WO2021259395A2

WO2021259395A2 - Method and apparatus for obtaining myocardial bridge image, and electronic device and storage medium

Info

Publication number: WO2021259395A2
Application number: PCT/CN2021/122321
Authority: WO
Inventors: 梁隆恺; 吴振洲; 刘盼
Original assignee: 北京安德医智科技有限公司
Priority date: 2021-05-17
Filing date: 2021-09-30
Publication date: 2021-12-30
Also published as: CN113139959B; CN113139959A; WO2021259395A3

Abstract

The present disclosure relates to a method and apparatus for obtaining a myocardial bridge image, and an electronic device and a storage medium. The method comprises: acquiring a CTA image, a coronary artery blood vessel image and a heart image of the same target object, wherein the CTA image includes a CT value; according to the coronary artery blood vessel image, dividing a coronary artery blood vessel region into a plurality of coronary artery sub-regions; in regions, which correspond to the coronary artery sub-regions, in the CTA image and the heart image, extracting image features representing a myocardial bridge, wherein the image features can reflect a CT value distribution situation in an extravascular region, a heart pixel distribution situation in the extravascular region and a heart pixel distribution situation in an intravascular region; and inputting the image features into a trained machine learning model, and determining a first region, which includes the myocardial bridge, in the coronary artery sub-regions, so as to generate a myocardial bridge image. According to the embodiments of the present disclosure, the efficiency and accuracy of the acquisition of a myocardial bridge image can be improved, and the manual workload is reduced.

Description

Method and device for obtaining myocardial bridge image, electronic equipment and storage medium

Technical field

The present disclosure relates to the field of image processing, and in particular to a method and device for obtaining a myocardial bridge image, electronic equipment, and storage medium.

Background technique

Coronary artery (coronary artery) myocardial bridge is a manifestation of congenital coronary artery dysplasia. Myocardial bridge is specifically manifested as a section of the main coronary artery or its branches covered by myocardium. Usually, the main coronary artery or its branches are distributed on the surface of the heart rather than in the myocardium. This myocardium covering the coronary artery is called a myocardial bridge. Myocardial bridge will cause the covered coronary artery to be compressed by myocardium during systole, which may cause myocardial ischemia. The local incidence of coronary artery disease may also be related to myocardial bridge. Therefore, obtaining accurate myocardial bridge images is of high value for both clinical applications and myocardial bridge research.

Usually, the myocardial bridge image comes from coronary angiography. Experienced physicians cooperate with other medical tests and clinical symptoms to mark the myocardial bridge part in the coronary angiography image to obtain the myocardial bridge image. However, this scheme is inefficient, and it is difficult to obtain images of the myocardial bridge where the coronary artery is covered with a shallower part.

Summary of the invention

In view of this, the present disclosure proposes an image processing technical solution.

According to an aspect of the present disclosure, there is provided a method for obtaining a myocardial bridge image, the method including:

Acquiring a CTA image, a coronary vascular image, and a heart image of the same target object, the CTA image containing CT values;

Divide the coronary vascular area into multiple coronary sub-areas according to the coronary vascular image;

Extract image features that characterize the myocardial bridge from the regions corresponding to the multiple coronary subregions in the CTA image and the heart image, and the regions corresponding to the multiple coronary subregions include the intravascular in the coronary subregions Area and extravascular area, the image features respectively representing the distribution of CT values in the extravascular area, the distribution of cardiac pixels in the extravascular area, and the distribution of cardiac pixels in the intravascular area ；

The image features are input to the trained machine learning model, and the first region containing the myocardial bridge in the multiple coronary subregions is determined to generate the myocardial bridge image.

In a possible implementation manner, the dividing the coronary blood vessel area into coronary sub-areas according to the coronary blood vessel image includes:

According to the coronary blood vessel image, extract the midline point connecting the midline points representing the coronary blood vessels;

Dividing the midline point connection into midline point line segments according to the connectivity of the midline point connection;

According to the first threshold, divide the midline point line segment into midline point sub-line segments;

According to the midline point sub-line segment, the coronary blood vessel area is divided into coronary artery sub-areas.

In a possible implementation manner, the regions corresponding to the multiple coronary subregions include regions with different distances from the line connecting the midline points, and the method further includes:

According to the coronary vascular image, generating a vascular mask corresponding to each coronary sub-region according to different distances from the line of the midline point;

According to the blood vessel mask, determine the regions corresponding to the multiple coronary subregions in the CTA image and the heart image,

Among them, the area located in the blood vessel mask and outside the coronary vessel wall is the extravascular area, and the area located in the blood vessel mask, and the area inside the coronary vessel wall is the intravascular area.

In a possible implementation manner, the extracting image features that characterize the myocardial bridge in the regions corresponding to the multiple coronary subregions in the CTA image and the heart image includes:

In the CTA image and the heart image in the regions corresponding to the multiple coronary artery subregions, extracting the image features that characterize the myocardial bridge includes:

Obtaining, according to the blood vessel mask and the CTA image, a first percentage of the CT value distribution in the extravascular area corresponding to each of the distances;

Obtaining, according to the blood vessel mask and the heart image, a second percentage of the distribution of cardiac pixels in the extravascular area corresponding to each of the distances;

According to the blood vessel mask and the heart image, a third percentage of the distribution of heart pixels in the intravascular area is obtained.

In a possible implementation manner, inputting the image features into a trained machine learning model, and determining a first region containing a myocardial bridge in the plurality of coronary subregions to generate a myocardial bridge image includes:

The midline point line segment corresponding to the first region containing the myocardial bridge is expanded to obtain an image of the myocardial bridge.

In a possible implementation manner, the method further includes:

Input the image features characterizing the myocardial bridge extracted from the CTA image sample, the coronary blood vessel image sample, and the heart image sample into the untrained machine learning model to obtain the second region containing the myocardial bridge;

Determining the degree of overlap between the second area and the third area, where the third area represents an artificially labeled myocardial bridge area;

According to the degree of overlap, the machine learning model is trained.

In a possible implementation manner, the CTA image, coronary blood vessel image, and heart image are 3D images.

According to another aspect of the present disclosure, there is provided an apparatus for obtaining a myocardial bridge image, including:

An image acquisition module for acquiring CTA images, coronary vascular images, and heart images of the same target object, the CTA images containing CT values;

The coronary artery sub-areas division module is used to divide the coronary blood vessel area into multiple coronary sub-areas according to the coronary blood vessel image;

The image feature extraction module is used to extract image features that characterize the myocardial bridge in the regions corresponding to the multiple coronary subregions in the CTA image and the heart image, and the regions corresponding to the multiple coronary subregions include The intravascular area and the extravascular area of the coronary subregion, the image features respectively representing the distribution of CT values in the extravascular area, the distribution of cardiac pixels in the extravascular area, and the intravascular area The distribution of heart pixels in the area;

The image generation module is configured to input the image features into the trained machine learning model, and determine the first region containing the myocardial bridge in the multiple coronary sub-regions to generate the myocardial bridge image.

In a possible implementation manner, the coronary artery sub-region dividing module is used to extract, according to the coronary vascular image, the midline point connection formed by the midline points representing the coronary blood vessels; and the connection according to the midline point Divide the midline point line into midline point line segments; divide the midline point line segment into midline point sub-line segments according to the first threshold; divide the coronary vascular area into coronary artery sub-regions according to the midline point sub-line segment .

In a possible implementation manner, the regions corresponding to the multiple coronary subregions include regions at different distances from the line connecting the midline points, and the device for obtaining images of the myocardial bridge further includes

The blood vessel mask module is used to generate a blood vessel mask corresponding to each coronary artery sub-region according to different distances from the line of the midline point;

The corresponding region module is used to determine the regions corresponding to the multiple coronary subregions in the CTA image and the heart image according to the blood vessel mask,

In a possible implementation manner, the image feature extraction module is configured to obtain a first percentage of the CT value distribution in the extravascular area corresponding to each of the distances according to the blood vessel mask and the CTA image; According to the blood vessel mask and the heart image, obtain the second percentage of the heart pixel distribution in the extravascular area corresponding to each of the distances; obtain the intravascular area according to the blood vessel mask and the heart image The third percentage of the central heart pixel distribution.

In a possible implementation manner, the image generation module is used to perform an expansion operation on the midline point line segment corresponding to the first region containing the myocardial bridge to obtain an image of the myocardial bridge.

In a possible implementation, the device for obtaining a myocardial bridge image further includes:

The feature input module is used to input the image features representing the myocardial bridge extracted from the CTA image sample, the coronary blood vessel image sample, and the heart image sample into the untrained machine learning model to obtain the second region containing the myocardial bridge;

The degree of coincidence determining module is used to determine the degree of coincidence between the second area and the third area, where the third area represents an artificially marked myocardial bridge area;

The training module is used to train the machine learning model according to the degree of overlap.

In a possible implementation manner, the image feature further includes: an imageomics feature.

According to another aspect of the present disclosure, there is provided an electronic device including: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to execute the above method.

According to another aspect of the present disclosure, there is provided a non-volatile computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions implement the above method when executed by a processor.

According to the embodiments of the present disclosure, the coronary vascular area on the coronary vascular image is divided into multiple coronary sub-regions, the image features representing the myocardial bridge are extracted from the CTA image and the heart image, and the machine learning model is used to bridge the myocardial bridge based on the image features. Predict where you are. According to the prediction result, the image of the myocardial bridge is regenerated. In this way, the efficiency and accuracy of acquiring images of the myocardial bridge can be improved, and the manual workload can be reduced.

According to the following detailed description of exemplary embodiments with reference to the accompanying drawings, other features and aspects of the present disclosure will become clear.

Description of the drawings

The drawings included in the specification and constituting a part of the specification together with the specification illustrate exemplary embodiments, features, and aspects of the present disclosure, and are used to explain the principle of the present disclosure.

Fig. 1 shows a flowchart of a method for obtaining a myocardial bridge image according to an embodiment of the present disclosure.

Fig. 2 shows a schematic diagram of the midline point and line segment division according to an embodiment of the present disclosure.

Fig. 3 shows a schematic diagram of dividing a coronary artery sub-areas by using midline point sub-line segments according to an embodiment of the present disclosure.

Fig. 4 shows a schematic diagram of blood vessel masks with different distances from the midline point in accordance with an embodiment of the present disclosure.

Fig. 5 shows a block diagram of a method for obtaining a myocardial bridge image according to an embodiment of the present disclosure.

Fig. 6 shows a block diagram of an electronic device according to an embodiment of the present disclosure.

detailed description

Hereinafter, various exemplary embodiments, features, and aspects of the present disclosure will be described in detail with reference to the drawings. The same reference numerals in the drawings indicate elements with the same or similar functions. Although various aspects of the embodiments are shown in the drawings, unless otherwise noted, the drawings are not necessarily drawn to scale.

The dedicated word "exemplary" here means "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" need not be construed as being superior or better than other embodiments. The "first" and "second" in the embodiments of the present disclosure are used to distinguish the described objects, and should not be understood as other restrictions on the order of the described objects or the like.

In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. Those skilled in the art should understand that the present disclosure can also be implemented without certain specific details. In some instances, the methods, means, elements, and circuits well-known to those skilled in the art have not been described in detail in order to highlight the gist of the present disclosure.

Myocardial bridge is a manifestation of congenital coronary artery (coronary artery) dysplasia. Myocardial bridge is specifically manifested as a section of the main coronary artery or its branches being covered by myocardium. Usually, the main coronary artery or its branches are distributed on the surface of the heart rather than in the myocardium. This myocardium covering the coronary artery is called a myocardial bridge. Myocardial bridge will cause the covered coronary artery to be compressed by myocardium during systole, which may cause myocardial ischemia. The local incidence of coronary artery disease may also be related to myocardial bridge.

Medically, the research and detection of myocardial bridge are mainly realized by imaging methods. Usually, the myocardial bridge image comes from coronary angiography. However, some myocardial bridges are almost completely occluded due to their proximal coronary artery, or fixed stenosis caused by atherosclerosis restricts coronary blood perfusion and masks the signs of transient stenosis during systole, or due to The presence of vasospasm is difficult to detect by angiography.

Therefore, in the prior art, doctors usually use coronary angiography with ultrasound images, Doppler images, clinical symptoms, electrocardiograms, etc., to comprehensively determine the position of the myocardial bridge in the coronary angiography image, and obtain the myocardial bridge through the doctor’s mark. image.

However, this scheme is inefficient, and it is difficult to obtain images of the myocardial bridge where the coronary artery is covered with a shallower part.

Therefore, the embodiments of the present disclosure propose a method for obtaining a myocardial bridge image, which can improve the efficiency and accuracy of obtaining the myocardial bridge image.

Fig. 1 shows a flowchart of a method for obtaining a myocardial bridge image according to an embodiment of the present disclosure. Through the method flow shown in FIG. 1, the process of obtaining the image of the myocardial bridge is exemplified.

In step 11, a CTA image, a coronary vascular image, and a heart image of the same target object are acquired, and the CTA image includes CT values.

The image scanning device scans the scanned object to obtain a scanned image with one or some characteristics. In some implementations, the image scanning device may be a computer tomography (Computed Tomography, CT) device. For example, under the action of an iodine contrast agent, a CT scan of the heart can obtain a computed tomography angiography (CT Angiography, CTA) image that can clearly show the blood vessels of the heart. It can be understood that based on the imaging principle of CT equipment, the pixels of the CTA image carry CT value information that characterizes the density of the scanned object.

The coronary blood vessel image is an image containing pixel information of the coronary blood vessel (that is, the pixel information of the pixel where the coronary blood vessel is located), and the heart image is an image containing the heart pixel information (that is, the pixel information of the pixel where the heart is located). Among them, the pixels that characterize coronary blood vessels are coronary blood vessel pixels; the pixels that characterize the heart are cardiac pixels.

In a possible implementation manner, the coronary blood vessel image can be obtained through a CTA image of the heart, and the heart image can be obtained through a CT image.

In a possible implementation, the coronary image can be binarized, which means that the pixel value of the coronary blood vessel is 1, and the other pixel values are 0; the heart image can be binarized to represent the pixels of the heart The value is 1, the other pixel values are 0.

In a possible implementation manner, the CTA image, the coronary blood vessel image, and the heart image are 3D images.

The use of 3D images facilitates the analysis of scanned objects from all directions, not only includes the two dimensions of the planar image, but also includes the depth, avoiding missing the myocardial bridge that wraps the coronary blood vessels in the superficial layer, and improving the accuracy of the judgment of the myocardial bridge .

In step S12, the coronary blood vessel area is divided into multiple coronary sub-areas according to the coronary blood vessel image.

The coronary blood vessel pixels in the coronary blood vessel image constitute the coronary blood vessel area, and the coronary blood vessel area is divided according to a certain rule. The divided areas are called coronary sub-areas. In the embodiment of the present disclosure, there is no limitation on the division rule of the coronary artery sub-areas.

In a possible implementation manner, the method for dividing coronary sub-regions in step S12 may include:

According to the coronary blood vessel image, extract the midline point line connecting the midline points of the coronary blood vessels; according to the connectivity of the midline point line, divide the midline point line into midline point line segments; according to the first Threshold, dividing the midline point line segment into midline point sub-line segments; according to the midline point sub-line segment, the coronary blood vessel area is divided into coronary artery sub-areas.

Exemplarily, the midline point of the coronary blood vessel is extracted from the coronary blood vessel image, where the midline point is the pixel point on the midline of the coronary blood vessel. All the pixels on the midline of the blood vessel can be determined as the midline point, or the midline point can be determined according to a certain rule, and the spacing between the midline points is equal. For example, one pixel is determined as a midline point every four pixels. The embodiment of the present disclosure does not limit the determination rule of the midline point.

The midline points are connected to form a midline point connection. According to the connectivity of the midline point connection, the midline point connection is divided into midline point line segments.

In the following, Figure 2 schematically illustrates how to divide the midline point line segment according to the connectivity of the midline point connection.

Use a certain midline point A as the starting point to extend the midline point on the line of coronary vascular traversing midline points. When traversing to another midline point B on the line of two or more midline points, pause the traversal; _{The L 1} segment between the end point A and the end point B is determined as a midline point line segment. Or, use a midline point B as the starting point to traverse the coronary artery, and traverse to another midline point C or midline point D or midline point E at the end of the midline point connection, pause the traversal, and set L ₂ or L ₃ Or L _{4 is} determined as a midline point line segment. In this way, any midline point on each divided midline point line segment, except for the two end points, belongs only to the midline point line segment.

Then, each midline point line segment can be divided according to the number of midline points to obtain midline point sub-segments, and the number of midline points contained in each midline point sub-line segment can be equal. The number of midline points in each midline point sub-line segment may be a preset first threshold. For example, the first threshold may be preset to 7, that is, each midline point sub-line segment contains 7 midline points. The embodiment of the present disclosure does not limit the value of the first threshold.

According to the midline point sub-line segment, the coronary vascular area is divided to obtain the coronary sub-areas. Each coronary sub-area can contain a midline point line segment; and the end point of the midline point line segment is on the boundary of the coronary sub-region. It is the boundary perpendicular to the midline of coronary vessels. Fig. 3 schematically shows the way of dividing the coronary artery sub-areas. The coronary blood vessel area is divided into three coronary artery sub-areas _{by the midline point line segments l 1} , l ₂ , and l _3.

In practice, the length of some coronary vessels wrapped by myocardial bridge is small or the length of coronary vessels wrapped by superficial layers is small. Therefore, dividing the coronary artery area into sub-regions for prediction and processing can avoid the need for shorter lengths. The missed judgment of the myocardial bridge improves the accuracy of judging the myocardial bridge. In addition, according to the first threshold, the midline points and line segments are further divided into equal lengths, which is convenient for batch operation and improves efficiency.

In step S13, the coronary artery sub-region obtained in step 12 is corresponding to the CTA image and the heart image, so that the CTA image and the cardiac image are also divided into multiple regions, and each region corresponds to a coronary sub-region. In each coronary subregion, the coronary vessel pixels within the coronary subregion boundary constitute the intravascular region of the coronary subregion, and the pixels within the coronary subregion boundary except for the coronary vessel pixels constitute the coronary artery The extravascular area of the area. In the above-mentioned CTA image and heart image, in the region corresponding to the coronary artery subregion, image features that can characterize the myocardial bridge are extracted.

Various parts of the human body, such as: human tissue, blood, blood vessels, organs, bones, etc., correspond to different CT value ranges. Therefore, the CT value of each pixel of the CT image is an indicator for calibrating each part of the human body. When the CT value is not within the corresponding CT value range of this part of the body, then this part of the body may have an abnormality. The CT value distribution of the coronary vascular segment wrapped by the myocardial bridge is different from the CT value distribution of the coronary vascular segment that normally walks under the epicardium. Therefore, the CT value distribution of the extravascular area in the coronary subregion can be used as a discrimination An indicator of the position of the myocardial bridge.

Normal coronary blood vessels are between the epimyocardium and myocardium, so the outer side of the normal coronary blood vessels is the myocardium. The outside of the coronary vessels wrapped by the myocardial bridge are all myocardium. Then, there will be a difference in the distribution of cardiac pixels between the extravascular area of normal coronary vessels and the extravascular area of coronary vessels wrapped by myocardial bridge. Therefore, the distribution of cardiac pixels in the extravascular area of the coronary subregion can be used as an index to distinguish the position of the myocardial bridge.

Since normal coronary vessels are located on the surface of the myocardium, no cardiac pixels appear inside the coronary vessels in the image. However, the coronary blood vessels wrapped by the myocardial bridge are reflected in the image, and there will be a phenomenon of heart pixels inside the coronary blood vessels. Therefore, the distribution of cardiac pixels in the intravascular area of the coronary artery subregion can be used as an index to distinguish the position of the myocardial bridge.

In a possible implementation manner, the feature of step S13 reflecting the distribution of CT values in the extravascular area of the coronary subregion, the extravascular area of the coronary subregion, and the distribution of cardiac pixels in the intravascular area can be extracted. Extracting according to different distances from the midline point, the specific method includes: according to the coronary vascular image, according to the different distances from the midline point to generate the blood vessel mask corresponding to each coronary sub-region; according to the blood vessel mask , Determine the regions corresponding to the multiple coronary sub-regions in the CTA image and the heart image, wherein the region located in the blood vessel mask but outside the coronary vessel wall is the extravascular area, located in the blood vessel mask, and the crown The area within the vessel wall is the intravascular area.

The different distances from the midline point can be determined in pixels. For example, a unit distance can be represented by a unit number of pixels (the number of units can be determined according to needs), then the blood vessel mask of a unit distance contains pixels within one unit distance from the center line point; the same is true, two units The inside of the blood vessel mask of the distance contains the pixels within two unit distances from the midline point.

Fig. 4 is a schematic diagram of a cross-sectional view of a blood vessel mask. The figure shows the blood vessel mask from a unit distance of 1h to five units of 5h from the center line of the coronary blood vessel. Among them, the solid circle represents the blood vessel wall, and the dashed circle represents the boundary of the blood vessel mask; the blood vessel mask 3h from the midline point coincides with the blood vessel wall.

When the image is a 3D image, the blood vessel mask may also be a 3D mask. The cross section perpendicular to the length direction is shown in FIG. 4, and the length direction range corresponds to each coronary artery sub-region, shown in FIG. 3.

It should be understood that this figure is only an embodiment, and a specific value of a unit distance can be adjusted as required in practical applications. The embodiments of the present disclosure do not limit the number of pixels included in a unit distance.

Through the vascular mask, the image characteristics of different distances from the coronary vessels and the depth of the coronary vessels can be obtained, so as to improve the accuracy of judging the myocardial bridge of the superficially wrapped coronary vessels.

In a possible implementation, the pixel value inside the blood vessel mask is 1, and the pixel value outside the blood vessel mask is 0.

The coronary blood vessel pixels adjacent to the pixels of the non-coronary blood vessels are the blood vessel wall pixels, and the coronary blood vessel wall pixels are used to characterize the coronary blood vessel wall in the image. The area formed by the non-coronary vascular pixels inside the vascular mask and outside the coronary vessel wall is called the extravascular area; the area formed by the coronary vascular pixels inside the vascular mask and inside the coronary vessel wall is called the blood vessel Inner area. In this way, the aforementioned image features can be extracted according to the distance based on the intravascular area and the extravascular area of the coronary artery subregion. A mask that coincides with the coronary vessel wall (for example, the blood vessel mask 3h from the midline point in Figure 4) can be used to assist in determining the extravascular area and the intravascular area. For example, for a blood vessel mask with a distance of 4h, the annular area between 3h and 4h in FIG. 4 is the extravascular area, and the area within 3h is the intravascular area.

Exemplarily, by superimposing the CTA image with the blood vessel mask, the area corresponding to the coronary sub-region can be determined in the CTA image; by superimposing the coronary blood vessel image or the heart image with the blood vessel mask, it can also be used in the coronary blood vessel image Or the area corresponding to the coronary sub-area is identified in the heart image. The region corresponding to each coronary subregion is taken as a range of image feature extraction, and the image feature extraction is performed according to the corresponding extravascular area and intravascular area.

In a possible implementation manner, the image feature that characterizes the myocardial bridge mentioned in step S13 may include: obtaining the CT in the extravascular area corresponding to each of the distances according to the blood vessel mask and the CTA image. The first percentage of the value distribution; according to the blood vessel mask and the heart image, obtain the second percentage of the heart pixel distribution in the extravascular area corresponding to each of the distances; according to the blood vessel mask and A cardiac image, obtaining a third percentage of the distribution of cardiac pixels in the intravascular area.

Exemplarily, a certain coronary artery subregion, a certain distance of the blood vessel mask is superimposed on the CTA image, and the pixel value of each point on the blood vessel mask is multiplied by the corresponding pixel value of each point in the CTA image (or with ) Operation to obtain a masked CTA image. The pixel value outside the blood vessel mask is 0, and the pixel value inside the blood vessel mask retains the original pixel value of the corresponding position on the original CTA image. Taking the blood vessel mask shown in Figure 4 as an example, the blood vessel mask at a distance of 4h and the CTA image can be ANDed to obtain a CTA image within 4h, and then take the image and the blood vessel mask at a distance of 3h The reverse phase "AND" can get the CTA image of the extravascular area. On the masked CTA image, the pixels within the blood vessel mask have at least one CT value. For the distance and the blood vessel mask of the coronary sub-region, obtain the percentage of the number of pixels of each CT value in the extravascular area within the blood vessel mask range to the number of pixels in the blood vessel mask. For example, in a CTA image, the number of pixels in a blood vessel mask is n, and the number of pixels with CT value x in the area outside the blood vessel mask is m, and the percentage corresponding to the CT value x is (m/n) %. These reflect the percentage values of the CT value pixels in the extravascular area of the coronary vascular subregion to the total pixels of the mask, and can reflect the distribution of CT values in the extravascular area. In the same way, for multiple blood vessel masks, the distribution of CT values in the extravascular area at various distances from the midline point can be obtained.

Exemplarily, the blood vessel mask is superimposed on the heart image, and the pixel value of each point on the blood vessel mask is multiplied (or ANDed) with the pixel value of each point in the heart image to obtain the heart image with the mask. The pixel value outside the blood vessel mask is 0, the heart pixel value inside the blood vessel mask is 1, and the remaining pixels are 0. For a certain distance, a certain coronary artery sub-region of the blood vessel mask, statistics in the extravascular area of the blood vessel mask (that is, within the mask range, the area outside the coronary vessel wall), the heart pixels (for example, the value is 1 pixel) the number of pixels accounted for the percentage of the total number of pixels inside the mask. For example, in a heart image, the number of pixels in a blood vessel mask is k, and the number of pixels with a heart pixel value of 1 in the area outside the blood vessel mask is h, and the corresponding percentage is (h/k)%. For multiple masks, obtain the percentage of the number of heart pixels at each distance from the line of the midline point to the total number of pixels in the mask. Image features that reflect the distribution of cardiac pixels in the extravascular area can be obtained. Based on a similar approach, it is possible to count the number of heart pixels (for example, pixels with a value of 1) in the intravascular area of the blood vessel mask (that is, the area within the mask and the coronary vessel wall). The percentage of the total number of pixels in the plate to obtain image characteristics reflecting the distribution of heart pixels in the intravascular area.

In this way, the above distributions obtained under different coronary subregions and different distance templates can be obtained, so that the position of the myocardial bridge can be determined from multiple dimensions, and the accuracy and efficiency of acquiring the image of the myocardial bridge can be improved.

In a possible implementation, the image features extracted in step S13 further include: imageomics features.

Generally, the imaging omics features are extracted through CT images. Imageomics features include: first-order statistical features, 3D-based image features, gray-level co-occurrence matrix, gray-level area size matrix, etc. The extraction tool is generally a software package, and there may be multiple programming languages for realizing the software package. The embodiment of the present disclosure does not limit the programming language for realizing the feature extraction function of imageomics.

Use a variety of image features to judge the position of the myocardial bridge, and the image features are mutually corroborated to improve the accuracy.

In step S14, the image features extracted in step S13 are input to the trained machine learning model, and the first region including the myocardial bridge in the multiple coronary subregions is determined to generate the myocardial bridge image.

Machine learning is a branch of artificial intelligence. Machine learning uses algorithms to enable machines to learn laws from a large amount of historical data, so as to intelligently identify or predict new samples. Use a lot of data and algorithms to train the machine, let the machine learn how to complete the task.

The machine learning model will contain some parameters, and the training of the machine learning model can be realized by tuning these parameters to make the prediction result of the machine learning model more accurate.

The training of the machine learning model is to use the machine learning model to predict the training data, compare the predicted value with the label value on the training data, obtain the difference between them (or called the loss value), and adjust the machine learning according to the difference The process of model parameters. Generally, the difference is represented by a loss function. After training, the machine learning model obtains optimized parameters.

In a possible implementation manner, the image features representing the myocardial bridge extracted from the CTA image sample, the coronary vascular image sample, and the heart image sample are input into the untrained machine learning model to obtain the second region containing the myocardial bridge; The degree of overlap between the second area and the third area is determined, where the third area represents a manually labeled myocardial bridge area; and the machine learning model is trained according to the degree of overlap.

Exemplarily, based on the CT image of the heart and the image of the coronary blood vessels, the myocardial bridge area can be marked by a professional physician. Here, the CT image of the heart and the image of coronary blood vessels are used to represent the same target object. Then, based on the location of these areas, a separate image is generated. The region representing the myocardial bridge included in the single image is defined as the third region. And, the position of the third area corresponds to the position of the myocardial bridge area on the CT image. The machine learning model determines the second area where the myocardial bridge is located according to the image characteristics of the image sample. The image samples here include: CTA image samples, coronary vascular image samples, and cardiac image samples. The image features may include the above-mentioned image features that characterize the myocardial bridge. Find the position corresponding to the second area in the CT image. The degree of overlap between the second area and the third area can be defined in a variety of ways. For example, the midline point in the second area can be Defined as a percentage. For example, if 50% of the midline points in the second area fall into the third area, the degree of overlap is 50%. The degree of overlap can also be defined in other ways, such as the percentage of the overlap area of the second region and the third region in the second region or the third region, which is not limited in this application. The above-mentioned difference or loss value can be determined according to the degree of overlap, and the parameters of the machine learning model can be updated to improve the accuracy of the machine learning model's prediction.

There are many types of machine learning models, each of which has its own characteristics, and has different prediction effects for different features and combinations of different features. Therefore, image features can be input into multiple machine learning models, and then the prediction results in each machine learning model can be obtained, and the accuracy of the prediction results can be compared to obtain a machine learning model suitable for prediction of myocardial bridge according to the image features in the embodiment of the present disclosure. .

In one possible implementation, the machine learning model can be random forest (Gradient Boost), gradient boosting decision tree (Random Forest), logistic regression (LR), support vector machine (Support Vector Machines, SVM), etc. The type of machine learning model is not limited in the embodiment of the present disclosure.

In a possible implementation manner, the image features of the image sample are input to multiple machine learning models for cross-validation, and a preferred machine learning model with a higher accuracy of the prediction result is selected.

Exemplarily, divide the midline point and line segment with the aforementioned image characteristics into 10 parts, choose 1 of the 10 midline point and line segments as the verification set, and use the remaining 9 as the test set, and each machine learning model performs 10 rounds each Prediction, so that each midline point line segment has a chance to be used as a verification set in each machine learning model. In addition, the parameters of the machine learning model are updated based on the predicted results. For the 10 rounds of prediction results of each machine learning model, the accuracy of the prediction can be the mean square error or other error value of the accuracy of each round of prediction results. Take the average of the accuracy of these 10 rounds of prediction results to obtain each machine The average value of the accuracy of the learning model's prediction results. Compare the average value of the accuracy of the prediction results of each machine learning model, and select a suitable machine learning model and parameters as the preferred machine learning model. The accuracy of the prediction result can be defined in many ways. For example, if the prediction result is whether each coronary artery subregion is the second region containing myocardial bridge, the accuracy of the prediction result can be determined by the second region and the corresponding third region. Measured by the degree of overlap, this application does not limit this.

In a possible implementation manner, the image features of the image sample can be input to the preferred machine learning model for cross-validation, and preferred image features that can improve the accuracy of the prediction result are selected.

Exemplarily, the first image feature (for example, the image feature representing the distribution of CT values in the extravascular area) of the image features of the sample image is temporarily added to the feature set, where the feature set is a set containing preferred image features. Use the image features in the feature set to predict the midline point and line segment in the preferred machine learning model in a cross-validation manner to obtain the first prediction result. The cross-validation process will not be repeated here. Since only one prediction result is obtained at this time, it can be understood that it is a good prediction result (for example, the degree of coincidence described above is high), so the first image feature is formally added to the feature set.

Next, add the second image feature (such as the feature representing the distribution of heart pixels in the extravascular area) temporarily into the feature set. At this time, the feature set contains the first image feature and the second image feature, continue to use the preferred machine learning The model is cross-validated and the second prediction result is obtained. Next, if the accuracy of the second prediction result is higher than that of the first prediction result, indicating that the addition of the second image feature makes the prediction result better, the second image feature is formally added to the feature set; otherwise, the second image feature is added to the feature set. The second image feature is deleted from the feature set.

By analogy, the image features of the sample image are temporarily added to the image feature set in turn, and cross-validation is performed until the number of features in the feature set reaches a preset threshold, and no more features are added to the feature set. Ensuring the addition of the n+1th feature in the image feature set can make the prediction result of the machine learning model better than the prediction result when the nth feature is added.

The image feature prediction in step S14 can be completed by using the preferred machine learning model and the preferred image features in the feature set to determine the first region including the myocardial bridge in the coronary artery subregion.

In a possible implementation, the image features extracted from the CTA image, coronary vascular image, and heart image of the same target object are input into the preferred machine learning model, and the preferred image features in the feature set are used for cross-validation. The midline point line segment is predicted and the neural optimization machine learning model is trained, and the prediction result of each midline point line segment in the verification set is output as the prediction result of the machine learning model. If the output result is 1, it means that the coronary artery subregion where the midline point line segment contains myocardial bridge; if the output result is 0, it means that the coronary artery subregion where the midline point line segment is located does not contain myocardial bridge to complete the alignment. Judgment of the position of the myocardial bridge in the image.

In a possible implementation manner, the midline point line segment corresponding to the first region containing the myocardial bridge is subjected to an expansion operation to obtain an image of the myocardial bridge.

Exemplarily, the midline point line segment whose output result of the preferred machine learning model is 1 is expanded within this range in the corresponding first region. The image obtained after the expansion operation is mapped to the coronary vascular image, and the pixels that fall into the coronary vascular area are retained. Calculate the number of pixels retained in each independent connected domain, and delete the independent connected domains whose pixel number is less than a preset threshold. The remaining independent connected domains constitute the myocardial bridge image. Among them, the expansion operation can be implemented based on the existing technology.

The first area is determined on the coronary blood vessel sub-region, and gives the range of the expansion operation, so that the obtained myocardial bridge image range is more accurate, and the pixels generated by the incorrect operation during the expansion operation are removed to improve the myocardial bridge image The accuracy rate.

According to the embodiment of the present disclosure, the coronary blood vessel area on the coronary blood vessel image is divided into multiple coronary sub-areas, and the image features representing the myocardial bridge are extracted from the CTA image and the heart image. A machine learning model is used to predict the location of the myocardial bridge based on image features. According to the prediction result, the image of the myocardial bridge is regenerated. In this way, the efficiency and accuracy of acquiring images of the myocardial bridge can be improved, and the manual workload can be reduced.

It should be noted that although the above-mentioned embodiment is taken as an example to introduce the method of obtaining myocardial bridge data, those skilled in the art can understand that the present disclosure should not be limited to this. In fact, the user can flexibly set each implementation manner according to personal preferences and/or actual application scenarios, as long as it conforms to the technical solution of the present disclosure.

Fig. 5 shows a block diagram of an apparatus for obtaining a myocardial bridge image according to an embodiment of the present disclosure. As shown in FIG. 5, the device 50 includes:

The image acquisition module 51 is configured to acquire CTA images, coronary vascular images, and cardiac images of the same target object, where the CTA image contains CT values;

The coronary artery sub-region dividing module 52 is used to divide the coronary vessel region into multiple coronary artery sub-regions according to the coronary artery blood vessel image;

The image feature extraction module 53 is configured to extract image features that characterize the myocardial bridge from the regions corresponding to the multiple coronary subregions in the CTA image and the heart image, and the regions corresponding to the multiple coronary subregions The intravascular area and the extravascular area including the coronary artery subregion, the image features respectively representing the distribution of CT values in the extravascular area, the distribution of cardiac pixels in the extravascular area, and the distribution of cardiac pixels in the extravascular area, and The distribution of heart pixels in the inner area;

The image generation module 54 is configured to input the image features into the trained machine learning model, and determine the first region containing the myocardial bridge in the multiple coronary sub-regions to generate the myocardial bridge image.

In a possible implementation manner, the regions corresponding to the multiple coronary artery subregions include regions with different distances from the line connecting the midline points, and the device for obtaining a myocardial bridge image further includes:

In a possible implementation manner, the image feature extraction module 53 is configured to obtain a first percentage of the CT value distribution in the extravascular area corresponding to each of the distances according to the blood vessel mask and the CTA image According to the blood vessel mask and the heart image, obtain the second percentage of the heart pixel distribution in the extravascular area corresponding to each of the distances; according to the blood vessel mask and the heart image, obtain the The third percentage of the distribution of cardiac pixels in the intravascular area.

In a possible implementation, the image generation module 54 is configured to perform an expansion operation on the midline point line segment corresponding to the first region containing the myocardial bridge to obtain an image of the myocardial bridge.

Fig. 6 is a block diagram showing a device 1900 for obtaining a myocardial bridge image according to an exemplary embodiment. For example, the apparatus 1900 may be provided as a server or terminal device. 5, the apparatus 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932, for storing instructions that can be executed by the processing component 1922, such as application programs. The application program stored in the memory 1932 may include one or more modules each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above-mentioned methods.

The device 1900 may also include a power component 1926 configured to perform power management of the device 1900, a wired or wireless network interface 1950 configured to connect the device 1900 to a network, and an input output (I/O) interface 1958. The device 1900 can operate based on an operating system stored in the memory 1932, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or the like.

An embodiment of the present disclosure also provides an electronic device, including: a processor; a memory for storing executable instructions of the processor; wherein the processor is configured to call the instructions stored in the memory to execute the above method. For structural examples of electronic equipment, please refer to the above-mentioned device 1900 for myocardial bridge image.

The embodiment of the present disclosure also proposes a non-volatile computer-readable storage medium on which computer program instructions are stored, characterized in that the computer program instructions implement the above-mentioned method when executed by a processor.

In an exemplary embodiment, a non-volatile computer-readable storage medium, such as the memory 1932 including the foregoing computer program instructions, can be executed by the processing component 1922 of the device 1900 to complete the foregoing methods.

The present disclosure may be a system, method and/or computer program product. The computer program product may include a computer-readable storage medium loaded with computer-readable program instructions for enabling a processor to implement various aspects of the present disclosure.

The computer-readable storage medium may be a tangible device that can hold and store instructions used by the instruction execution device. The computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) Or flash memory), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, mechanical encoding device, such as a printer with instructions stored thereon The protruding structure in the hole card or the groove, and any suitable combination of the above. The computer-readable storage medium used here is not interpreted as a transient signal itself, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (for example, light pulses through fiber optic cables), or through wires Transmission of electrical signals.

The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to various computing/processing devices, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers. The network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network, and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing/processing device .

The computer program instructions used to perform the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, status setting data, or in one or more programming languages. Source code or object code written in any combination, the programming language includes object-oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as "C" language or similar programming languages. Computer-readable program instructions can be executed entirely on the user's computer, partly on the user's computer, executed as a stand-alone software package, partly on the user's computer and partly executed on a remote computer, or entirely on the remote computer or server implement. In the case of a remote computer, the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect to the user’s computer) connect). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), can be customized by using the status information of the computer-readable program instructions. The computer-readable program instructions are executed to realize various aspects of the present disclosure.

Here, various aspects of the present disclosure are described with reference to flowcharts and/or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each block of the flowcharts and/or block diagrams, and combinations of blocks in the flowcharts and/or block diagrams, can be implemented by computer-readable program instructions.

These computer-readable program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine that makes these instructions when executed by the processor of the computer or other programmable data processing device , A device that implements the functions/actions specified in one or more blocks in the flowchart and/or block diagram is produced. It is also possible to store these computer-readable program instructions in a computer-readable storage medium. These instructions make computers, programmable data processing apparatuses, and/or other devices work in a specific manner, so that the computer-readable medium storing the instructions includes An article of manufacture, which includes instructions for implementing various aspects of the functions/actions specified in one or more blocks in the flowcharts and/or block diagrams.

It is also possible to load computer-readable program instructions onto a computer, other programmable data processing device, or other equipment, so that a series of operation steps are executed on the computer, other programmable data processing device, or other equipment to produce a computer-implemented process , So that the instructions executed on the computer, other programmable data processing apparatus, or other equipment realize the functions/actions specified in one or more blocks in the flowcharts and/or block diagrams.

The flowcharts and block diagrams in the accompanying drawings show the possible implementation architecture, functions, and operations of the system, method, and computer program product according to multiple embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram can represent a module, program segment, or part of an instruction, and the module, program segment, or part of an instruction contains one or more components for realizing the specified logical function. Executable instructions. In some alternative implementations, the functions marked in the block may also occur in a different order from the order marked in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, or they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and/or flowchart, and the combination of the blocks in the block diagram and/or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or actions Or it can be realized by a combination of dedicated hardware and computer instructions.

The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Without departing from the scope and spirit of the described embodiments, many modifications and changes are obvious to those of ordinary skill in the art. The choice of terms used herein is intended to best explain the principles, practical applications, or technical improvements to technologies in the market for each embodiment, or to enable those of ordinary skill in the art to understand the various embodiments disclosed herein.

Claims

A method for obtaining a myocardial bridge image, characterized in that it comprises:

Acquiring a CTA image, a coronary vascular image, and a heart image of the same target object, the CTA image containing CT values;

Divide the coronary vascular area into multiple coronary sub-areas according to the coronary vascular image;

Extract image features that characterize the myocardial bridge from the regions corresponding to the multiple coronary subregions in the CTA image and the heart image, and the regions corresponding to the multiple coronary subregions include the intravascular in the coronary subregions Area and extravascular area, the image features respectively representing the distribution of CT values in the extravascular area, the distribution of cardiac pixels in the extravascular area, and the distribution of cardiac pixels in the intravascular area ；

The image features are input to the trained machine learning model, and the first region containing the myocardial bridge in the multiple coronary subregions is determined to generate the myocardial bridge image.
The method for obtaining a myocardial bridge image according to claim 1, wherein the dividing the coronary blood vessel area into coronary sub-areas according to the coronary blood vessel image comprises:

According to the coronary blood vessel image, extract the midline point connecting the midline points representing the coronary blood vessels;

Dividing the midline point connection into midline point line segments according to the connectivity of the midline point connection;

According to the first threshold, divide the midline point line segment into midline point sub-line segments;

According to the midline point sub-line segment, the coronary blood vessel area is divided into coronary artery sub-areas.
The method for obtaining a myocardial bridge image according to claim 2, wherein the regions corresponding to the multiple coronary artery subregions include regions with different distances from the line connecting the midline points,

The method also includes:

According to the coronary vascular image, generating a vascular mask corresponding to each coronary sub-region according to different distances from the line of the midline point;

According to the blood vessel mask, determine the regions corresponding to the multiple coronary subregions in the CTA image and the heart image,

Among them, the area located in the blood vessel mask and outside the coronary vessel wall is the extravascular area, and the area located in the blood vessel mask, and the area inside the coronary vessel wall is the intravascular area.
The method for obtaining a myocardial bridge image according to claim 3, wherein:

In the CTA image and the heart image in the regions corresponding to the multiple coronary artery subregions, extracting the image features that characterize the myocardial bridge includes:

Obtaining, according to the blood vessel mask and the CTA image, a first percentage of the CT value distribution in the extravascular area corresponding to each of the distances;

Obtaining, according to the blood vessel mask and the heart image, a second percentage of the distribution of cardiac pixels in the extravascular area corresponding to each of the distances;

According to the blood vessel mask and the heart image, a third percentage of the distribution of heart pixels in the intravascular area is obtained.
The method for obtaining a myocardial bridge image according to claim 1, wherein the image features are input to a trained machine learning model to determine the first region containing the myocardial bridge among the plurality of coronary subregions To generate images of myocardial bridge, including:

The midline point line segment corresponding to the first region containing the myocardial bridge is expanded to obtain an image of the myocardial bridge.
The method for obtaining a myocardial bridge image according to claim 2, wherein the method further comprises:

Input the image features characterizing the myocardial bridge extracted from the CTA image sample, the coronary blood vessel image sample, and the heart image sample into the untrained machine learning model to obtain the second region containing the myocardial bridge;

Determining the degree of overlap between the second area and the third area, where the third area represents an artificially labeled myocardial bridge area;

According to the degree of overlap, the machine learning model is trained.
The method for obtaining a myocardial bridge image according to any one of claims 1 to 6, characterized in that,

The CTA image, coronary blood vessel image, and heart image are 3D images.
A device for obtaining an image of a myocardial bridge is characterized in that it comprises:

An image acquisition module for acquiring CTA images, coronary vascular images, and heart images of the same target object, the CTA images containing CT values;

The coronary artery sub-areas division module is used to divide the coronary blood vessel area into multiple coronary artery sub-areas according to the coronary blood vessel image;

The image feature extraction module is used to extract image features that characterize the myocardial bridge in the regions corresponding to the multiple coronary subregions in the CTA image and the heart image, and the regions corresponding to the multiple coronary subregions include The intravascular area and the extravascular area of the coronary subregion, the image features respectively representing the distribution of CT values in the extravascular area, the distribution of cardiac pixels in the extravascular area, and the intravascular area The distribution of heart pixels in the area;

The image generation module is configured to input the image features into the trained machine learning model, and determine the first region containing the myocardial bridge in the multiple coronary sub-regions to generate the myocardial bridge image.
An electronic device, characterized in that it comprises:

processor;

A memory for storing processor executable instructions;

Wherein, the processor is configured to call instructions stored in the memory to execute the method according to any one of claims 1 to 7.
A non-volatile computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions implement the method according to any one of claims 1 to 7 when the computer program instructions are executed by a processor.