CN112336365A - Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipment - Google Patents
Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipmentDownload PDFInfo
- Publication number
- CN112336365A CN112336365ACN202011251260.8ACN202011251260ACN112336365ACN 112336365 ACN112336365 ACN 112336365ACN 202011251260 ACN202011251260 ACN 202011251260ACN 112336365 ACN112336365 ACN 112336365A
- Authority
- CN
- China
- Prior art keywords
- image
- blood flow
- flow distribution
- myocardial blood
- patient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computed tomography [CT]
- A61B6/032—Transmission computed tomography [CT]
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/48—Diagnostic techniques
- A61B6/481—Diagnostic techniques involving the use of contrast agents
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
- A61B6/503—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for diagnosis of the heart
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
- A61B6/504—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for diagnosis of blood vessels, e.g. by angiography
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
- A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Biophysics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Optics & Photonics (AREA)
- Pathology (AREA)
- Radiology & Medical Imaging (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Molecular Biology (AREA)
- Physics & Mathematics (AREA)
- High Energy & Nuclear Physics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Cardiology (AREA)
- Vascular Medicine (AREA)
- Pulmonology (AREA)
- Theoretical Computer Science (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
Description
Technical Field
The present invention relates to an image processing method, and more particularly, to a myocardial blood flow distribution image acquisition method, system, medium, and electronic device.
Background
Cardiovascular diseases are one of the disease types with the highest death probability in the current society, medical images of heart parts of patients are obtained by scanning medical imaging equipment such as CT (computed tomography) and magnetic resonance equipment, and relevant indexes of human bodies are obtained by processing and calculating the medical images by using an image processing technology, so that the cardiovascular diseases are important bases for clinical diagnosis of the cardiovascular diseases. A series of clinical studies show that the MBF (myocardial blood flow) value not only can more accurately evaluate the myocardial perfusion condition and distinguish ischemic myocardium from infarcted myocardium, but also can early identify abnormal myocardial blood flow distribution of a diabetic patient and a hypertensive patient. Clinically, the MBF value of each point of the myocardium can be obtained by processing and calculating based on the CT dynamic myocardial perfusion image, so that the myocardial MBF distribution image can be obtained, and doctors are helped to diagnose the myocardial blood supply condition of patients.
In the prior art, the MBF value is usually obtained by calculating based on a CTP (CT perfusion) image sequence, that is, obtaining a plurality of continuous images within a period of time after the contrast agent is injected into the heart, drawing a time-density curve according to the dynamic change of the intramuscular contrast agent concentration of the snack at different times, and deriving a Myocardial Blood Flow (MBF) (myocardial perfusion flow) parameter according to the time-density curve data by using a corresponding model. At present, a method for quantitatively evaluating myocardial perfusion is mainly based on a double-chamber model, and a deconvolution or Patlakplot analysis method is utilized to deduce myocardial blood flow parameters according to time-density curve data.
However, in practical applications, the inventor finds that, in the prior art, the MBF parameter is calculated based on the CTP image, on one hand, the CT perfusion scanning process needs to perform CT scanning on the same part of the patient for a plurality of times within a period of time after the contrast medium is injected into the patient, so as to obtain the change of the concentration of the contrast medium within a period of time. The essence of the CT scanning process is the process of reconstructing a signal of a multi-angle X-ray penetrating through a human body to obtain a medical image, and the CTP scanning process has long scanning time and more times, so that the dose of X-ray radiation absorbed by the human body is higher than that of the common CT scanning, and a certain damage can be caused to the human body. On the other hand, CT perfusion scanning has high requirements on CT equipment hardware, can be realized only by high-end CT equipment with a wide detector, and has high cost, and many lower-level hospitals are only equipped with medium-end and low-end CT equipment, and do not have conditions for CT perfusion scanning.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, it is an object of the present invention to provide a myocardial blood flow distribution image acquisition method, system, medium and electronic device for solving the problems caused by the need of acquiring myocardial blood flow distribution images based on CTP images in the prior art.
To achieve the above and other related objects, a first aspect of the present invention provides a myocardial blood flow distribution image acquisition method including: acquiring a CT angiographic image of a patient, the CT angiographic image including a cardiac region of the patient; processing the CT angiography image by using an image generation model based on machine learning to obtain a myocardial blood flow distribution image of the patient; wherein the image generation model is trained by the following method: acquiring a first training data set, wherein the first training data set comprises a plurality of training data pairs, and each training data pair comprises a CT angiography image and a myocardial blood flow distribution image corresponding to the CT angiography image; and training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model.
In an embodiment of the first aspect, a method for obtaining an image of myocardial blood flow distribution of a patient includes: processing the CT angiography image to obtain a myocardial image of the patient; and processing the myocardial image by using the image generation model to acquire a myocardial blood flow distribution image of the patient.
In an embodiment of the first aspect, a method of acquiring an image of a myocardium of a patient includes: segmenting the CT angiography image by utilizing an AI image segmentation model to obtain a myocardial image of the patient; the AI image segmentation model is obtained by training according to the following method: acquiring a second training data set, wherein the second training data set comprises a plurality of CT angiography images and myocardial images contained in the CT angiography images; and training a neural network model by using the second training data set, and taking the trained neural network model as the AI image segmentation model.
In an embodiment of the first aspect, another implementation method for acquiring an image of a myocardium of a patient includes: processing the CT angiographic image such that only a cardiac region is contained in the CT angiographic image; removing blood vessels in the heart region using a thresholding method to obtain an image of the myocardium of the patient.
In an embodiment of the first aspect, after acquiring the first training data set, the training method for the image generation model further includes: and carrying out image registration on the CT angiography image in the first training data set and the corresponding myocardial blood flow distribution image.
In an embodiment of the first aspect, after acquiring the CT angiography image of the patient, the myocardial blood flow distribution image acquiring method further includes: and preprocessing the CT angiography image.
In an embodiment of the first aspect, the image generation model is a generation countermeasure network model.
A second aspect of the present invention provides a myocardial blood flow distribution image acquisition system including: an image acquisition module to acquire a CT angiographic image of a patient, the CT angiographic image including a heart of the patient; the image generation module is connected with the image acquisition module and used for processing the CT angiography image by utilizing an image generation model based on machine learning so as to acquire a myocardial blood flow distribution image of the patient; wherein the image generation model is trained by the following method: acquiring a first training data set, wherein the first training data set comprises a plurality of training data pairs, and each training data pair comprises a CT angiography image and a myocardial blood flow distribution image corresponding to the CT angiography image; and training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model.
A third aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the myocardial blood flow distribution image acquisition method according to any one of the first aspects of the present invention.
A fourth aspect of the present invention provides an electronic apparatus, comprising: a memory having a computer program stored thereon; a processor, communicatively connected to the memory, for executing the computer program and implementing the myocardial blood flow distribution image acquisition method according to any one of the first aspect of the present invention; and the display is in communication connection with the memory and the processor and is used for displaying a related GUI (graphical user interface) of the myocardial blood flow distribution image acquisition method.
As described above, one technical solution of the myocardial blood flow distribution image acquisition method, system, medium, and electronic apparatus according to the present invention has the following advantageous effects:
the myocardial blood flow distribution image acquisition method is characterized in that a CT angiography image of a patient is processed by utilizing an image generation model to acquire the myocardial blood flow distribution image of the patient, and a CTP image of the patient is not required to be acquired in the process, so that the radiation dose absorbed by the patient in the scanning process can be reduced, the requirement on the hardware of scanning equipment is lowered, and the diagnosis cost is lowered.
Drawings
FIG. 1A is a flowchart illustrating a myocardial blood flow distribution image acquisition method according to an embodiment of the present invention.
FIG. 1B is a flowchart illustrating the training of an image generation model according to the myocardial blood flow distribution image acquisition method of the present invention in one embodiment.
Fig. 2 is a flowchart of step S12 according to an embodiment of the myocardial blood flow distribution image acquisition method of the present invention.
Fig. 3 is a flowchart illustrating the method for acquiring myocardial blood flow distribution images according to an embodiment of the invention for training an AI image segmentation model.
FIG. 4 is a flowchart illustrating a method for acquiring myocardial blood flow distribution images according to an embodiment of the present invention.
FIG. 5 is a flowchart illustrating image registration in an embodiment of the myocardial blood flow distribution image acquisition method of the present invention.
FIG. 6 is a flowchart illustrating the training of an impedance network model in an embodiment of the myocardial blood flow distribution image acquisition method of the present invention.
Fig. 7A and 7B are schematic diagrams illustrating the myocardial blood flow distribution image acquisition method according to an embodiment of the present invention for training an impedance network model.
Fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the invention.
Description of the element reference numerals
800 electronic device
810 memory
820 processor
830 display
S11-S12
S21-S22
S121 to S122
S31-S32
S41-S42
S51-S53
S61-S62
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the drawings only show the components related to the present invention rather than being drawn according to the number, shape and size of the components in actual implementation, and the type, number and proportion of the components in actual implementation may be changed arbitrarily, and the layout of the components may be more complicated. Moreover, in this document, relational terms such as "first," "second," and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The existing technology for acquiring the MBF image is usually realized based on the CTP image, and on one hand, the CT perfusion scanning process needs to perform CT scanning on the same part of a patient for multiple times continuously within a period of time after a contrast medium is injected into the patient, so as to obtain the change condition of the concentration of the contrast medium within a period of time. The essence of the CT scanning process is the process of reconstructing a signal of a multi-angle X-ray penetrating through a human body to obtain a medical image, and the CTP scanning process has long scanning time and more times, so that the dose of X-ray radiation absorbed by the human body is higher than that of the common CT scanning, and a certain damage is caused to the human body. On the other hand, CT perfusion scanning has high requirements on CT equipment hardware, can be realized only by high-end CT equipment with a wide detector, and has high cost. Many lower-level hospitals are equipped with only mid-low CT facilities and do not have the conditions for performing CT perfusion scans.
In order to solve the problem, the invention provides a myocardial blood flow distribution image acquisition method, which acquires a myocardial blood flow distribution image of a patient by processing a CT angiography image of the patient by using an image generation model, wherein a CTP image of the patient is not required to be acquired in the process, so that the radiation dose absorbed by the patient in the scanning process can be reduced, the requirement on the hardware of scanning equipment is reduced, and the diagnosis cost is reduced.
Referring to fig. 1A and 1B, in an embodiment of the invention, the method for acquiring the myocardial blood flow distribution image includes:
s11, a CT angiographic image of the patient is acquired, the CT angiographic image including a cardiac region of the patient. The CT angiography combines the CT enhancement technology with the thin-layer, large-range and quick scanning technology, clearly displays details of the cardiac vessels through reasonable post-processing, has the characteristics of no wound and simple and convenient operation, and has lower requirements on scanning equipment.
S12, the CT angiography image is processed using a machine learning based image generation model to obtain a myocardial blood flow distribution image of the patient. Machine learning is a multi-domain interdiscipline, and is a special study on how a computer simulates or realizes human learning behaviors to acquire new knowledge or skills and reorganize an existing knowledge structure to continuously improve the performance of the computer. Therefore, by properly training the image generation model, the image generation model can be provided with the capability of processing the CT angiography image to generate a myocardial blood flow distribution image of the patient. Specifically, a training method of the image generation model in this embodiment includes:
s21, a first training data set is obtained, the first training data set includes a plurality of training data pairs, each training data pair includes a CT angiography image and its corresponding myocardial blood flow distribution image. Each CT angiography image in the first training data set corresponds to a myocardial blood flow distribution image, and in a specific application, a plurality of CT angiography images may be acquired first, and the myocardial blood flow distribution images corresponding to the CT angiography images may be acquired in an existing manner, so as to acquire the first training data set.
And S22, training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model. The training of the machine learning model by using the training data can be realized by adopting the prior art, and details are not repeated here.
As can be seen from the above description, the method for acquiring an image of myocardial blood flow distribution according to this embodiment can acquire an image of myocardial blood flow distribution of a patient by processing a CT angiography image of the patient using an image generation model, and the acquisition process of the CT angiography image has the advantages of being noninvasive, simple and convenient to operate, and low in requirements on scanning equipment, so that the radiation dose absorbed by the patient during the scanning process can be reduced, the requirements on hardware of the scanning equipment can be reduced, and the diagnosis cost can be reduced.
Referring to fig. 2, in an embodiment of the present invention, a method for obtaining an image of myocardial blood flow distribution of a patient includes:
and S121, processing the CT angiography image to acquire a myocardial image of the patient. As described above, since the CT angiography image includes the cardiac region of the patient including the myocardium and the myocardial blood flow distribution image often includes only the pixel data of the myocardial region, in order to reduce the data processing amount of the image generation model and eliminate the possible noise of the myocardial blood flow distribution image due to the pixel points other than the myocardial region as much as possible, step S121 first processes the CT angiography image to separate the myocardial image of the patient from the CT angiography image.
In this step, the image of the myocardium of the patient may be a mask of the myocardium region, and in a specific application, pixels outside the myocardium region may be set to 0 or other constant values, and at this time, the mask of the myocardium region may be directly obtained according to the image pixel values.
And S122, processing the myocardial image by using the image generation model to acquire a myocardial blood flow distribution image of the patient.
In this embodiment, in order to ensure that the image generation model can process the myocardial images, in the process of training the image generation model, it is necessary to process the CT angiography images in the first training data set to acquire the myocardial images included in the CT angiography images, and train the image generation model using the myocardial images and the corresponding myocardial blood flow distribution images as training data.
As can be seen from the above description, in the method for acquiring a myocardial blood flow distribution image according to this embodiment, a myocardial image of a patient is first acquired from the CT angiography image, so that the image generation model only needs to process pixel points in the myocardial image, thereby reducing the data processing amount of the image generation model, reducing or even eliminating noise brought by pixel points outside a myocardial area to the myocardial blood flow distribution image, and facilitating to improve the quality of the acquired myocardial blood flow distribution image.
In an embodiment of the present invention, a method for acquiring an image of a myocardium of a patient includes: the CT angiographic image is segmented using an AI image segmentation model to obtain an image of the myocardium of the patient. Referring to fig. 3, a training method of the AI image segmentation model according to the embodiment includes:
s31, a second training data set is acquired, the second training data set including a plurality of CT angiography images and a myocardium image included in each of the CT angiography images. The myocardial image included in each CT angiography image may be acquired by, for example, manual segmentation.
And S32, training a neural network model by using the second training data set, and taking the trained neural network model as the AI image segmentation model. The method for training the neural network model by using the second training data set can be implemented by using the existing gradient descent method, conjugate gradient method, and the like, and details are not repeated here.
Referring to fig. 4, in an embodiment of the present invention, another implementation method for acquiring a myocardial image of a patient includes:
s41, processing the CT angiography image to remove a background region in the CT angiography image, so that only a cardiac region is included in the CT angiography image. The step S41 can be implemented by using pattern recognition, UNet model, VNet model, etc., and is not limited herein.
S42, removing blood vessels in the heart region by adopting a threshold method to obtain a myocardial image of the patient. Specifically, the gray-scale values of the blood vessels in the heart region are within an interval range, and the thresholding method includes: and deleting all pixel points in the interval range in the CT angiography image, wherein the rest pixel points are the myocardial image of the patient.
In an embodiment of the invention, it is considered that the spatial coordinates of the CT angiographic image and the myocardial blood flow distribution image in the first training dataset may change and that the spatial resolution may differ, which may affect the performance of the image generation model. To address this problem, in this embodiment, after acquiring the first training data set, the training method for the image generation model further includes: and carrying out image registration on the CT angiography image in the first training data set and the corresponding myocardial blood flow distribution image.
Referring to fig. 5, taking any one of the training data pairs in the first training data set as an example, one implementation method for image registration of the myocardial blood flow distribution image includes:
s51, initializing a transformation center; specifically, the centroids of the CT angiography image and the myocardial blood flow distribution image in the myocardial region are respectively found, and the coordinates of the centroids of the two images in the myocardial region are aligned to serve as the transformation center of the subsequent registration. The centroid of the image is also called the centroid of the image, for the image, the pixel value of each point in the image can be regarded as the mass of the point, and the centroid of the image can be directly obtained according to the mass of each point in the image.
S52, 3D affine transformation. And on the basis of the transformation center, carrying out rigid registration of 3D affine transformation on the expiratory phase CT angiography image. Wherein the 3D affine transformation includes rotation, scaling, translation, shear, reflection, and the like. The transformed CT angiographic image and the myocardial blood flow distribution image have the same spatial resolution and spatial dimensions, wherein the transformed expiratory phase CT angiographic image is used as an input image for subsequent elastic registration.
In this step, one implementation method for performing 3D radiation transformation on the expiratory phase CT angiography image is as follows:
wherein (x, y, z) is the pixel coordinates before transformation, (x ', y ', z ') is the pixel coordinates after transformation,
is a transformation matrix, in which aijAnd expressing the pixel value of each pixel point in the image.
S53, B-spline geometric transformation. Specifically, firstly, the result of 3D affine transformation is used as input, and B-spline geometric transformation elastic registration is performed using the image of the myocardial region, so that the contours of the myocardial region in the CT angiography image and the myocardial blood flow distribution image are substantially fitted.
Wherein the B-spline curve is
PkRepresenting the kth control point, N representing the polynomial of the B-spline, k-th order canonical B-spline basis functions
In this embodiment, the registration of the CT angiography image and the myocardial blood flow distribution image in any training data pair can be realized through the above steps of transformation center initialization, 3D affine transformation, B-spline geometric transformation, and the like. After all training data pairs in the first training data set are registered, the first training data set is used for training the image generation model, so that the processing performance of the image generation model is improved, and the accuracy and precision of the obtained myocardial blood flow distribution image are improved.
In an embodiment of the present invention, after acquiring the CT angiography image of the patient, the method for acquiring the myocardial blood flow distribution image further includes: preprocessing the CT angiography image; wherein the preprocessing comprises denoising the CT angiography image.
In this embodiment, the gray value of each pixel point may be set as the median of the gray values of all pixel points in a certain neighborhood window of the point by adopting a median filtering method, that is: the gray value of each point in the image is replaced by the median of the gray values of all pixel points in a neighborhood of the point, so that the gray value of the pixel point is close to the true value, and the isolated noise point is eliminated. Specifically, in this embodiment, a two-dimensional sliding template may be adopted, and the pixel points in the template are sorted according to the gray scale value to generate a monotonously rising (or falling) two-dimensional data sequence; the result of median filtering is g (x, y) ═ med { f (x-k, y-l), (k, l ∈ W) }; wherein g (x, y) is the gray value of the pixel (x, y) after filtering, f (x, y) is the gray value of the pixel (x, y) before filtering, and W is a two-dimensional template, for example, a 5 × 5 rectangular area may be selected as the two-dimensional template; med is a median function, the function value of which is the median of the two or more subsequent data.
In this embodiment, the CT angiography image is preprocessed, so that isolated noise points therein can be reduced, and accuracy of the acquired myocardial blood flow distribution image is improved. The CT angiography image and the myocardial blood flow distribution image in the first training data set may be preprocessed so that the image generation model can generate a myocardial blood flow distribution image with higher accuracy.
In an embodiment of the invention, the image generation model is a generation countermeasure network model. Referring to fig. 6, in the present embodiment, the method for training the image generation model includes:
s61, constructing a generative confrontation network model which comprises a generative model G and a discriminant model D.
S62, training the generative confrontation network model with the first training data set. In the training process, the generating model G is used for generating a virtual MBF image according to the CT angiography image, and D is used for distinguishing and distinguishing the virtual MBF image and the real myocardial blood flow distribution image. And (3) performing iterative training on the generated countermeasure network model through a certain amount of training data, wherein G and D are respectively and continuously optimized: the virtual MBF image generated by the G according to the CT angiography image is closer and closer to the real myocardial blood flow distribution image, and the ability of the D for distinguishing the virtual MBF image from the real myocardial blood flow distribution image is stronger and stronger. The training process continues until the virtual MBF image generated by the G according to the CT angiography image is very close to a real myocardial blood flow distribution image, at the moment, the training is completed, and the generated confrontation network model can be used as the image generation model; in use, the image generation model is capable of generating a corresponding virtual MBF image from an acquired CT angiographic image of the patient, which virtual MBF image may be used in place of a real myocardial blood flow distribution image.
Preferably, the generation of the confrontation network model is based on pix2pixHD, and a pyramid method is adopted, and the input CT angiography image outputs a low-resolution virtual MBF image firstly, and then the low-resolution MBF image is used as the input of another network, so as to generate a virtual MBF image with higher resolution.
Referring to fig. 7A and 7B, the generative model G is composed of two parts: g1 and G2. Wherein, G1 is an end-to-end UNet structure, G2 has two input parts, the left half part of G2 inputs high-resolution images and extracts features, then the high-resolution images and the feature images of the previous layer of the output layer of G1 are added and fused, and the fused information is sent to the back half part of G2, so that G2 outputs high-resolution images. The discriminative model D uses multi-scale discriminators including D1, D2, and D3. Three dimensions of discrimination are for example: 1/4 downsampling of the original, 1/2 downsampling of the original and the original are finally distinguished on three different scales and the results are averaged.
The generative models G1 and G2 are structurally similar, again using the UNet architecture. The multi-scale discriminant models D1, D2, and D3 are structurally similar, using markov discriminants (PatchGAN), namely: the image is equally divided into N multiplied by N slices, the authenticity of each slice is judged respectively, and finally, the average is taken.
Generating antagonistic network models as described in the present embodimentThe LOSS function consists of generating both a challenge network model LOSS (GAN LOSS) and a Feature Matching (FM) LOSS. The loss of the generated confrontation network model is calculated by the loss of a Markov discriminator, and the loss of the feature matching can be obtained by respectively inputting the generated sample and the real sample into the discriminator to extract features and acquiring Element-wise loss according to the features. The loss function can be expressed as
Wherein L isGAN(G,Dk) Representation Generation versus network model loss, LFM(G,Dk) Represents a loss of feature matching, an
T is the total number of layers, NiRepresenting the number of elements of the i-th layer, λ being important for controlling the two losses, E(s,x)Expressing expectation, the value of the method can be adjusted according to actual requirements, s is an input image, and x is an output image. Note that for feature matching loss DkAs a feature extractor only, without maximizing the loss LFM。
Based on the above description of the myocardial blood flow distribution image acquisition method, the present invention also provides a myocardial blood flow distribution image acquisition system, and the myocardial blood flow distribution image acquisition module can implement the myocardial blood flow distribution image acquisition method of the present invention. In an embodiment of the present invention, the myocardial blood flow distribution image acquisition system includes an image acquisition module and an image generation module; wherein the image acquisition module is configured to acquire a CT angiographic image of a patient, the CT angiographic image including a heart of the patient; the image generation module is connected with the image acquisition module and used for processing the CT angiography image by utilizing an image generation model based on machine learning so as to acquire a myocardial blood flow distribution image of the patient.
In this embodiment, the image generation model may be obtained by training according to the following method: acquiring a first training data set, wherein the first training data set comprises a plurality of training data pairs, and each training data pair comprises a CT angiography image and a myocardial blood flow distribution image corresponding to the CT angiography image; and training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model.
Based on the above description of the myocardial blood flow distribution image acquisition method, the present invention also provides a computer-readable storage medium having stored thereon a computer program. The computer program, when executed by a processor, implements the myocardial blood flow distribution image acquisition method of the present invention.
Based on the description of the myocardial blood flow distribution image acquisition method, the invention also provides electronic equipment. Referring to fig. 8, in an embodiment of the invention, the electronic device 800 includes: amemory 810 on which a computer program is stored; aprocessor 820, communicatively coupled to thememory 810, for executing the computer program and implementing the myocardial blood flow distribution image acquisition method of the present invention; adisplay 830, communicatively coupled to thememory 810 and theprocessor 820, for displaying a GUI interactive interface associated with the myocardial blood flow distribution image acquisition method.
The scope of the myocardial blood flow distribution image acquisition method according to the present invention is not limited to the execution sequence of the steps illustrated in this embodiment, and all the solutions of the prior art including step addition, step subtraction, and step replacement according to the principles of the present invention are included in the scope of the present invention.
The present invention also provides a myocardial blood flow distribution image acquisition system, which can implement the myocardial blood flow distribution image acquisition method of the present invention, but the implementation apparatus of the myocardial blood flow distribution image acquisition method of the present invention includes, but is not limited to, the structure of the myocardial blood flow distribution image acquisition system described in this embodiment, and all structural modifications and substitutions of the prior art made according to the principles of the present invention are included in the protection scope of the present invention.
The myocardial blood flow distribution image acquisition method does not depend on CTP scanning, and can acquire the myocardial blood flow distribution image only according to the CT angiography image. Compared with the prior art of acquiring MBF based on CTP scanning, the technical scheme provided by the invention reduces the radiation dose absorbed by a patient in the scanning process, reduces the requirement on the hardware of scanning equipment, and is beneficial to reducing the diagnosis cost.
In conclusion, the present invention effectively overcomes various disadvantages of the prior art and has high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (10)
1. A myocardial blood flow distribution image acquisition method is characterized by comprising the following steps:
acquiring a CT angiographic image of a patient, the CT angiographic image including a cardiac region of the patient;
processing the CT angiography image by using an image generation model based on machine learning to obtain a myocardial blood flow distribution image of the patient;
wherein the image generation model is trained by the following method:
acquiring a first training data set, wherein the first training data set comprises a plurality of training data pairs, and each training data pair comprises a CT angiography image and a myocardial blood flow distribution image corresponding to the CT angiography image;
and training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model.
2. The myocardial blood flow distribution image acquisition method according to claim 1, wherein the realization method of acquiring the myocardial blood flow distribution image of the patient includes:
processing the CT angiography image to obtain a myocardial image of the patient;
and processing the myocardial image by using the image generation model to acquire a myocardial blood flow distribution image of the patient.
3. The myocardial blood flow distribution image acquisition method of claim 2, wherein one implementation of acquiring an image of a myocardium of a patient comprises:
segmenting the CT angiography image by utilizing an AI image segmentation model to obtain a myocardial image of the patient;
the AI image segmentation model is obtained by training according to the following method:
acquiring a second training data set, wherein the second training data set comprises a plurality of CT angiography images and myocardial images contained in the CT angiography images;
and training a neural network model by using the second training data set, and taking the trained neural network model as the AI image segmentation model.
4. The myocardial blood flow distribution image acquisition method according to claim 1, wherein another implementation method of acquiring an image of the myocardium of a patient includes:
processing the CT angiographic image such that only a cardiac region is contained in the CT angiographic image;
removing blood vessels in the heart region using a thresholding method to obtain an image of the myocardium of the patient.
5. The myocardial blood flow distribution image acquisition method according to claim 1, wherein the training method of the image generation model further includes, after acquiring the first training data set:
and carrying out image registration on the CT angiography image in the first training data set and the corresponding myocardial blood flow distribution image.
6. The myocardial blood flow distribution image acquisition method according to claim 1, further comprising, after acquiring a CT angiography image of a patient: and preprocessing the CT angiography image.
7. The myocardial blood flow distribution image acquisition method according to claim 1, characterized in that: the image generation model is a generation antagonizing network model.
8. A myocardial blood flow distribution image acquisition system, characterized by comprising:
an image acquisition module to acquire a CT angiographic image of a patient, the CT angiographic image including a heart of the patient;
the image generation module is connected with the image acquisition module and used for processing the CT angiography image by utilizing an image generation model based on machine learning so as to acquire a myocardial blood flow distribution image of the patient;
wherein the image generation model is trained by the following method:
acquiring a first training data set, wherein the first training data set comprises a plurality of training data pairs, and each training data pair comprises a CT angiography image and a myocardial blood flow distribution image corresponding to the CT angiography image;
and training a machine learning model by using the first training data set, and taking the trained machine learning model as the image generation model.
9. A computer-readable storage medium having stored thereon a computer program, characterized in that: the computer program, when executed by a processor, implements the myocardial blood flow distribution image acquisition method of any one of claims 1-7.
10. An electronic device, characterized in that the electronic device comprises:
a memory having a computer program stored thereon;
a processor, communicatively connected to the memory, for executing the computer program and implementing the myocardial blood flow distribution image acquisition method of any one of claims 1-7;
a display, communicatively coupled to the memory and the processor, for displaying a GUI interface associated with the myocardial blood flow distribution image acquisition method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011251260.8ACN112336365B (en) | 2020-11-11 | 2020-11-11 | Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011251260.8ACN112336365B (en) | 2020-11-11 | 2020-11-11 | Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112336365Atrue CN112336365A (en) | 2021-02-09 |
| CN112336365B CN112336365B (en) | 2021-09-28 |
Family
ID=74363240
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011251260.8AActiveCN112336365B (en) | 2020-11-11 | 2020-11-11 | Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112336365B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114420286A (en)* | 2022-01-14 | 2022-04-29 | 深圳睿心智能医疗科技有限公司 | Method and device for acquiring blood vessel physiological parameters, electronic equipment and storage medium |
| WO2024229165A1 (en)* | 2023-05-02 | 2024-11-07 | Cleerly, Inc. | Systems, methods, and devices for image-based plaque analysis and risk determination |
| US12245882B2 (en) | 2020-01-07 | 2025-03-11 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12283046B2 (en) | 2020-01-07 | 2025-04-22 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12299885B2 (en) | 2022-03-10 | 2025-05-13 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
| US12324696B2 (en) | 2022-03-10 | 2025-06-10 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
| US12324695B2 (en) | 2020-01-07 | 2025-06-10 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12380560B2 (en) | 2022-03-10 | 2025-08-05 | Cleerly, Inc. | Systems, methods, and devices for image-based plaque analysis and risk determination |
| US12440180B2 (en) | 2024-02-29 | 2025-10-14 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090129536A1 (en)* | 2007-11-20 | 2009-05-21 | Toshiba Medical Systems Corporation | X-ray ct apparatus and myocardial perfusion image generating system |
| US20110110488A1 (en)* | 2007-09-28 | 2011-05-12 | The Johns Hopkins University | Combined multi-detector ct angiography and ct myocardial perfusion imaging for the diagnosis of coronary artery disease |
| CN105096388A (en)* | 2014-04-23 | 2015-11-25 | 北京冠生云医疗技术有限公司 | Computational Fluid Dynamics (CFD) based coronary artery blood flow simulating system and method |
| CN105517492A (en)* | 2013-09-06 | 2016-04-20 | 皇家飞利浦有限公司 | Processing apparatus for processing cardiac data |
| JP2019111282A (en)* | 2017-12-26 | 2019-07-11 | 国立大学法人千葉大学 | Blood flow distribution generation program, blood flow distribution generation method, and blood flow distribution display device |
| CN110276748A (en)* | 2019-06-12 | 2019-09-24 | 上海移视网络科技有限公司 | Analysis Method of Blood Flow Velocity and Fractional Flow Reserve in Myocardial Ischemia Region |
| CN110428420A (en)* | 2018-09-05 | 2019-11-08 | 深圳科亚医疗科技有限公司 | The method, apparatus and medium of flowing information coronarius are determined based on the coronary artery CT angiographic image of patient |
| US20190350538A1 (en)* | 2018-05-16 | 2019-11-21 | Case Western Reserve University | System-independent quantitative perfusion imaging |
| CN110853111A (en)* | 2019-11-05 | 2020-02-28 | 上海杏脉信息科技有限公司 | Medical image processing system, model training method and training device |
| WO2020154398A1 (en)* | 2019-01-22 | 2020-07-30 | Arizona Board Of Regents On Behalf Of The University Of Arizona | Noninvasive real-time patient-specific assessment of stroke severity |
- 2020
- 2020-11-11CNCN202011251260.8Apatent/CN112336365B/enactiveActive
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110110488A1 (en)* | 2007-09-28 | 2011-05-12 | The Johns Hopkins University | Combined multi-detector ct angiography and ct myocardial perfusion imaging for the diagnosis of coronary artery disease |
| US20090129536A1 (en)* | 2007-11-20 | 2009-05-21 | Toshiba Medical Systems Corporation | X-ray ct apparatus and myocardial perfusion image generating system |
| CN105517492A (en)* | 2013-09-06 | 2016-04-20 | 皇家飞利浦有限公司 | Processing apparatus for processing cardiac data |
| CN105096388A (en)* | 2014-04-23 | 2015-11-25 | 北京冠生云医疗技术有限公司 | Computational Fluid Dynamics (CFD) based coronary artery blood flow simulating system and method |
| JP2019111282A (en)* | 2017-12-26 | 2019-07-11 | 国立大学法人千葉大学 | Blood flow distribution generation program, blood flow distribution generation method, and blood flow distribution display device |
| US20190350538A1 (en)* | 2018-05-16 | 2019-11-21 | Case Western Reserve University | System-independent quantitative perfusion imaging |
| CN110428420A (en)* | 2018-09-05 | 2019-11-08 | 深圳科亚医疗科技有限公司 | The method, apparatus and medium of flowing information coronarius are determined based on the coronary artery CT angiographic image of patient |
| WO2020154398A1 (en)* | 2019-01-22 | 2020-07-30 | Arizona Board Of Regents On Behalf Of The University Of Arizona | Noninvasive real-time patient-specific assessment of stroke severity |
| CN110276748A (en)* | 2019-06-12 | 2019-09-24 | 上海移视网络科技有限公司 | Analysis Method of Blood Flow Velocity and Fractional Flow Reserve in Myocardial Ischemia Region |
| CN110853111A (en)* | 2019-11-05 | 2020-02-28 | 上海杏脉信息科技有限公司 | Medical image processing system, model training method and training device |
Non-Patent Citations (1)
| Title |
|---|
| 刘嘉策，张华: "CT心肌灌注成像的临床应用及展望", 《中国中西医结合影像学杂志》* |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US12245882B2 (en) | 2020-01-07 | 2025-03-11 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12283046B2 (en) | 2020-01-07 | 2025-04-22 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12324695B2 (en) | 2020-01-07 | 2025-06-10 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| US12396695B2 (en) | 2020-01-07 | 2025-08-26 | Cleerly, Inc. | Systems, methods, and devices for medical image analysis, diagnosis, risk stratification, decision making and/or disease tracking |
| CN114420286A (en)* | 2022-01-14 | 2022-04-29 | 深圳睿心智能医疗科技有限公司 | Method and device for acquiring blood vessel physiological parameters, electronic equipment and storage medium |
| US12299885B2 (en) | 2022-03-10 | 2025-05-13 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
| US12324696B2 (en) | 2022-03-10 | 2025-06-10 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
| US12380560B2 (en) | 2022-03-10 | 2025-08-05 | Cleerly, Inc. | Systems, methods, and devices for image-based plaque analysis and risk determination |
| US12406365B2 (en) | 2022-03-10 | 2025-09-02 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
| WO2024229165A1 (en)* | 2023-05-02 | 2024-11-07 | Cleerly, Inc. | Systems, methods, and devices for image-based plaque analysis and risk determination |
| US12440180B2 (en) | 2024-02-29 | 2025-10-14 | Cleerly, Inc. | Systems, devices, and methods for non-invasive image-based plaque analysis and risk determination |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112336365B (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112336365B (en) | Myocardial blood flow distribution image acquisition method, myocardial blood flow distribution image acquisition system, myocardial blood flow distribution image acquisition medium and electronic equipment | |
| JP7246866B2 (en) | medical image processor | |
| Kumar et al. | Medical images classification using deep learning: a survey | |
| Yang et al. | CT image denoising with perceptive deep neural networks | |
| Ni et al. | Global channel attention networks for intracranial vessel segmentation | |
| Liu et al. | Automatic whole heart segmentation using a two-stage u-net framework and an adaptive threshold window | |
| CN110163809A (en) | Confrontation network DSA imaging method and device are generated based on U-net | |
| Koundal et al. | Challenges and future directions in neutrosophic set-based medical image analysis | |
| Li et al. | Automatic quantification of epicardial adipose tissue volume | |
| CN114565711B (en) | Cardiac image reconstruction method and system based on deep learning | |
| CN114155208B (en) | A method and device for evaluating atrial fibrillation based on deep learning | |
| CN114241077B (en) | CT image resolution optimization method and device | |
| CN112562058B (en) | Method for quickly establishing intracranial vascular simulation three-dimensional model based on transfer learning | |
| CN119206038A (en) | A method for reconstructing cardiac interventional surgery scenes based on 3D ultrasound imaging rendering | |
| CN112315493A (en) | Myocardial blood flow distribution image acquisition method, system, medium and electronic device | |
| Kshirsagar et al. | Generative ai-assisted novel view synthesis of coronary arteries for angiography | |
| Zarei et al. | A physics-informed deep neural network for harmonization of CT images | |
| CN114612484A (en) | Retinal OCT Image Segmentation Method Based on Unsupervised Learning | |
| Travaglini et al. | A mathematical model for the analysis of eye fundus images in healthy and diabetic patients | |
| CN112508881A (en) | Intracranial blood vessel image registration method | |
| Pal et al. | Panoptic Segmentation and Labelling of Lumbar Spine Vertebrae using Modified Attention Unet | |
| CN116342726A (en) | Sparse angle CT image artifact suppression method based on structure enhancement and artifact estimation | |
| Dimililer et al. | Image preprocessing phase with artificial intelligence methods on medical images | |
| Ruhaiyem et al. | Cerebrovascular Segmentation Based on Edge Preserving Filters Technique in Magnetic Resonance Angiography Images: A Systematic Review. | |
| Wang et al. | GAN-Based Architecture for Low-dose Computed Tomography Imaging Denoising |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |