Heart image processing method and device, storage medium and computer equipment

Abstract

The invention discloses a heart image processing method and device, a storage medium and computer equipment, and relates to the technical field of nuclear medicine image processing and quantitative analysis. The method comprises the steps that a heart image to be processed is acquired, four cavities of the heart image are segmented based on a pre-trained segmentation model, segmentation masks are obtained, and the heart image is a CT image or an MR image; aligning the heart image into a PET space, so that the segmentation mask is mapped to a PET image corresponding to the heart image, and a PET segmentation image is obtained; and cutting the PET segmented image based on the segmentation mask to obtain a PET target anatomical region, and performing three-dimensional rendering on the PET target anatomical region to obtain a dynamic rendered image of tracer agent distribution. According to the method, the segmentation precision of the myocardial subregion can be improved, and the three-dimensional visualization effect of the heart image and the processing efficiency of the heart image are improved.

Description

Technical Field

[0001] This application relates to the field of nuclear medicine image processing and quantitative analysis technology, and in particular to a method, apparatus, storage medium and computer equipment for processing cardiac images. Background Technology

[0002] In the field of nuclear medicine imaging, positron emission tomography (PET) and computed tomography (CT) fusion imaging technology has become a core tool for cardiac function assessment and metabolic analysis. This technology, through the uptake characteristics of radioactive tracers, can quantitatively reflect the metabolic activity of myocardial tissue, thereby providing data support for the assessment of cardiovascular diseases.

[0003] Currently, PET / CT cardiac imaging technology primarily relies on static image analysis. This involves first roughly segmenting the overall cardiac contour using semi-automatic or manual methods, and then employing 2D tomographic image overlay or basic volume rendering techniques to achieve 3D visualization of the static image. However, these imaging methods only support coarse assessments of the entire heart or large areas, lacking the ability to precisely segment and quantify myocardial subregions such as the atria and ventricular walls. Furthermore, the 3D visualization achieved based on static images yields poor results. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method, apparatus, storage medium, and computer device for processing cardiac images, with the main purpose of solving the technical problems of insufficient myocardial subregion segmentation and quantization capabilities in cardiac images, and poor three-dimensional visualization effects based on static images.

[0005] According to one aspect of this application, a method for processing cardiac images is provided, the method comprising:

[0006] A cardiac image to be processed is acquired, and based on a pre-trained segmentation model, the four chambers of the cardiac image are segmented to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image;

[0007] The cardiac image is aligned to PET space so that the segmentation mask is mapped onto the PET image corresponding to the cardiac image, thereby obtaining a PET segmentation image;

[0008] The PET segmentation image is cropped based on the segmentation mask to obtain the PET target anatomical region. The PET target anatomical region is then rendered in three dimensions to obtain a dynamic rendering image of the tracer distribution.

[0009] According to another aspect of this application, a cardiac image processing apparatus is provided, the apparatus comprising:

[0010] The image segmentation module is used to acquire the cardiac image to be processed and segment the four chambers of the cardiac image based on a pre-trained segmentation model to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image.

[0011] The image mapping module is used to align the cardiac image to the PET space so that the segmentation mask is mapped onto the PET image corresponding to the cardiac image to obtain a PET segmentation image;

[0012] The image rendering module is used to crop the PET segmentation image based on the segmentation mask to obtain the PET target anatomical region, and to perform three-dimensional rendering of the PET target anatomical region to obtain a dynamic rendering image of the tracer distribution.

[0013] According to another aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described method for processing cardiac images.

[0014] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described method for processing cardiac images.

[0015] By employing the above technical solutions, the cardiac image processing method, apparatus, storage medium, and computer equipment provided in this application, through a pre-trained segmentation model, segment the four chambers of a cardiac image, achieving precise segmentation and quantification of myocardial subregions such as the four chambers. Simultaneously, by aligning the cardiac image to PET space, a segmentation mask is mapped onto the PET image. Then, based on the segmentation mask, the PET segmented image is cropped to obtain the PET target anatomical region. Finally, 3D rendering is performed based on the PET target anatomical region to obtain a dynamically rendered image of tracer distribution, which can dynamically and intuitively present the spatial distribution characteristics of the tracer, thereby enhancing the 3D visualization effect of the cardiac image. Furthermore, by integrating multiple image processing steps onto the same platform, the above method can effectively simplify the processing flow of cross-modal data, thereby reducing manual intervention and improving the processing efficiency of cardiac images.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 A schematic flowchart of a method for processing cardiac images provided in an embodiment of this application is shown;

[0019] Figure 2 A schematic flowchart of a method for processing cardiac images provided in an embodiment of this application is shown;

[0020] Figure 3 This illustration shows a scenario diagram of a cardiac image processing method provided in an embodiment of this application;

[0021] Figure 4 A schematic diagram of a cardiac image processing device provided in an embodiment of this application is shown. Detailed Implementation

[0022] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0023] In one embodiment, refer to Figure 1 and Figure 3 This paper provides a method for processing cardiac images, and illustrates the method using a computer device as an example. The method includes the following steps:

[0024] Step 101: Obtain the cardiac image to be processed, and segment the four chambers of the cardiac image based on the pre-trained segmentation model to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image.

[0025] The PET / CT or PET / MR equipment will obtain two modal images after capturing images: CT or MR modal images and PET modal images. Here, the cardiac image to be processed refers to the CT or MR modal image, i.e., CT image or MR image. The PET image in step 102 refers to the PET modal image corresponding to the CT image or MR image.

[0026] Furthermore, the acquired cardiac images to be processed are typically in DICOM format, which contains images and metadata (such as patient information, equipment parameters, scan order, etc.), but its structure is redundant and unsuitable for direct input into neural network models. Therefore, after acquiring CT or MR images, they can be converted to NIfTI format to facilitate batch processing, 3D array operations, tensor input, and other operations, and voxel-level alignment and resampling can be achieved through affine matrices.

[0027] Specifically, after acquiring cardiac images to be processed, computer equipment can perform format conversion. For example, the computer can read cardiac image data from DICOM sequences using an open-source image processing library, and then convert the cardiac images to NIfTI format through metadata parsing to ensure uniform voxel sizes. Further, after format conversion, a pre-trained segmentation model can be used to segment the four chambers (left atrium, right atrium, left ventricle, and right ventricle) of the cardiac image. For example, the computer can input NIfTI format image data into the TotalSegmentator model to output a segmentation mask for the four chambers. Through this method, high-precision identification of the four cardiac chambers and other myocardial subregions can be achieved, thereby reducing manual intervention and improving image processing efficiency and accuracy.

[0028] Step 102: Align the cardiac image into the PET space so that the segmentation mask is mapped onto the corresponding PET image to obtain the PET segmentation image.

[0029] Specifically, after segmenting the cardiac image to obtain a segmentation mask, the cardiac image can be aligned to the PET space so that the segmentation mask is mapped onto the corresponding PET image, thus obtaining a precisely segmented PET image. During the alignment process, the computer equipment can perform affine registration on the cardiac image, combined with B-spline nonlinear transformation, and then use nearest neighbor interpolation to preserve the anatomical structures in the PET image, thereby obtaining a precisely segmented PET image.

[0030] Step 103: Based on the segmentation mask, the PET segmentation image is cropped to obtain the PET target anatomical region, and the PET target anatomical region is rendered in three dimensions to obtain a dynamic rendering image of the tracer distribution.

[0031] Specifically, after obtaining the segmentation mask and PET segmentation images, the PET segmentation images can be cropped based on the segmentation mask to obtain the PET target anatomical region focusing on the cardiac subregion. The PET target anatomical region is the cropped 3D data volume. Based on the voxel values ​​of the PET target anatomical region, a 3D uniform mesh can be constructed, and the PET intensity can be mapped to a scalar field, such as mapping SUV values ​​to a color gradient. This allows for the generation of a 360° rotating animation sequence through rendering, which is then compiled into an MP4 video image to visually demonstrate the tracer distribution. This method can intuitively visualize spatial acquisition patterns and reduce manual rendering time.

[0032] By applying the technical solution of this embodiment, the four chambers of a heart image are segmented using a pre-trained segmentation model, achieving accurate segmentation and quantification of myocardial subregions such as the four chambers. Simultaneously, by aligning the heart image to PET space, a segmentation mask is mapped onto the PET image. The segmented PET image is then cropped based on the segmentation mask to obtain the target anatomical region. Finally, 3D rendering is performed based on the target anatomical region to obtain a dynamically rendered image of the tracer distribution. This dynamically and intuitively presents the spatial distribution characteristics of the tracer, thereby enhancing the 3D visualization effect of the heart image. Furthermore, by integrating multiple image processing steps onto a single platform, the above method effectively simplifies the processing flow of cross-modal data, reducing manual intervention and improving the processing efficiency of heart images.

[0033] In one embodiment, the method of aligning the cardiac image to PET space in step 102 can be implemented through the following steps: First, using the PET image as the reference image, the cardiac image is affine registered as a floating image to align the position and orientation of the PET image and the cardiac image. Then, B-spline registration is performed on the aligned PET image and the cardiac image to adjust the deformation of the cardiac image. Finally, the position of each pixel in the cardiac image in the PET image is calculated, and the registered pixel coordinates of the cardiac image are calculated based on the position of each pixel in the PET image. The PET segmentation image is obtained based on the registered pixel coordinates.

[0034] Specifically, during registration, the corresponding PET image for the CT or MR image can be used as the reference image. The CT or MR image is treated as a floating image and affine registered. Then, by calculating rigid transformation parameters such as translation, rotation, scaling, and shearing, the CT or MR image and the PET image are initially aligned in position, orientation, and scale. Subsequently, based on the aligned image pair, a B-spline free deformation registration method can be used. This involves constructing a control point grid and optimizing the deformation field to adjust the local non-rigid deformation of the cardiac image, making it more closely match the anatomical structure of the PET image. Finally, nearest neighbor interpolation can be used to calculate the corresponding coordinates of each pixel in the cardiac image in the PET image to generate the registered PET segmentation image. The above embodiment, through affine registration, can quickly correct global geometric differences. By combining B-spline registration and nearest neighbor interpolation, local deformation can be finely adjusted, ensuring the fidelity of the anatomical structure in the PET space for the segmentation results, thereby achieving a high-precision mapping of CT / MR segmentation results to PET space.

[0035] In one embodiment, step 104 can be implemented by the following method: First, Boolean operations are performed on the PET segmentation image and the segmentation mask, and the voxel values ​​in the segmentation mask are retained to obtain the PET target anatomical region. Then, the coordinates of each voxel value in the PET target anatomical region are read, and a three-dimensional uniform grid is constructed based on the coordinates of each voxel value. Further, the SUV values ​​in the PET target anatomical region are normalized, and the normalized SUV values ​​are mapped to each grid point of the three-dimensional uniform grid. Finally, the mapped three-dimensional uniform grid is rendered from multiple angles to generate a rotation animation frame sequence, and the rotation animation frame sequence is compiled into a video image and output.

[0036] Specifically, during dynamic image rendering, a Boolean intersection operation is first performed on the PET segmentation image and the segmentation mask. Then, a logical judgment is performed on a voxel-by-voxel basis to retain voxel regions with non-zero values ​​in the segmentation mask, thereby extracting the target anatomical region. Subsequently, the 3D coordinates (x, y, z) of all voxels within this region are read, and a uniformly spaced 3D mesh structure is constructed based on the coordinate distribution range. The mesh cell size can be dynamically adjusted according to the maximum spatial span of the target anatomical region to ensure coverage of all voxel positions. Next, the SUV values ​​of the target region are standardized using a min-max normalization method, mapping the original SUV values ​​to a preset interval (e.g., [0, 1]). Then, the normalized SUV values ​​are assigned to the corresponding mesh points in the 3D mesh using a nearest neighbor interpolation algorithm. Finally, multi-view ray tracing rendering technology is used to perform a 360° rotation projection on the mesh model to generate a continuous frame sequence containing front, side, and axonometric views. This continuous frame sequence is then compressed into video images in MP4 or other formats, and dynamically annotated information is overlaid before output.

[0037] The above embodiments, through Boolean operations, can accurately extract the target anatomical region. By combining uniform mesh modeling and SUV normalization, the geometric accuracy and numerical consistency of the 3D image reconstruction can be significantly improved. Simultaneously, the rotational animation generated by multi-angle rendering can intuitively demonstrate the spatial characteristics of metabolic distribution, thus providing standardized visualization output for clinical practice.

[0038] In one embodiment, refer to Figure 2 and Figure 3 The above-mentioned method for processing cardiac images also includes the following steps:

[0039] Step 201: Based on the PET images corresponding to the cardiac images, the blood pool regions of the four chambers are marked, and the mean SUV and standard deviation of the SUV of the blood pool regions of the four cardiac chambers are calculated respectively.

[0040] Step 202: Based on the mean SUV and standard deviation of the SUV in the blood pool region of the four cardiac chambers, the blood pool calculation threshold of the four cardiac chambers is obtained.

[0041] Step 203: Calculate the threshold based on the blood pools of the four cardiac chambers, segment the myocardial regions of the four chambers in the segmentation mask, and obtain the myocardial segmentation mask.

[0042] Step 204: Perform quantitative analysis based on the myocardial segmentation mask to obtain the quantitative analysis results of the cardiac images.

[0043] Specifically, in quantitative analysis, the blood pool regions of the four cardiac chambers can first be manually or model-assistedly labeled in PET images. Then, based on the labeled blood pool regions, the mean (SUV_{mean(blood pool)}) and standard deviation (SD_{blood pool}) of the standardized uptake value (SUV) for each cardiac chamber can be calculated. The mean SUV can be obtained through global intensity histogram analysis, and the standard deviation SUV can be calculated through local neighborhood statistics to reflect the metabolic heterogeneity of the blood pool regions. Next, a pre-defined dynamic threshold calculation model (such as a linear combination formula based on the mean and standard deviation of the blood pool SUV) can be used, and the calculation weights can be adjusted according to the metabolic characteristics of different cardiac chambers to generate an adaptive blood pool calculation threshold. Furthermore, after obtaining the blood pool calculation threshold for each cardiac chamber, the blood pool calculation threshold can be used as the segmentation basis to segment the myocardial region of the four chambers in the segmentation mask to obtain the myocardial segmentation mask. Finally, quantitative analysis can be performed based on the myocardial segmentation mask to obtain the quantitative analysis results of cardiac images.

[0044] The above embodiments, through a dynamic threshold calculation model, can accurately identify myocardial regions, thereby achieving adaptive segmentation at the cardiac chamber level and avoiding misjudgments caused by fixed thresholds. Simultaneously, by combining blood pool statistics, the accuracy of quantitative indicator calculations can be improved, thus avoiding the need to calculate various indicator data across multiple platforms, thereby reducing the proportion of manual intervention and improving image analysis efficiency.

[0045] In one embodiment, the blood pool calculation threshold of the four cardiac chambers in step 202 can be calculated by the following method: First, obtain the first weight coefficient and the second weight coefficient corresponding to the four cardiac chambers respectively. Then, for each cardiac chamber, sum the product of the first weight coefficient of the cardiac chamber and the mean SUV of the blood pool region of the cardiac chamber and the product of the second weight coefficient of the cardiac chamber and the standard deviation of the SUV of the blood pool region of the cardiac chamber to obtain the blood pool calculation threshold of the cardiac chamber.

[0046] Specifically, when calculating the blood pool threshold for the four cardiac chambers (left atrium, right atrium, left ventricle, and right ventricle), the first and second weighting coefficients (w1 and w2) for each chamber are first obtained. Since the atria have lower metabolic activity, the threshold can be lowered to avoid missing low-grade abnormalities; for example, the weighting coefficients for the atrial chambers can be set to w1 = 1.25 and w2 = 1. In contrast, the ventricles have higher metabolic activity, so the weighting coefficients can be relatively higher to suppress background noise interference in the blood pool and improve specificity; for example, the weighting coefficients for the ventricular chambers can be set to w1 = 1.5 and w2 = 1. Subsequently, for each chamber, the dynamic threshold (cutoff) can be calculated based on the normalized uptake value (SUV) mean (SUV_{mean(blood pool)}) and SUV standard deviation (SD_{blood pool}) of its blood pool region using the dynamic threshold calculation model "cutoff = w1 × SUV_{mean(blood pool)} + w2 × SD_{blood pool}". The mean SUV can be obtained through global intensity histogram analysis, and the standard deviation of the SUV can be calculated through local neighborhood statistics to reflect the metabolic heterogeneity of the blood pool. Finally, the calculated dynamic threshold can be used as the segmentation basis to identify myocardial regions above the dynamic threshold, and a myocardial segmentation mask can be generated through binarization processing.

[0047] The above embodiments differentiate the weight coefficients in the model by combining the metabolic distribution characteristics of different cardiac chambers, which can accurately segment the myocardial regions of different cardiac chambers, thereby avoiding misjudgment caused by fixed thresholds and improving the accuracy of myocardial region identification.

[0048] In one embodiment, the method for obtaining quantitative analysis results based on myocardial segmentation mask in step 204 can be implemented through the following steps: First, based on myocardial segmentation mask, calculate the SUV index and metabolic tissue volume of the four cardiac chambers, wherein the SUV index includes myocardial metabolic intensity index and metabolic blood flow ratio index. Then, based on the SUV index, metabolic tissue volume and blood pool of the four cardiac chambers, calculate the threshold, generate the quantitative analysis results of cardiac images, and output the quantitative analysis results.

[0049] Specifically, for each heart chamber in the myocardial segmentation mask, myocardial metabolic intensity indicators (including the mean and maximum SUV values ​​of the myocardial region within the chamber) and metabolic blood flow ratio indicators (i.e., the ratio between the maximum SUV value of the myocardial region and the mean SUV value of the blood pool region within the chamber) can be calculated chamber by chamber. Simultaneously, the metabolic tissue volume (MTV) of the heart chamber can be calculated using three-dimensional voxel counting. The mean SUV value of the myocardial region can be obtained by averaging the SUV values ​​of all voxels in the myocardial region, the maximum SUV value of the myocardial region is the largest SUV value within the myocardial region, and the MTV can be obtained by multiplying the number of voxels in the binary mask by the voxel volume. Finally, the SUV indicators, MTV indicators, and dynamic threshold cutoff of the four heart chambers (LA / LV / RA / RV) can be integrated into a structured data table. A visualization module is used to generate bar charts containing each indicator, as well as quantitative analysis results including a comparison of myocardial metabolic intensity indicators for each chamber, a heatmap showing the distribution of the metabolic blood flow ratio indicator, and a pie chart showing the metabolic tissue volume. The quantitative analysis results are referenced... Figure 3 As shown.

[0050] The above embodiments achieve automated integration and report generation of cardiac-level SUV and MTV indicators through a single platform, enabling quantitative analysis of cardiac-level SUV and MTV indicators. This shortens the image analysis process, reduces the need for manual intervention, and avoids the inefficiency caused by traditional multi-platform decentralized operations.

[0051] In one embodiment, the SUV index and metabolic tissue volume of the four cardiac chambers can be calculated as follows: For each cardiac chamber in the myocardial segmentation mask, firstly, the mean and maximum SUV values ​​of the myocardial region of the cardiac chamber are calculated to obtain the myocardial metabolic intensity index of the cardiac chamber. Then, based on the ratio between the maximum SUV value of the myocardial region of the cardiac chamber and the mean SUV value of the blood pool region of the cardiac chamber, the metabolic blood flow ratio index of the cardiac chamber is obtained. Finally, based on the myocardial segmentation mask, the number of voxels in the myocardial region of the cardiac chamber is counted, and based on the number of voxels and the voxel volume, the metabolic tissue volume of the cardiac chamber is obtained.

[0052] Specifically, based on the myocardial segmentation mask, the normalized mean uptake value (i.e., the SUV mean of the myocardial region, denoted as SUVmean(myocardial)) and the maximum normalized uptake value (i.e., the maximum SUV value of the myocardial region, denoted as SUVmax(myocardial)) of the myocardial region in each ventricle can be calculated through global intensity histogram analysis. The SUV mean and maximum SUV value of the myocardial region are then used as indicators of myocardial metabolic intensity. The SUV mean of the myocardial region can be obtained by averaging the SUV values ​​of all high-metabolic voxels, and the maximum SUV value of the myocardial region can be the maximum SUV value of a single voxel within the myocardium. Next, the metabolic blood flow ratio (denoted as SUVr(myocardial)) can be calculated using the formula SUVr(myocardial) = SUVmax(myocardial) / SUVmean(myocardial) to characterize the difference in metabolic activity between the myocardium and the blood pool, by combining the SUV mean (denoted as SUVmean(blood pool)) of the blood pool region within the same ventricle. Finally, the number of voxels in the myocardial region can be counted using the binarized segmentation mask and multiplied by the voxel volume (which can be calculated from the spatial resolution of the PET segmentation image) to obtain the metabolic tissue volume (MTV).

[0053] The above embodiments quantify indicators such as myocardial metabolic intensity, metabolic blood flow ratio, and metabolic tissue volume, and integrate them into a structured data table. This can effectively improve the calculation efficiency of cardiac chamber-level metabolic parameters and achieve accurate characterization of cardiac chamber-level metabolic features, thereby providing data support for the assessment of cardiovascular diseases.

[0054] Furthermore, as Figures 1 to 3 In terms of specific implementation, this application provides a cardiac image processing apparatus, such as... Figure 4 As shown, the device includes:

[0055] The image segmentation module 31 is used to acquire the cardiac image to be processed and segment the four chambers of the cardiac image based on a pre-trained segmentation model to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image.

[0056] Image mapping module 32 is used to align the cardiac image to PET space so that the segmentation mask is mapped onto the PET image corresponding to the cardiac image to obtain a PET segmentation image;

[0057] The image rendering module 33 is used to crop the PET segmentation image based on the segmentation mask to obtain the PET target anatomical region, and to perform three-dimensional rendering of the PET target anatomical region to obtain a dynamic rendering image of the tracer distribution.

[0058] In a specific application scenario, the image mapping module 32 is specifically used to perform affine registration of the heart image as a floating image, using the PET image as a reference image, to align the position and orientation of the PET image and the heart image; perform B-spline registration on the aligned PET image and the heart image to adjust the deformation of the heart image; calculate the position of each pixel in the heart image in the PET image, and calculate the registered pixel coordinates of the heart image based on the position of each pixel in the PET image, and obtain the PET segmented image based on the registered pixel coordinates.

[0059] In a specific application scenario, the image rendering module 33 can be used to perform Boolean operations on the PET segmentation image and the segmentation mask, and retain the voxel values ​​in the segmentation mask to obtain the PET target anatomical region; read the coordinates of each voxel value in the PET target anatomical region, and construct a three-dimensional uniform mesh based on the coordinates of each voxel value; normalize the SUV values ​​in the PET target anatomical region, and map the normalized SUV values ​​to each grid point of the three-dimensional uniform mesh; perform multi-angle image rendering on the mapped three-dimensional uniform mesh to generate a rotation animation frame sequence, and compile the rotation animation frame sequence into a video image for output.

[0060] In specific application scenarios, the device further includes a quantitative analysis module 34, which can be used to annotate the blood pool regions of the four chambers based on the PET images corresponding to the cardiac images, and calculate the mean SUV and standard deviation of the blood pool regions of the four cardiac chambers respectively; obtain the blood pool calculation threshold of the four cardiac chambers based on the mean SUV and standard deviation of the blood pool regions of the four cardiac chambers; segment the myocardial regions of the four chambers in the segmentation mask based on the blood pool calculation threshold of the four cardiac chambers to obtain the myocardial segmentation mask; and perform quantitative analysis based on the myocardial segmentation mask to obtain the quantitative analysis results of the cardiac images.

[0061] In specific application scenarios, the quantitative analysis module 34 can also be used to obtain the first weighting coefficient and the second weighting coefficient corresponding to the four heart chambers respectively; for each heart chamber, the product of the first weighting coefficient of the heart chamber and the mean SUV of the blood pool region of the heart chamber and the product of the second weighting coefficient of the heart chamber and the standard deviation of the SUV of the blood pool region of the heart chamber are summed to obtain the blood pool calculation threshold of the heart chamber.

[0062] In specific application scenarios, the quantitative analysis module 34 can also be used to calculate the SUV index and metabolic tissue volume of the four cardiac chambers based on the myocardial segmentation mask, wherein the SUV index includes the myocardial metabolic intensity index and the metabolic blood flow ratio index; based on the SUV index, metabolic tissue volume and blood pool calculation threshold of the four cardiac chambers, the quantitative analysis results of the cardiac image are generated and the quantitative analysis results are output.

[0063] In specific application scenarios, the quantitative analysis module 34 can also be used to calculate the mean and maximum SUV values ​​of the myocardial region of each ventricle in the myocardial segmentation mask to obtain the myocardial metabolic intensity index of the ventricle; obtain the metabolic blood flow ratio index of the ventricle based on the ratio between the maximum SUV value of the myocardial region of the ventricle and the mean SUV value of the blood pool region of the ventricle; and count the number of voxels in the myocardial region of the ventricle based on the myocardial segmentation mask, and obtain the metabolic tissue volume of the ventricle based on the number and volume of voxels.

[0064] It should be noted that other corresponding descriptions of the functional units involved in the cardiac image processing apparatus provided in this application embodiment can be found by referring to... Figures 1 to 3 The corresponding descriptions in the method will not be repeated here.

[0065] This application also provides a computer device, specifically a personal computer, server, network device, etc. The computer device includes a bus, processor, memory, and communication interface, and may also include input / output interfaces and a display device. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores location information. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the various method embodiments.

[0066] Those skilled in the art will understand that the structure of the computer device described above is only a partial structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. A specific computer device may include more or fewer components, or combine certain components, or have different component arrangements.

[0067] In one embodiment, a computer-readable storage medium is provided, which may be non-volatile or volatile, having stored thereon a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0068] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0069] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0070] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0071] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for processing cardiac images, characterized in that, The method includes: A cardiac image to be processed is acquired, and based on a pre-trained segmentation model, the four chambers of the cardiac image are segmented to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image; The cardiac image is aligned to PET space so that the segmentation mask is mapped onto the PET image corresponding to the cardiac image, thereby obtaining a PET segmentation image; The PET segmentation image is cropped based on the segmentation mask to obtain the PET target anatomical region. The PET target anatomical region is then rendered in three dimensions to obtain a dynamic rendering image of the tracer distribution.

2. The method according to claim 1, characterized in that, Aligning the cardiac image to PET space so that the segmentation mask is mapped onto the corresponding PET image to obtain a PET segmentation image includes: Using the PET image as a reference image, the cardiac image is used as a floating image for affine registration to align the position and orientation of the PET image and the cardiac image; B-spline registration was performed on the aligned PET image and cardiac image to adjust the deformation of the cardiac image; The position of each pixel in the cardiac image is calculated in the PET image, and the registered pixel coordinates of the cardiac image are calculated based on the position of each pixel in the PET image. The PET segmentation image is obtained based on the registered pixel coordinates.

3. The method according to claim 1, characterized in that, The process of cropping the PET segmented image based on the segmentation mask to obtain the PET target anatomical region, and performing three-dimensional rendering on the PET target anatomical region to obtain a dynamic rendering image of tracer distribution includes: Boolean operations are performed on the PET segmented image and the segmentation mask, and the voxel values ​​in the segmentation mask are retained to obtain the PET target anatomical region; The coordinates of each voxel value of the PET target anatomical region are read, and a three-dimensional uniform mesh is constructed based on the coordinates of each voxel value; The SUV values ​​within the PET target anatomical region are normalized, and the normalized SUV values ​​are mapped to each grid point of the three-dimensional uniform grid. The mapped 3D uniform mesh is rendered from multiple angles to generate a sequence of rotating animation frames, which are then compiled into video images for output.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Based on the PET images corresponding to the cardiac images, the blood pool regions of the four chambers are marked, and the mean SUV and standard deviation of the SUV of the blood pool regions of the four cardiac chambers are calculated respectively. Based on the mean and standard deviation of the SUV in the blood pool region of the four cardiac chambers, the blood pool calculation thresholds for the four cardiac chambers are obtained. Based on the blood pools of the four cardiac chambers, thresholds are calculated to segment the myocardial regions of the four chambers in the segmentation mask, thus obtaining the myocardial segmentation mask; Quantitative analysis was performed based on the myocardial segmentation mask to obtain the quantitative analysis results of the cardiac images.

5. The method according to claim 4, characterized in that, The blood pool calculation thresholds for the four cardiac chambers are obtained based on the mean and standard deviation of the SUV values ​​of the blood pool regions of the four cardiac chambers, including: Obtain the first and second weighting coefficients corresponding to the four cardiac chambers respectively; For each heart chamber, the product of the first weighting coefficient of the heart chamber and the mean SUV of the blood pool region of the heart chamber is summed with the product of the second weighting coefficient of the heart chamber and the standard deviation of the SUV of the blood pool region of the heart chamber to obtain the blood pool calculation threshold of the heart chamber.

6. The method according to claim 4, characterized in that, The quantitative analysis based on the myocardial segmentation mask to obtain the quantitative analysis results of the cardiac image includes: Based on the myocardial segmentation mask, the SUV index and metabolic tissue volume of the four cardiac chambers are calculated, wherein the SUV index includes the myocardial metabolic intensity index and the metabolic blood flow ratio index. Based on the SUV index, metabolic tissue volume, and blood pool calculation threshold of the four cardiac chambers, quantitative analysis results of the cardiac images are generated and output.

7. The method according to claim 6, characterized in that, The calculation of the SUV index and metabolic tissue volume of the four cardiac chambers based on the myocardial segmentation mask includes: For each ventricle in the myocardial segmentation mask, the mean and maximum SUV values ​​of the myocardial region of the ventricle are calculated to obtain the myocardial metabolic intensity index of the ventricle. The metabolic blood flow ratio index of the heart chamber is obtained by the ratio between the maximum SUV value of the myocardial region of the heart chamber and the average SUV value of the blood pool region of the heart chamber. Based on the myocardial segmentation mask, the number of voxels in the myocardial region of the cardiac chamber is counted, and the metabolic tissue volume of the cardiac chamber is obtained based on the number and volume of voxels.

8. A cardiac image processing device, characterized in that, The device includes: The image segmentation module is used to acquire the cardiac image to be processed and segment the four chambers of the cardiac image based on a pre-trained segmentation model to obtain a segmentation mask, wherein the cardiac image is a CT image or an MR image. The image mapping module is used to align the cardiac image to the PET space so that the segmentation mask is mapped onto the PET image corresponding to the cardiac image to obtain a PET segmentation image; The image rendering module is used to crop the PET segmentation image based on the segmentation mask to obtain the PET target anatomical region, and to perform three-dimensional rendering of the PET target anatomical region to obtain a dynamic rendering image of the tracer distribution.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

Images