Method and system for acquiring cardiac muscle CT value of left ventricle segment
The 3DnnUNet model and 16-segment model segmentation and partitioning of cardiac CT delay enhancement images is solved, and the problem of insufficient segmentation accuracy and consistency in myocardial CT value measurement is achieved, achieving more accurate CT value calculation and more efficient diagnostic support .
Patent Information
- Application Number
- CN202410990694.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-23
AI Technical Summary
The prior art lacks segmentation accuracy and consistency in the measurement of myocardial CT value, making it difficult to accurately divide the apical, papillary muscle, cardiac basal and other areas, resulting in inaccurate calculation of CT value, affecting the accuracy of clinical diagnosis and treatment.
The 3DnnUNet model was used to segment the cardiac CT delay enhancement images, combined with the American Heart Association's 16-segment model, the left ventricular myocardial area was divided in detail, and the image quality was improved through image preprocessing, and CT values were automatically calculated.
It improves the accuracy and consistency of myocardial region segmentation, provides more accurate CT values, reduces artificial errors, saves time, and improves the efficiency of diagnosis and treatment.
Smart Images

Figure BDA0004958787980000101 
Figure HDA0004958788010000011 
Figure HDA0004958788010000012
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of myocardial CT value measurement, and in particular to a method and system for acquiring left ventricular segment myocardial CT values. Background Art
[0002] As the incidence of cardiovascular disease continues to rise, early diagnosis and precise treatment of heart disease have become particularly important. Cardiac CT (computed tomography) imaging technology, with its high resolution and rapid scanning characteristics, has become a key tool for evaluating cardiac structure and function. In particular, cardiac CT delayed enhancement (CT-LE) imaging technology can clearly show myocardial fibrosis and the extent of myocardial infarction by injecting contrast agents, thereby more clearly showing the extent of myocardial damage.
[0003] The 17-segment model and the improved 16-segment model proposed by the American Heart Association (AHA) are important division methods for standardized assessment of accurate positioning of myocardial injury and overall and segmental function of the heart in clinical and scientific research. The 16-segment model divides the left ventricle (LV) into 16 anatomical segments for detailed assessment of the function and pathological conditions of each segment. Specifically, the left ventricle is divided vertically into three equal parts along the long axis of the heart, generating three circular parts of the left ventricle: the basal plane, the papillary muscle plane, and the apical plane. Only left ventricular myocardial slices that include all 360° are taken into consideration. The basal image is further divided into six 60° segments, named: basal anterior wall, basal anterior septum, basal inferior septum, basal inferior wall, basal inferior lateral wall, and basal segment anterior lateral wall. Similarly, the papillary muscle image is also divided into six 60° segments, called the middle anterior wall, middle anterior septum, middle inferior septum, middle inferior wall, middle inferior lateral wall, and middle anterior lateral wall. Because the left ventricle is conical, the apical image uses only four 90° segments, named the apical anterior wall, apical septum, apical inferior wall, and apical lateral wall. The apical cap represents the true muscle at the extreme tip of the ventricle, where the chamber no longer exists, and this section is called the apex.
[0004] Accurately assessing the location and extent of left ventricular myocardial damage is of great significance for the diagnosis and treatment of cardiovascular diseases. However, the existing technology still has the following shortcomings: First, the accuracy and consistency of the manual segmentation method are insufficient. Traditional myocardial region segmentation mainly relies on manual delineation, but due to the low contrast between the blood pool and the myocardium in the delayed enhancement image of cardiac CT, and the presence of interference factors such as trabeculae and papillary muscles in the endocardium, manual segmentation is time-consuming and prone to subjective errors, and the accuracy and consistency of the obtained segmentation results are low. Secondly, the segmentation accuracy is not high. The existing segmentation method is difficult to accurately divide the myocardial regions at different levels such as the apex, papillary muscle, and cardiac base, resulting in rough segmentation results and poor consistency, and unable to provide stable and reliable CT value calculation results, which may affect the clinician's accurate assessment of the condition and prognosis. Finally, the CT value calculation is inaccurate. In the prior art, the conversion between pixel values and CT values often depends on the automatic processing of the equipment. However, due to the poor signal-to-noise ratio and image quality of delayed enhancement images of cardiac CT, if the equipment fails to accurately convert, manual calculation is required, which further increases the workload and error risk.
[0005] In recent years, deep learning technology, especially convolutional neural networks (CNN), has made significant progress in the field of medical image segmentation. Among them, U-Net and its variants have become the mainstream methods for medical image segmentation. 3D nnU-Net, as a three-dimensional network architecture based on U-Net, further improves the accuracy and robustness of medical image segmentation. Summary of the invention
[0006] The purpose of the present invention is to propose a method and system for acquiring left ventricular segment myocardial CT values, which can provide more objective and accurate myocardial CT value measurement results and provide reliable support for clinical diagnosis and treatment.
[0007] According to a first aspect of an embodiment of the present disclosure, a method for acquiring a left ventricular segment myocardial CT value is provided, comprising the following steps:
[0008] Perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0009] Preprocessing the cardiac CT delayed enhancement image, including image normalization and noise removal, to improve image quality;
[0010] The 3DnnUNet model is used to segment the preprocessed delayed enhancement CT images of the heart, and the segmentation results of the left ventricular myocardium area are obtained; 3DnnUNet is a segmentation model based on deep learning, which can accurately segment images in three-dimensional space;
[0011] The segmented left ventricular myocardial area was further divided: the apical CT delayed enhancement image was divided into 4 areas: the apical anterior wall, the apical septum, the apical inferior wall and the apical lateral wall; the papillary muscle CT delayed enhancement image was divided into 6 areas: the middle anterior wall, the middle anterior septum, the middle inferior septum, the middle inferior wall, the middle inferior lateral wall and the middle anterior lateral wall; the basal CT delayed enhancement image was divided into 6 areas: the basal anterior wall, the basal anterior septum, the basal inferior septum, the basal inferior wall, the basal inferior lateral wall and the basal segment anterior lateral wall;
[0012] The pixel mean of each region was obtained and then converted into the corresponding CT mean.
[0013] According to a second aspect of an embodiment of the present disclosure, a system for acquiring CT values of segmental myocardium of a left ventricle is provided, comprising:
[0014] Construct a data set module to perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0015] An image preprocessing module, which preprocesses the cardiac CT delayed enhanced image, including image normalization and noise removal, to improve image quality;
[0016] The myocardial region segmentation module uses the 3DnnUNet model to segment the preprocessed cardiac CT delayed enhancement image to obtain the segmentation result of the left ventricular myocardial region; 3DnnUNet is a segmentation model based on deep learning that can accurately segment images in three-dimensional space;
[0017] The regional division module further divides the segmented left ventricular myocardial area: the delayed CT enhancement image of the apex is divided into 4 areas: the anterior wall of the apex, the apical septum, the inferior wall of the apex and the lateral wall of the apex; the delayed CT enhancement image of the papillary muscle is divided into 6 areas: the anterior wall of the middle part, the anterior septum of the middle part, the inferior septum of the middle part, the inferior wall of the middle part, the inferior lateral wall of the middle part and the anterior lateral wall of the middle part; the delayed CT enhancement image of the base is divided into 6 areas: the anterior wall of the base, the anterior septum of the base, the inferior septum of the base, the inferior wall of the base, the inferior lateral wall of the base and the anterior lateral wall of the base segment;
[0018] The CT value acquisition module calculates the CT value of each area separately: obtains the pixel mean of each area and then converts it into the corresponding CT mean.
[0019] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored and running on the memory, wherein when the processor executes the program, the method for acquiring CT values of left ventricular segment myocardium is implemented.
[0020] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method for acquiring CT values of left ventricular segment myocardium is implemented.
[0021] Compared with the prior art, the above technical solution adopted by the present invention has the following advantages: 1. Improving the accuracy and consistency of segmentation: Using the 3DnnUNet model for automated segmentation greatly improves the accuracy and consistency of myocardial region segmentation and reduces the errors caused by human operation.
[0022] 2. Detailed regional division and CT value calculation: Based on the 16-segment model proposed by the American Heart Association (AHA), the myocardial area is divided in detail to provide more accurate CT values to quantify myocardial damage, which helps clinicians to make accurate diagnosis and treatment.
[0023] 3. Save time and improve efficiency: Compared with the traditional manual segmentation method, the present invention greatly shortens the time of image processing and data analysis and improves work efficiency.
[0024] 4. Automated CT value conversion and calculation: Automatically read DICOM information and convert CT values to ensure the accuracy of CT value calculation, reduce the error risk of manual calculation, and further improve the accuracy and efficiency of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The drawings in the specification, which constitute a part of the present application, are used to provide further understanding of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.
[0026] Figure 1 This is a flow chart of the method for obtaining segmental myocardial CT values of the left ventricle;
[0027] Figure 2 This is the structure diagram of the 3DnnUNet model;
[0028] Figure 3 This is a flow chart of the CT value acquisition method. Specific implementation methods
[0030] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.
[0031] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.
[0032] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.
[0033] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and systems according to various embodiments of the present disclosure. It should be noted that each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of a code may include one or more executable instructions for implementing the logical functions specified in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the flowchart and / or block diagram, and the combination of boxes in the flowchart and / or block diagram can be implemented using a dedicated hardware-based system that performs a specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.
[0034] Embodiment 1:
[0035] This embodiment provides a method for obtaining left ventricular segment myocardial CT values, comprising the following steps:
[0036] S1: Perform delayed CT enhanced scanning on the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0037] Specifically, the doctor selects multiple patients to ensure the representativeness and diversity of the samples. Then, a delayed CT enhancement scan is performed on the left ventricle of each patient to obtain complete left ventricular image data. Different algorithms can be the Aice (Advanced intelligent Clear-IQ Engine) algorithm based on deep learning reconstruction, the MBIR (Model-Based Iterative Reconstruction) algorithm for model iterative reconstruction, and the HIR (Hybrid Iterative Reconstruction) algorithm for hybrid iterative reconstruction. The left ventricle is sliced and reconstructed through different algorithms. The thickness of each slice can be about 10 mm, covering the entire left ventricle from top to bottom. Three different algorithms ensure the diversity and comprehensiveness of the data. Finally, a delayed CT enhanced image of the heart of each patient is obtained, providing basic data for subsequent image processing and analysis.
[0038] S2: preprocessing the cardiac CT delayed enhancement image, including image normalization and noise removal, to improve image quality;
[0039] Specifically, a Gaussian filter is used to smooth the delayed enhanced cardiac CT image; the Gaussian filter can effectively reduce high-frequency noise while maintaining edge information of the image.
[0040] I'=G*I
[0041] Where I is the original image, G is the Gaussian kernel, I' is the filtered image, and * represents the convolution operation. The parameters of the Gaussian kernel (such as the standard deviation σ) are adjusted to achieve the best noise removal effect while avoiding over-smoothing that causes loss of details.
[0042] S3: Use the 3DnnUNet model to segment the preprocessed delayed enhancement CT cardiac image to obtain the segmentation result of the left ventricular myocardium area; specifically, it includes:
[0043] S31: dividing the preprocessed cardiac CT delayed enhancement image into apical category, papillary muscle category and basal category;
[0044] S32: training a 3DnnUNet model of the apex category: using the first 3DnnUNet architecture, setting network parameters, such as convolution kernel size, number of layers, learning rate, etc.; training the model using a cross entropy loss function and an Adam optimizer, performing data enhancement (such as rotation, translation, etc.) during the training process to increase the generalization ability of the model, and using delayed enhanced cardiac CT images of the apex category and their corresponding labels as training data; after the training is completed, saving the best model weights, inputting delayed enhanced cardiac CT images of the apex category into the best 3DnnUNet model, and obtaining a segmented image of the apical myocardial area;
[0045] S33: training 3DnnUNet models of papillary muscle and substrate categories: using the second 3DnnUNet architecture, setting network parameters such as convolution kernel size, number of layers, learning rate, etc.; performing model training through cross entropy loss function and Adam optimizer, performing data enhancement (such as rotation, translation, etc.) during training to increase the generalization ability of the model, and the data used for training are delayed enhanced cardiac CT images of the papillary muscle category, delayed enhanced cardiac CT images of the substrate category, and labels corresponding to the two; after the training is completed, saving the best model weights, inputting delayed enhanced cardiac CT images of the papillary muscle category and the substrate category into the best 3DnnUNet model, and obtaining segmented images of the papillary muscle myocardial region and segmented images of the substrate myocardial region;
[0046] Since there are significant differences in anatomical structure and image features between the apex category and the papillary muscle and base categories, two independent 3DnnUNet models were trained separately.
[0047] S34: The DICOM original images of the apex, papillary muscle and base and the myocardial region segmentation labels corresponding to the three images are spliced to obtain the segmentation result of the left ventricular myocardial region.
[0048] S4: further divide the segmented left ventricular myocardial area: divide the apical CT delayed enhancement image into 4 areas: apical anterior wall, apical septum, apical inferior wall and apical lateral wall; divide the papillary muscle CT delayed enhancement image into 6 areas: middle anterior wall, middle anterior septum, middle inferior septum, middle inferior wall, middle inferior lateral wall and middle anterior lateral wall; divide the basal CT delayed enhancement image into 6 areas: basal anterior wall, basal anterior septum, basal inferior septum, basal inferior wall, basal inferior lateral wall and basal segment anterior lateral wall;
[0049] It should be noted that after the classification is completed, three DICOM images and segmentation mask images are selected for each patient. These three images are the most representative images of the apex, papillary muscle, and base categories that the doctor considers. The specific steps are as follows:
[0050] S41: Use the pydicom.dcmread function to read three DICOM images; use the nib.load function to read the segmentation mask image label (NIFTI format) corresponding to each image; the three DICOM images are the most representative segmentation images of the apical myocardium, papillary muscle myocardium, and basal myocardium regions;
[0051] S42: adjusting the read DICOM image and the segmentation mask image labels so that their positions correspond to each other;
[0052] S43: traverse each segmentation mask image label, record the position where the pixel is 0; the pixel value of the corresponding pixel on the corresponding DICOM image is also set to 0, so as to determine the area on the DICOM image where the CT value needs to be calculated;
[0053] S44: Use the trim.py function to crop the redundant black background in the image and adjust the cropped image to a square;
[0054] S45: Get two functions get6.py and get4.py for dividing regions; in the get6.py function, divide the delayed enhancement CT images of the base and papillary muscles into 6 parts according to the 16-segment model proposed by the American Heart Association (AHA), first define the vertex coordinates of the 6 sector-shaped regions, and then divide the input image into 6 parts and store them in the segments list; similarly, in the get4.py function, divide the delayed enhancement CT image of the apex into 4 parts according to the 16-segment model proposed by the American Heart Association (AHA), first define the vertex coordinates of the 4 sector-shaped regions, and then divide the input image into 4 parts and store them in the segments list.
[0055] S5: Obtain the pixel mean of each region, and then convert it into the corresponding CT mean.
[0056] Medical CT images are different from natural images. Their pixel matrix represents the reconstructed data of CT value or inter-tissue relaxation time. Taking CT as an example, the CT value of water is 0, and its calculation formula is:
[0057]
[0058] In the formula, μ and μ w are the attenuation coefficients of the object being tested and water, respectively. Some devices do not convert pixel values into CT values after scanning, so you need to convert them yourself. The conversion formula between pixel mean and CT mean is:
[0059] CT=pixel×slope+intercept
[0060] Among them, pixel is the pixel of the cardiac CT delayed enhancement image, slope is the slope of the cardiac CT delayed enhancement image, and intercept is the intercept.
[0061] In order to obtain the CT value of the image, you first need to read two DICOM tag information: rescaleintercept and rescale slope. This conversion is very useful in practice because the severity of certain diseases is related to the CT value of the CT image. After using machine learning for segmentation or positioning, the size of the CT value often becomes one of the important indicators of the output. The specific steps are as follows:
[0062] S51: Obtain the rescale intercept and rescale slope of each DICOM image;
[0063] S52: traverse the six regions of the delayed CT enhancement image of the basal region and the six regions of the delayed CT enhancement image of the papillary muscle output by the get6.py function, read the pixel values of the pixels in each region, and then obtain the pixel mean, and use the conversion formula to convert the pixel mean of each region into the CT mean; traverse the four regions of the delayed CT enhancement image of the apex output by the get4.py function, read the pixel values of the pixels in each region, and then obtain the pixel mean, and use the conversion formula to convert the pixel mean of each region into the CT mean;
[0064] S53: storing the CT averages of the six regions of the delayed enhancement image of the basal CT, the six regions of the papillary muscle category, and the four regions of the delayed enhancement image of the apical CT into a list averagesCT;
[0065] S54: making the CT mean values correspond to 16 regions respectively and drawing a 16-segment model;
[0066] S55: Use the stack.py function to stack the segmentation mask images of each myocardial area together to facilitate doctor observation.
[0067] Embodiment 2:
[0068] This embodiment provides a system for acquiring CT values of left ventricular segment myocardium, comprising:
[0069] Construct a data set module to perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0070] An image preprocessing module, which preprocesses the cardiac CT delayed enhanced image, including image normalization and noise removal, to improve image quality;
[0071] The myocardial region segmentation module uses the 3DnnUNet model to segment the preprocessed cardiac CT delayed enhancement image to obtain the segmentation result of the left ventricular myocardial region;
[0072] The regional division module further divides the segmented left ventricular myocardial area: the delayed apical CT enhanced image is divided into 4 areas: the anterior wall of the apex, the apical septum, the inferior apical wall and the lateral apical wall; the delayed papillary muscle CT enhanced image is divided into 6 areas: the anterior wall of the middle part, the anterior septum of the middle part, the inferior septum of the middle part, the inferior wall of the middle part, the inferior lateral wall of the middle part and the anterior lateral wall of the middle part; the myocardial part of the basal slice is divided into 6 areas: the anterior wall of the basal part, the anterior septum of the basal part, the inferior septum of the basal part, the inferior wall of the basal part, the inferior lateral wall of the basal part and the anterior lateral wall of the basal segment;
[0073] The CT value acquisition module calculates the CT value of each area separately: obtains the pixel mean of each area and then converts it into the corresponding CT mean.
[0074] Embodiment three:
[0075] An electronic device includes a memory, a processor, and a computer program stored and running on the memory, wherein the processor implements the above-mentioned method for obtaining the segmental myocardial CT value of the left ventricle when executing the program, including:
[0076] Perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0077] Preprocessing the cardiac CT delayed enhancement image, including image normalization and noise removal;
[0078] The 3DnnUNet model is used to segment the preprocessed delayed enhancement CT images of the heart and obtain the segmentation results of the left ventricular myocardium area.
[0079] The segmented left ventricular myocardial area was further divided: the apical CT delayed enhancement image was divided into 4 areas: the apical anterior wall, the apical septum, the apical inferior wall and the apical lateral wall; the papillary muscle CT delayed enhancement image was divided into 6 areas: the middle anterior wall, the middle anterior septum, the middle inferior septum, the middle inferior wall, the middle inferior lateral wall and the middle anterior lateral wall; the basal CT delayed enhancement image was divided into 6 areas: the basal anterior wall, the basal anterior septum, the basal inferior septum, the basal inferior wall, the basal inferior lateral wall and the basal segment anterior lateral wall;
[0080] The pixel mean of each region was obtained and then converted into the corresponding CT mean.
[0081] Embodiment 4:
[0082] A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the method for obtaining the CT value of the left ventricular segment myocardium is implemented, comprising:
[0083] Perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient;
[0084] Preprocessing the cardiac CT delayed enhancement image, including image normalization and noise removal;
[0085] The 3DnnUNet model is used to segment the preprocessed delayed enhancement CT images of the heart and obtain the segmentation results of the left ventricular myocardium area.
[0086] The segmented left ventricular myocardial area was further divided: the apical CT delayed enhancement image was divided into 4 areas: the apical anterior wall, the apical septum, the apical inferior wall and the apical lateral wall; the papillary muscle CT delayed enhancement image was divided into 6 areas: the middle anterior wall, the middle anterior septum, the middle inferior septum, the middle inferior wall, the middle inferior lateral wall and the middle anterior lateral wall; the basal CT delayed enhancement image was divided into 6 areas: the basal anterior wall, the basal anterior septum, the basal inferior septum, the basal inferior wall, the basal inferior lateral wall and the basal segment anterior lateral wall;
[0087] The pixel mean of each region was obtained and then converted into the corresponding CT mean.
[0088] Those skilled in the art should understand that the modules or steps of the present disclosure can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present disclosure is not limited to any specific combination of hardware and software.
[0089] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
[0090] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.
Claims
1. A method for obtaining segmental myocardial CT values of the left ventricle, characterized in that: The following steps are involved: Perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient; Preprocessing the cardiac CT delayed enhancement image, including image normalization and noise removal; The 3DnnUNet model is used to segment the preprocessed delayed enhancement CT images of the heart and obtain the segmentation results of the left ventricular myocardium area. The segmented left ventricular myocardial area was further divided: the apical CT delayed enhancement image was divided into 4 areas: the apical anterior wall, the apical septum, the apical inferior wall and the apical lateral wall; the papillary muscle CT delayed enhancement image was divided into 6 areas: the middle anterior wall, the middle anterior septum, the middle inferior septum, the middle inferior wall, the middle inferior lateral wall and the middle anterior lateral wall; the basal CT delayed enhancement image was divided into 6 areas: the basal anterior wall, the basal anterior septum, the basal inferior septum, the basal inferior wall, the basal inferior lateral wall and the basal segment anterior lateral wall; The pixel mean of each region was obtained and then converted into the corresponding CT mean.
2. A method for acquiring segmental myocardial CT values of left ventricle according to claim 1, characterized in that: The 3DnnUNet model is used to segment the preprocessed cardiac CT delayed enhancement images, specifically: The pre-processed delayed enhancement CT images of the heart are divided into apical category, papillary muscle category and basal category; Train the 3DnnUNet model for the apex category: Use the first 3DnnUNet architecture and set the network parameters; train the model using the cross entropy loss function and the Adam optimizer, and perform data augmentation during training to increase the generalization ability of the model; After the training is completed, the best model weights are saved, and the delayed enhanced cardiac CT images of the apical category are input into the best 3DnnUNet model to obtain the segmented image of the apical myocardial area; Train the 3DnnUNet model for papillary muscle and base categories: Use the second 3DnnUNet architecture and set the network parameters; train the model using the cross entropy loss function and the Adam optimizer, and perform data augmentation during training to increase the generalization ability of the model; After the training is completed, the best model weight is saved, and the delayed enhanced cardiac CT images of the papillary muscle category and the basal category are input into the best 3DnnUNet model to obtain the segmentation images of the papillary muscle myocardial region and the segmentation images of the basal myocardial region; The segmentation result of the left ventricular myocardial area is obtained by splicing the original DICOM images of the apex, papillary muscle and base and the myocardial region segmentation labels corresponding to these three images.
3. A method for acquiring segmental myocardial CT values of left ventricle according to claim 2, characterized in that: The data used to train the 3DnnUNet model of the apex category are delayed enhanced cardiac CT images of the apex category and their corresponding labels; The data used to train the 3DnnUNet model of papillary muscle and substrate categories are delayed enhanced cardiac CT images of the papillary muscle category, delayed enhanced cardiac CT images of the substrate category, and their corresponding labels.
4. A method for acquiring segmental myocardial CT values of left ventricle according to claim 1, characterized in that: When the segmented left ventricular myocardial area is further divided, the specific steps are as follows: Use the pydicom.dcmread function to read three DICOM images; use the nib.load function to read the segmentation mask image label corresponding to each image; the three DICOM images are the most representative images of the apex, papillary muscle, and base categories; Adjust the read DICOM image and segmentation mask image labels so that their positions correspond to each other; Traverse each segmentation mask image label and record the position where the pixel is 0; the pixel value of the corresponding pixel on the corresponding DICOM image is also set to 0, so as to determine the area on the DICOM image where the CT value needs to be calculated; Crop the extra black background part of the image and adjust the cropped image to a square; Get two functions get6.py and get4.py for dividing regions; in the get6.py function, the delayed CT enhancement images of the base and papillary muscle are divided into 6 parts according to the 16-segment model. It first defines the vertex coordinates of the 6 sector-shaped regions, and then divides the input image into 6 parts and stores them in the segments list; similarly, in the get4.py function, the delayed CT enhancement image of the apex is divided into 4 parts according to the 16-segment model. It first defines the vertex coordinates of the 4 sector-shaped regions, and then divides the input image into 4 parts and stores them in the segments list.
5. The method for acquiring segmental myocardial CT values of left ventricle according to claim 1, characterized in that: The pixel mean is converted to the corresponding CT mean as follows: CT=pixel×slope+intercept Among them, pixel is the pixel of the cardiac CT delayed enhancement image, slope is the slope of the cardiac CT delayed enhancement image, and intercept is the intercept.
6. A method for acquiring segmental myocardial CT values of left ventricle according to claim 5, characterized in that: Get the pixel mean of each area and then convert it into the corresponding CT mean, specifically: Get the rescale intercept and rescale slope of each DICOM image; Traverse the six regions of the delayed CT enhancement image of the basal area and the six regions of the delayed CT enhancement image of the papillary muscle output by the get6.py function, read the pixel values of the pixels in each region, and then obtain the pixel mean, and use the conversion formula to convert the pixel mean of each region to the CT mean; traverse the four regions of the delayed CT enhancement image of the apex output by the get4.py function, read the pixel values of the pixels in each region, and then obtain the pixel mean, and use the conversion formula to convert the pixel mean of each region to the CT mean; The CT averages of the six areas of the delayed enhancement image of the basal CT, the six areas of the delayed enhancement image of the papillary muscle CT, and the four areas of the delayed enhancement image of the apex CT are stored in the list averagesCT; The CT mean values were mapped to 16 regions and a 16-segment model was drawn; The segmentation mask images of each myocardial region are stacked together.
7. A system for acquiring CT values of left ventricular segment myocardium, characterized in that: include: Construct a data set module to perform delayed CT enhanced scanning of the left ventricle of multiple patients to obtain complete left ventricle image data; use different algorithms to slice and reconstruct the left ventricle image data to obtain delayed CT enhanced images of the heart of each patient; An image preprocessing module, which preprocesses the cardiac CT delayed enhanced image, including image normalization and noise removal, to improve image quality; The myocardial region segmentation module uses the 3DnnUNet model to segment the preprocessed cardiac CT delayed enhancement image to obtain the segmentation result of the left ventricular myocardial region; The regional division module further divides the segmented left ventricular myocardial area: the delayed CT enhancement image of the apex is divided into 4 areas: the anterior wall of the apex, the apical septum, the inferior wall of the apex and the lateral wall of the apex; the delayed CT enhancement image of the papillary muscle is divided into 6 areas: the anterior wall of the middle part, the anterior septum of the middle part, the inferior septum of the middle part, the inferior wall of the middle part, the inferior lateral wall of the middle part and the anterior lateral wall of the middle part; the delayed CT enhancement image of the base is divided into 6 areas: the anterior wall of the base, the anterior septum of the base, the inferior septum of the base, the inferior wall of the base, the inferior lateral wall of the base and the anterior lateral wall of the base segment; The CT value acquisition module calculates the CT value of each area separately: obtains the pixel mean of each area and then converts it into the corresponding CT mean.
8. An electronic device comprising a memory, a processor and a computer program stored and running on the memory, characterized in that: When the processor executes the program, the method for acquiring the CT value of the left ventricular segment myocardium is implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a method for acquiring CT values of left ventricular segment myocardium is implemented.
Citation Information
Patent Citations
Interactive three-dimensional medical image segmentation method based on bidirectional gated memory network
CN115937220A
Lung medical CT (Computed Tomography) image segmentation and classification device and equipment
CN116703901A
An imaging toolbox for guiding cardiac resynchronization therapy implantation from patient-specific imaging and body surface potential mapping data
US20170071675A1
