automatic classification method for pulmonary blood perfusion based on dual-energy CT according to claim 2, characterized in that The preset threshold of local blood flow in the perfusion defect area is less than that in the perfusion reduction area, and the preset threshold of local blood flow in the perfusion reduction area is less than that in the healthy area.

9. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 2, wherein When the local blood flow is less than 10 ml / 100 ml / min, it is determined as the perfusion defect area, and when the local blood flow is greater than or equal to 30 ml / 100 ml / min, it is determined as the healthy area.

10. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, wherein, The division is to first compare the local blood flow with a preset threshold to obtain a division area, and then input the division area and functional information into a classifier for fine division of the area to obtain a division result.

11. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, wherein The registration is performed by an algorithm combining rigidity and non-rigidity for registering the segmented CT images and PBV images; first, rough registration is performed by a rigid algorithm to obtain a rough registration result, and then local shape constraints are applied to the rough registration result by a non-rigid algorithm to obtain the registered functional images.

12. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 11, wherein The registration also includes fine adjustment. After local shape constraints are applied by a non-rigid algorithm, precise alignment of local regions is performed by a block matching algorithm to obtain the registered functional images.

13. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 11, wherein The rigid algorithm obtains a rough registration result by calculating the rigid transformation parameters between the segmented CT images and PBV images and performing rough registration based on the rigid transformation parameters.

14. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 11, characterized in that, The rigid algorithm further includes parameter optimization. The rigid transformation parameters are optimized by gradient descent to obtain optimized rigid transformation parameters, and rough registration is performed based on the optimized rigid transformation parameters to obtain a rough registration result.

15. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 11, wherein The non-rigid algorithm performs local deformation constraint through a B-spline deformation model.

16. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 12, characterized in that, The regions for fine adjustment include: the boundaries of lung lobes and vascular structures.

17. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, wherein The registered functional image is quantitatively evaluated by the structural similarity index and the mutual information value after registration. When the quantitative evaluation result is unqualified, registration or manual adjustment is performed again until the quantitative evaluation result is qualified.

18. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 1, characterized in that, The segmentation is performed through a pre-trained segmentation model to obtain a segmentation result, and the segmentation result includes the left lung and the right lung.

19. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 18, wherein The segmentation model includes one or more of the following: U-Net, FCN, SegNet, DeepLabV3+, Enet.

20. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 18, wherein The segmentation model includes an encoder and a decoder. The encoder consists of N residual blocks, where N is a natural number greater than 1. Input data extracts features through the N residual blocks connected in series in the encoder to obtain multi-scale features. The features of the Nth residual block are input into the decoder, and the multi-scale features of the first N-1 are input into the decoder through skip connections to fuse features of different scales to obtain decoded features. Then, weight distribution is performed on the decoded features through an attention mechanism to obtain important regions, and segmentation is performed on the important regions to obtain a segmentation result.

21. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 18, wherein The method further includes data preprocessing. The CT image is preprocessed to obtain a processed CT image, and the processed image is segmented to obtain a segmented CT image.

22. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 21, wherein The data preprocessing includes pixel normalization, size unification, image denoising, content highlighting, and data augmentation.

23. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 22, wherein, The content highlighting highlights the lung tissue through window width and window level.

24. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 22, wherein The image denoising removes image noise through Gaussian filtering.

25. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 18, wherein The segmentation further includes data postprocessing. The segmented CT image is postprocessed to obtain a postprocessed CT image, and the postprocessed CT image is registered with the PBV image; the postprocessing is to perform morphological processing on the segmented CT image to obtain a postprocessed CT image.

26. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 25, wherein The morphological processing includes opening and closing operations, removing isolated regions, and hole filling.

27. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 25, wherein The postprocessing further includes edge optimization. The edge of the CT image after morphological processing is optimized to obtain a postprocessed CT image.

28. The automatic classification method of pulmonary blood perfusion based on dual-energy CT according to claim 27, wherein, The edge optimization is completed through a conditional random field.

29. A method for evaluating pulmonary blood perfusion based on dual-energy CT, characterized in that, Obtain the dual-energy CT scan data of the lungs of the subject to be tested, and the dual-energy CT scan data includes CT images and PBV images; Obtain a classification result based on the dual-energy CT-based automatic lung blood perfusion classification method according to any one of claims 1-28; Perform a lung blood perfusion evaluation based on the classification result to obtain an evaluation result.

30. A computer product, comprising a computer program or instructions, characterized in that, The computer program or instruction is executed by a processor to implement the dual-energy CT-based automatic lung blood perfusion classification method according to any one of claims 1-28 or execute the dual-energy CT-based lung blood perfusion evaluation method according to claim 29.

31. A computer device, comprising a memory, a processor, and a computer program or instruction stored on the memory, characterized in that, The computer program or instruction, when executed by a processor, implements the automatic classification method of pulmonary blood perfusion based on dual-energy CT according to any one of claims 1-28 or implements the evaluation method of pulmonary blood perfusion based on dual-energy CT according to claim 29.

32. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, The computer program or instruction, when executed by a processor, implements the automatic classification method of pulmonary blood perfusion based on dual-energy CT according to any one of claims 1-28 or implements the evaluation method of pulmonary blood perfusion based on dual-energy CT according to claim 29.

Images