An object volume determination method, system, device and storage medium

Abstract

The application discloses an object volume determination method, system, device and storage medium, which is applied to the technical field of network and comprises the following steps: after obtaining three-dimensional body data of an object to be analyzed, determining a segmentation direction of the three-dimensional body data; along the segmentation direction, performing N times of segmentation on the three-dimensional body data by N normal planes to obtain N section images; based on a labeling instruction, performing contour labeling on the object to be analyzed of at least one section image; performing feature analysis on the contour of each section image which has been subjected to the contour labeling, and determining the contour in each section image of the remaining section images based on the obtained feature analysis result; based on the contour in the N section images, obtaining the contour in the section image of each pixel position along the segmentation direction as obtained contour information of the object to be analyzed, so as to determine the volume of the object to be analyzed. According to the scheme, the volume of the object to be analyzed can be conveniently and effectively determined, and the determination is relatively accurate.

Description

Technical Field

[0001] This invention relates to the field of network technology, and in particular to a method, system, device, and storage medium for determining the volume of an object. Background Technology

[0002] In the field of network technology, irregular spatial volumes are sometimes constructed, and it is necessary to determine the volume of these irregular volumes. For example, with the widespread application of ultrasound, it is necessary to perform volume analysis on different tissues and organs such as tumors, follicles, uterus, and bladder to calculate information such as volume, grayscale, and blood flow.

[0003] In traditional methods, a common approach to volume analysis of 3D volumes is the rotational contour method. Specifically, this involves selecting a rotation axis, rotating the analysis target to obtain different cross-sections of the lesion tissue, drawing a contour on these rotated cross-sections, and finally reconstructing and filling the volume to obtain the 3D volume. This method requires the user to define and delineate the tissue information for different cross-sections, making it complex, labor-intensive, and unsuitable for rapid analysis. Furthermore, the accuracy of this method depends on the user's setting of the rotation angle, which can lead to significant volume errors in some cases.

[0004] In conclusion, how to conveniently, effectively, and accurately determine the volume of the object to be analyzed is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, device, and storage medium for determining the volume of an object, so as to conveniently, effectively, and accurately determine the volume of the object to be analyzed.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for determining the volume of an object, comprising:

[0008] After obtaining the three-dimensional volume data of the object to be analyzed, the segmentation direction of the three-dimensional volume data is determined;

[0009] Along the segmentation direction, the three-dimensional volume data is segmented N times using N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction;

[0010] Based on the received annotation instructions, at least one of the N cross-sectional images is used to annotate the contour of the object to be analyzed; where N is a positive integer greater than 1.

[0011] Feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated, and the contours of the object to be analyzed in the remaining cross-sectional images are determined based on the obtained feature analysis results.

[0012] Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded, and these contours are used as the obtained contour information of the object to be analyzed.

[0013] Based on the contour information of the object to be analyzed, the volume of the object to be analyzed is determined.

[0014] In one implementation, based on received annotation instructions, contour annotation of the object to be analyzed is performed on at least one of the N cross-sectional images, including:

[0015] Based on the received annotation instructions, at least 3 of the N cross-sectional images are used to annotate the contours of the object to be analyzed.

[0016] Among them, the cross-sectional images with contour annotations include at least the first cross-sectional image, the m-th cross-sectional image, and the N-th cross-sectional image; m is a positive integer;

[0017] The first cross-sectional image represents the cross-sectional image segmented at the first segmentation point in the segmentation direction.

[0018] The Nth cross-sectional image represents the cross-sectional image segmented at the last segmentation point in the segmentation direction;

[0019] The m-th cross-sectional image represents the cross-sectional image segmented at the middle segmentation point in the segmentation direction.

[0020] In one implementation, feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated, and the contours of the object to be analyzed in the remaining cross-sectional images are determined based on the obtained feature analysis results, including:

[0021] Feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated to obtain the feature analysis results of each of the cross-sectional images that have been contour-annotated.

[0022] For each cross-sectional image without contour annotation, the contour of the object to be analyzed in the current cross-sectional image is determined based on the feature analysis results of the two adjacent cross-sectional images with contour annotation, and the similarity of the contour regions.

[0023] In one implementation, for each of the unannotated cross-sectional images, the contour of the object to be analyzed in the current cross-sectional image is determined based on the similarity of the contour regions using the feature analysis results of two adjacent annotated cross-sectional images. This includes:

[0024] For each of the cross-sectional images that have not been contour-annotated, the initial contour of the object to be analyzed in the current cross-sectional image is obtained by interpolation calculation based on the contours of the two adjacent cross-sectional images that have been contour-annotated.

[0025] The reference feature analysis results of this cross-sectional image are obtained by analyzing the features of two adjacent cross-sectional images with contour annotations.

[0026] The initial contour is dynamically adjusted. When, during the adjustment process, it is determined that the feature analysis result obtained under the current contour is similar to the reference feature analysis result, the current contour is taken as the contour of the object to be analyzed in the current cross-sectional image.

[0027] In one implementation, after obtaining the initial contour, the method further includes:

[0028] Based on the initial contour, a contour analysis region is established;

[0029] Wherein, the outer contour of the contour analysis region encloses the initial contour, and the initial contour encloses the inner contour of the contour analysis region;

[0030] Accordingly, the initial contour is dynamically adjusted, including:

[0031] The initial contour is dynamically adjusted according to the principle that no pixel in the adjusted contour exceeds the contour analysis area.

[0032] In one implementation, based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded to obtain the contour information of the object to be analyzed, including:

[0033] Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded by interpolation or fitting, and are used as the contour information of the object to be analyzed.

[0034] In one implementation, it further includes:

[0035] After determining the outline of the object to be analyzed in N cross-sectional images, K connection points are set on the outline of each of the N cross-sectional images.

[0036] For any two adjacent cross-sectional images among the N cross-sectional images, the connection points of the two adjacent cross-sectional images are connected according to the connection rule that each connection point must participate in the connection, so as to establish multiple non-overlapping triangular patches between the two adjacent cross-sectional images, thereby constructing the patch model of the object to be analyzed.

[0037] In one implementation, determining the segmentation direction of the three-dimensional volume data includes:

[0038] The three-dimensional volume data of the object to be analyzed is displayed on the display interface;

[0039] When a dividing line drawn by the user on the display interface is detected, the dividing direction of the three-dimensional volume data is determined based on the starting and ending positions of the dividing line.

[0040] The three-dimensional volume data of the object to be analyzed displayed on the display interface is rotated and / or translated so that the segmentation direction is parallel to the specified axis of the display interface, and the midpoint of the segmentation line coincides with the center point of the display interface.

[0041] In a second aspect, the present invention provides a system for determining the volume of an object, comprising:

[0042] The segmentation direction determination module is used to determine the segmentation direction of the three-dimensional volume data after obtaining the three-dimensional volume data of the object to be analyzed.

[0043] The segmentation module is used to segment the three-dimensional volume data N times along the segmentation direction using N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction.

[0044] The contour annotation module is used to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images based on the received annotation instructions; where N is a positive integer greater than 1.

[0045] The contour prediction module is used to perform feature analysis on the contours of each of the cross-sectional images that have been contour-annotated, and to determine the contours of the object to be analyzed in the remaining cross-sectional images based on the obtained feature analysis results.

[0046] The contour information determination module is used to obtain the contour of the object to be analyzed in the cross-sectional image at each pixel position along the segmentation direction based on the contour of the object to be analyzed in N cross-sectional images, and use it as the obtained contour information of the object to be analyzed.

[0047] The volume calculation module is used to determine the volume of the object to be analyzed based on the contour information of the object.

[0048] Thirdly, the present invention provides an object volume determination device, comprising:

[0049] Memory, used to store computer programs;

[0050] A processor for executing the computer program to implement the steps of the object volume determination method as described above.

[0051] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the object volume determination method as described in any of the preceding claims.

[0052] By applying the technical solution provided in the embodiments of the present invention, after obtaining the three-dimensional volume data of the object to be analyzed, it is necessary to determine the segmentation direction of the three-dimensional volume data. Subsequently, along the segmentation direction, the three-dimensional volume data will be segmented N times by N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction.

[0053] For these N cross-sectional images, the user only needs to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images. Subsequently, the system can automatically perform feature analysis on the contours of each annotated cross-sectional image. Since the image features of the contours in different cross-sectional images are similar, the contours of the object to be analyzed in the remaining cross-sectional images can be determined based on the obtained feature analysis results. Finally, because of the similarity of the contours in different cross-sectional images, based on the contours of the object to be analyzed in the N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction can be obtained as the contour information of the object to be analyzed. Having obtained the contour information of the object to be analyzed, the volume of the object to be analyzed can be determined.

[0054] As can be seen, the solution in this application only requires the user to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images, making the operation simple. Furthermore, it automatically determines the volume of the object to be analyzed, eliminating the accuracy issues caused by inappropriate rotation angle settings found in traditional solutions. Therefore, in summary, the solution in this application can conveniently, effectively, and relatively accurately determine the volume of the object to be analyzed. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a flowchart illustrating the implementation of a method for determining the volume of an object according to a specific embodiment of the present invention.

[0057] Figure 2a This is a schematic diagram of a dividing line drawn by a user on a display interface in a specific embodiment of the present invention.

[0058] Figure 2b This is a schematic diagram showing the position of the dividing line after three-dimensional volume data movement in one specific embodiment of the present invention;

[0059] Figure 2c This is a schematic diagram illustrating N-fold segmentation of three-dimensional volume data in a specific embodiment of the present invention;

[0060] Figure 3 This is a schematic diagram showing the internal pixels of the outline of the object to be analyzed in a cross-sectional image after being marked, according to a specific embodiment of the present invention.

[0061] Figure 4 This is a schematic diagram illustrating the setting of connection points on the outline of the object to be analyzed in a cross-sectional image according to a specific embodiment of the present invention.

[0062] Figure 5 This is a schematic diagram of the structure of an object volume determination system provided in a specific embodiment of the present invention;

[0063] Figure 6 This is a schematic diagram of the structure of an object volume determination device provided in a specific embodiment of the present invention. Detailed Implementation

[0064] The core of this invention is to provide a method for determining the volume of an object, which can conveniently, effectively, and relatively accurately determine the volume of the object to be analyzed.

[0065] To enable those skilled in the art to better understand the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of a method for determining the volume of an object according to a specific embodiment of the present invention. The method for determining the volume of an object may include the following steps:

[0067] Step S101: After obtaining the three-dimensional volume data of the object to be analyzed, determine the segmentation direction of the three-dimensional volume data.

[0068] Specifically, each step of the object volume determination method of this application can be executed by a computer.

[0069] In practical applications, the object to be analyzed can be a three-dimensional structure with an irregular shape, such as a tumor, follicle, uterus, or bladder. After detecting the object using equipment such as ultrasound, CT (Computed Tomography), MRI, or endoscopy, three-dimensional volume data of the object to be analyzed is obtained, i.e., a three-dimensional image of the object. A three-dimensional image can be considered as a stack of multiple two-dimensional images. It is understandable that the three-dimensional volume data of the object to be analyzed includes not only the object itself but also its surrounding objects. Therefore, it is necessary to extract the three-dimensional contour of the object to be analyzed in order to perform volume calculations.

[0070] In this regard, based on the principle of the present application, it is necessary to segment the three-dimensional volume data of the object to be analyzed. Before segmentation, it is necessary to determine the segmentation direction of the three-dimensional volume data. There are many ways to determine the segmentation direction. For example, a direction can be randomly specified as the segmentation direction. Or, after detecting the three-dimensional volume data of the object to be analyzed, any direction passing through the center point can be used as the segmentation direction of the three-dimensional volume data.

[0071] Furthermore, in one specific embodiment of the present invention, step S101 may include:

[0072] The three-dimensional volume data of the object to be analyzed is displayed on the display interface;

[0073] When a dividing line drawn by the user on the display interface is detected, the dividing direction of the three-dimensional volume data is determined based on the starting and ending positions of the dividing line.

[0074] The three-dimensional volume data of the object to be analyzed displayed on the display interface will be rotated and / or translated so that the segmentation direction is parallel to the specified axis of the display interface and the midpoint of the segmentation line coincides with the center point of the display interface.

[0075] In this implementation, the three-dimensional volume data of the object to be analyzed is first displayed on the display interface. Then, the user is allowed to draw dividing lines on the display interface, which serve as the dividing direction for the three-dimensional volume data. This setup is chosen because the proposed solution requires dividing the three-dimensional volume data along these dividing directions to subsequently determine the contour information of the object to be analyzed. Due to the irregular shape of the object, different dividing directions can affect the accuracy of volume determination to some extent. For example, a suitable dividing direction is along the long axis of the object to be analyzed.

[0076] Therefore, in this embodiment, allowing users to draw dividing lines on the display interface helps improve the accuracy of volume determination and also ensures the flexibility of implementation of the solution in this application.

[0077] Furthermore, this implementation method will rotate and / or translate the three-dimensional volume data so that after the movement, the segmentation direction is parallel to the specified axis of the display interface, and the midpoint of the segmentation line coincides with the center point of the display interface, which is also beneficial for users to view it better.

[0078] For easier understanding, please refer to the following: Figure 2a and Figure 2b , Figure 2a This is a diagram illustrating the dividing lines drawn by the user on the display interface. (pt) s The starting position of this dividing line is pt. e The endpoint of this dividing line determines the segmentation direction of the 3D volume data. The length of the dividing line is expressed in pt. s and pt e The specific location can be flexibly set by the user.

[0079] Then the 3D volume data needs to be rotated and translated, so the dividing lines will also move accordingly. Figure 2b This shows the position of the dividing line after the 3D volume data has been moved. It can be seen that the dividing direction is now parallel to the specified axis of the display interface. In this example, the specified axis is the x-axis of the display interface, and the midpoint of the dividing line coincides with the center point of the display interface. In other implementations, for certain types of objects to be analyzed, the specified axis can be set to the y-axis of the display interface, or it can be set by the user in the display settings. Figure 2a and Figure 2b In this code, the center point of the display interface is denoted as pt0, and the midpoint of the dividing line is denoted as pt. m .

[0080] Step S102: Along the segmentation direction, the three-dimensional volume data is segmented N times by N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction.

[0081] In this application, the 3D volume data needs to be segmented N times along the segmentation direction. The specific value of N and the specific positions of the N segmentations can be set according to actual needs. For example, in practical applications, N segmentation points can usually be evenly set on the segmentation line, with the starting position and ending position of the segmentation line being the first and last segmentation points, respectively. For any segmentation point, a vertical plane passing through the segmentation point and perpendicular to the segmentation line can be used to segment the 3D volume data once, obtaining a cross-sectional image perpendicular to the segmentation direction.

[0082] See also Figure 2c This is a schematic diagram illustrating N-fold segmentation of three-dimensional volume data in one specific implementation method. Figure 2c In the example, N dividing points are evenly set on the dividing line to achieve N divisions of the three-dimensional volume data.

[0083] Step S103: Based on the received annotation instructions, perform contour annotation of the object to be analyzed on at least one of the N cross-sectional images; where N is a positive integer greater than 1.

[0084] It is understandable that although the object to be analyzed is irregular, the image features of the outline of the object to be analyzed are usually different from those of its surroundings in any cross-sectional image. That is, the pixel values ​​on the outline line are usually different from the pixel values ​​of the surrounding pixels, so that when the user observes the cross-sectional image with the naked eye, the outline of the object to be analyzed in the cross-sectional image can be determined.

[0085] In this application, the user needs to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images. For example, the user can select a cross-sectional image on the display interface, and the display interface can then show that cross-sectional image. The user can then draw the contour on the cross-sectional image by dragging the mouse or by other means. The front end can detect the user's operation and generate an annotation instruction, which is then sent to the back end processor. This allows the processor to annotate the contour of the object to be analyzed in the cross-sectional image based on the annotation instruction, i.e., based on the user's operation.

[0086] Furthermore, it's understandable that the specific number of cross-sectional images used for contour annotation can be set as needed. The more images annotated, the better the accuracy of the contours of other determined cross-sectional images is ensured during subsequent operations, thus guaranteeing the accuracy of the determined volume. Of course, this also increases the amount of user work. Therefore, in practical applications, selecting only a few cross-sectional images for contour annotation is sufficient. It should also be emphasized that for users, selecting several cross-sectional images for contour annotation is not complicated and is very simple to perform.

[0087] In one specific embodiment of the present invention, step S103 may specifically include:

[0088] Based on the received annotation instructions, at least 3 of the N cross-sectional images are used to annotate the contours of the object to be analyzed.

[0089] Among them, the cross-sectional images with contour annotations include at least the first cross-sectional image, the m-th cross-sectional image, and the N-th cross-sectional image; m is a positive integer;

[0090] The first cross-sectional image represents the cross-sectional image segmented at the first segmentation point in the segmentation direction;

[0091] The Nth slice image represents the slice image segmented at the last segmentation point in the segmentation direction.

[0092] The m-th slice image represents the slice image segmented at the midpoint of the segmentation direction.

[0093] This implementation takes into account that the purpose of annotating the outline of the object to be analyzed in at least one of the N cross-sectional images is to use this as a reference to obtain the outline of the object to be analyzed in the remaining cross-sectional images. Therefore, in order to avoid annotating too many cross-sectional images and to ensure the accuracy of subsequent steps, several cross-sectional images should be selected in a relatively dispersed manner for annotating the outline of the object to be analyzed.

[0094] For example, in this implementation, the user needs to annotate the outline of the object to be analyzed in at least 3 of the N cross-sectional images. These three cross-sectional images are the cross-sectional images segmented at the first segmentation point, the middle segmentation point, and the last segmentation point in the segmentation direction, respectively. Subsequently, the outline of the object to be analyzed in the remaining cross-sectional images can be obtained with relatively high accuracy by using these 3 cross-sectional images as references.

[0095] It should also be noted that the m-th slice image corresponds to the middle segmentation point in the segmentation direction. m can be equal to N / 2. In practical applications, when N / 2 is not an integer, it can be rounded down or up to get the value of m.

[0096] For example, regarding the above text Figure 2c For example, the starting position pt on the dividing line. s End position pt e and the midpoint position pt m In this implementation method, the outline of the object to be analyzed is annotated in the segmented cross-sectional image of the location.

[0097] Step S104: Perform feature analysis on the contours of each cross-sectional image with contour annotation, and determine the contours of the object to be analyzed in the remaining cross-sectional images based on the obtained feature analysis results.

[0098] When performing feature analysis on the contours of each cross-sectional image with contour annotation, the specific feature analysis method can be set and adjusted according to actual needs. One feature analysis method or a combination of multiple feature analysis methods can be used.

[0099] For example, feature analysis of the contours of each cross-sectional image with contour annotation can be performed using one or more methods in deep learning, such as grayscale statistics, edge extraction, corner extraction, and image moment calculation.

[0100] Furthermore, in practical applications, after performing feature analysis on the contour of the object to be analyzed in any cross-sectional image, the feature analysis results of the image can usually be represented in the form of a vector. This allows for the convenient insertion of results obtained from different feature analysis methods into the vector, and also facilitates subsequent similarity comparisons.

[0101] Taking grayscale statistics as an example, in a specific implementation, when performing grayscale statistical feature analysis on the contour of any cross-sectional image, the grayscale of the cross-sectional image can first be histogram equalized to make the image brightness distribution uniform. For example, in some cases, the grayscale of a cross-sectional image is generally between 120 and 150, making the grayscale difference between the contour position and the non-contour position small. After the histogram equalization operation, the grayscale of the entire cross-sectional image will be distributed in the grayscale range of 0 to 255, which is conducive to increasing the grayscale difference between the contour position and the non-contour position, so that the subsequent grayscale statistical feature analysis results can more accurately reflect the contour position of the cross-sectional image.

[0102] For any cross-sectional image with contour annotation, when performing feature analysis using grayscale statistics, for example, the mean grayscale value within the contour lines can be calculated as the first parameter in the feature analysis result of the cross-sectional image. Alternatively, the variance of grayscale values ​​within the contour lines can be calculated as the second parameter in the feature analysis result of the cross-sectional image. Furthermore, the difference between the maximum and minimum grayscale values ​​within the contour lines can be calculated as the third vector parameter in the feature analysis result of the cross-sectional image.

[0103] For example, when performing feature analysis on the contour of the cross-sectional image, specific parameters such as geometric moments, central moments, and Hu moments can be calculated and used as corresponding parameters in the feature analysis results of the cross-sectional image.

[0104] For example, when performing edge extraction feature analysis on the contour of the cross-sectional image, specific methods may include using Canny, high-pass filters, or edge detection methods based on phase consistency to obtain edge extraction feature analysis results that can reflect the boundary information of the contour, which can then be used as corresponding parameters in the feature analysis results of the cross-sectional image.

[0105] When performing feature analysis on contours based on deep learning, specific methods such as convolutional neural networks, generative adversarial networks, autoencoders, variational autoencoders, attention mechanisms, and visual transformers can be used to extract image features at different levels, obtain the feature analysis results of deep learning, and use them as the corresponding parameters in the feature analysis results of the cross-sectional image.

[0106] After performing feature analysis on the contours of each cross-sectional image with contour annotation, the feature analysis results of each cross-sectional image can be obtained. Then, based on these feature analysis results, the contours of the object to be analyzed in the remaining cross-sectional images need to be determined, that is, the contours of the object to be analyzed in each of the N cross-sectional images that have not been annotated with the contours of the object to be analyzed.

[0107] In practical implementation, based on these feature analysis results, contour detection can be performed on each of the N cross-sectional images where the contour is not determined, so that the detected contour can conform to the contour features reflected by these feature analysis results. Of course, there are many algorithms that can be used to perform contour detection on each cross-sectional image where the contour is not determined. For example, template matching, deep learning, etc., can be used to determine the contour of the object to be analyzed in each cross-sectional image that has not been contour-annotated.

[0108] In one specific embodiment of the present invention, step S104 may specifically include:

[0109] Step 1: Perform feature analysis on the contours of each cross-sectional image with contour annotation to obtain the feature analysis results of each cross-sectional image with contour annotation.

[0110] Step 2: For each cross-sectional image without contour annotation, the contour of the object to be analyzed in the current cross-sectional image is determined based on the feature analysis results of the two adjacent cross-sectional images with contour annotation, and the similarity of the contour regions.

[0111] In this implementation, after obtaining the feature analysis results of each contour-annotated cross-sectional image, when determining the contour of the object to be analyzed for each unannotated cross-sectional image based on the similarity of the contour regions, the feature analysis results of the two adjacent contour-annotated cross-sectional images of the current cross-sectional image are specifically used. This is because, theoretically, for the current cross-sectional image, the feature analysis results of any two or more contour-annotated cross-sectional images from the aforementioned steps can all be used to obtain the contour of the object to be analyzed in the current cross-sectional image based on the similarity of the contour regions. However, in this implementation, the feature analysis results of the two adjacent contour-annotated cross-sectional images of the current cross-sectional image are specifically used, that is, the feature analysis results of the two closest contour-annotated cross-sectional images on both sides of the current cross-sectional image are used. This helps to ensure the accuracy of the obtained contour of the object to be analyzed in the current cross-sectional image.

[0112] Furthermore, in one specific embodiment of the present invention, step two may specifically include:

[0113] For each cross-sectional image without contour annotation, the initial contour of the object to be analyzed in the current cross-sectional image is obtained by interpolation calculation based on the contours of the two adjacent cross-sectional images with contour annotation.

[0114] The reference feature analysis results of this cross-sectional image are obtained by analyzing the features of two adjacent cross-sectional images with contour annotations.

[0115] The initial contour is dynamically adjusted. When the feature analysis result obtained under the current contour is similar to the reference feature analysis result during the adjustment process, the current contour is taken as the contour of the object to be analyzed in the current cross-sectional image.

[0116] Specifically, in this implementation, the initial contour is first obtained through interpolation calculation. For ease of understanding, let's take the example of annotating the contour of the object to be analyzed in the first, m, and Nth cross-sectional images from N cross-sectional images described above. For example, N=10, with 10 evenly spaced dividing points. The contour of the object to be analyzed in the first cross-sectional image is, for example, a circle with a radius of 10cm, where m equals 5 in this example. The contour of the object to be analyzed in the fifth cross-sectional image is, for example, a circle with a radius of 6cm, and the contour of the object to be analyzed in the tenth cross-sectional image is, for example, a circle with a radius of 2cm.

[0117] In this example, for the first to fourth cross-sectional images, the contour of the object to be analyzed in the corresponding cross-sectional image can be determined based on the feature analysis results of the first cross-sectional image and the m-th cross-sectional image. Similarly, for the sixth to ninth cross-sectional images, the contour of the object to be analyzed in the corresponding cross-sectional image can be determined based on the feature analysis results of the m-th cross-sectional image and the N-th cross-sectional image.

[0118] Taking the second cross-sectional image as an example, based on the outline of the object to be analyzed in the first cross-sectional image and the outline of the object to be analyzed in the m-th cross-sectional image, the initial outline of the object to be analyzed in the second cross-sectional image can be obtained by interpolation calculation. It is a circle with a radius of 10 + (6 - 10) × (1 / 4) = 9 cm. Similarly, the initial outline of the object to be analyzed in the third cross-sectional image is a circle with a radius of 10 + (6 - 10) × (2 / 4) = 8 cm, and the initial outline of the object to be analyzed in the fourth cross-sectional image is a circle with a radius of 10 + (6 - 10) × (3 / 4) = 7 cm.

[0119] Taking the second cross-sectional image as an example, after obtaining a circle with a radius of 9cm as the initial contour, it is understandable that the initial contour is not accurate. That is, there is still a certain degree of difference between the initial contour and the actual contour of the object to be analyzed in the cross-sectional image. At this time, in this implementation method, the initial contour will be dynamically adjusted so that the adjusted contour is close to the actual contour. The judgment criterion is based on the similarity between the feature analysis results.

[0120] Specifically, based on the feature analysis results of the first and fifth cross-sectional images, the reference feature analysis results of the second cross-sectional image can be obtained. It can be understood that the reference feature analysis results represent the theoretical feature analysis results of the second cross-sectional image inferred from the feature analysis results of adjacent cross-sectional images.

[0121] When obtaining the reference feature analysis results, a simple and convenient approach is to weightedly sum the parameters from the feature analysis results of two adjacent contour-annotated cross-sectional images to obtain the corresponding parameters in the reference feature analysis results. Furthermore, as described above, the feature analysis results are typically in vector form; for example, the feature analysis results of the first cross-sectional image can be represented as a vector (a...). 11 a 12 a 13 , ..., a 1（z-1） a 1z ), z is the total number of parameters in the vector, and correspondingly, the feature analysis result of the m-th cross-section image is represented as a vector (a m1 a m2 a m3 , ..., a m（z-1） a mz The reference feature analysis results of the obtained second-section image are represented as vectors (a 21 a 22 a 23 , ..., a 2（z-1） a 2z ).

[0122] Taking the weighted summation of the first parameter in the feature analysis results as an example, the weighted summation operation can be represented as a 21 =k1×a 11 +k2×a m1 k1 and k2 are weighting coefficients, and similarly, a 22 =k3×a 12 +k4×a m2K3 and k4 are weighting coefficients. It's understandable that the weighting coefficients need to be set adaptively for different parameter items. For example, the first parameter represents the average gray level of each pixel within the contour. In different cross-sectional images, although the contours of the object being analyzed differ, the average gray level of each pixel within the contour is roughly the same. Therefore, k1 and k2 can both be set to 0.5. This means that for this parameter item in the feature analysis result, the value of this parameter item in the reference feature analysis result is obtained by averaging.

[0123] For example, the first parameter represents the total number of pixels inside the contour. It's understandable that the contour of the object being analyzed differs in different cross-sectional images, and the size of the contour varies systematically. Therefore, the values ​​of k1 and k2 can be set using interpolation. Since this is the result of the reference feature analysis of the second cross-sectional image, the first cross-sectional image is closer to the second cross-sectional image than the m-th cross-sectional image. Therefore, k1 should be larger and k2 smaller, so that through a... 21 =k1×a 11 +k2×a m1 Get a 21 At that time, the feature analysis results of the first cross-section image have a greater impact. Furthermore, it can be seen that the principle of setting the values ​​of relevant parameters through interpolation here is similar to the principle of obtaining the initial contour of the corresponding cross-section image based on interpolation mentioned above.

[0124] After obtaining the reference feature analysis results, it is necessary to compare the feature analysis results obtained under the current contour condition with the reference feature analysis results during the dynamic adjustment of the initial contour. It is understandable that during the dynamic adjustment of the initial contour, the feature analysis results of the current contour also change as the contour changes. If the feature analysis results obtained under the current contour condition are similar to the reference feature analysis results, then the current contour can be used as the contour of the object to be analyzed in this cross-sectional image.

[0125] There are several ways to compare the similarity between the feature analysis results obtained under the current contour condition and the reference feature analysis results. For example, by summing the errors of each parameter, and if the sum of errors is below a specified threshold, the feature analysis results under the current contour condition can be considered similar to the reference feature analysis results. Of course, considering the large differences in the values ​​of different parameters, if this method is used, in practical applications, it is usually necessary to normalize the parameters or sum the errors after weighting them. These settings and adjustments can be made according to actual needs.

[0126] When dynamically adjusting the initial contour, there can be multiple adjustment methods, such as random adjustment. Of course, other more efficient methods can be set in other implementations. For example, to ensure efficiency, if the feature analysis result obtained after an adjustment is closer to the reference feature analysis result, the adjustment is retained; otherwise, the adjustment is withdrawn.

[0127] Furthermore, in one specific embodiment of the present invention, after obtaining the initial contour, the process may further include:

[0128] Based on the initial contour, establish the contour analysis region;

[0129] The outer contour of the contour analysis area encloses the initial contour, and the initial contour encloses the inner contour of the contour analysis area.

[0130] Accordingly, the initial contour is dynamically adjusted, including:

[0131] The initial contour is dynamically adjusted according to the principle that no pixel in the adjusted contour exceeds the contour analysis area.

[0132] This implementation takes into account that although the object to be analyzed is usually an irregular object, the contour of the object to be analyzed in different cross-sectional images still has a certain regularity. In addition, considering that the above implementation requires dynamic adjustment of the initial contour, this implementation considers that a contour analysis area can be established as a range limit for dynamic adjustment, so that during the dynamic adjustment process, any pixel in the contour will not exceed the contour analysis area, which is conducive to further ensuring the reliability of the solution of this application.

[0133] The contour analysis region is based on the initial contour. For example, by enlarging the initial contour by a certain factor, the outer contour of the contour analysis region is obtained. Similarly, by shrinking the initial contour by a certain factor, the inner contour of the contour analysis region is obtained.

[0134] Step S105: Based on the contours of the object to be analyzed in N cross-sectional images, expand to obtain the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction, as the obtained contour information of the object to be analyzed.

[0135] As described above, the scheme of this application can obtain the outline of the object to be analyzed in N cross-sectional images along the segmentation direction.

[0136] In addition, in practical applications, to improve the flexibility of the solution, users are also allowed to combine images to adjust the contour of the object to be analyzed in any one or more cross-sectional images from N cross-sectional images.

[0137] After obtaining the contours of the object to be analyzed in N cross-sectional images along the segmentation direction, it is understood that these N cross-sectional images are not all the cross-sectional images along the segmentation direction. Therefore, based on the contours of the object to be analyzed in these N cross-sectional images, it is necessary to further determine the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction. For example, for the cross-sectional images at each pixel position, the contours of the object to be analyzed can be predicted by feature analysis and similarity comparison based on the principle mentioned above.

[0138] In one specific embodiment of the present invention, step S105 may specifically include:

[0139] Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are extended by interpolation or fitting, and these contours are used as the contour information of the object to be analyzed.

[0140] This implementation takes into account that after obtaining the contours of the object to be analyzed in N cross-sectional images along the segmentation direction, the contour information of the object to be analyzed is basically determined. The positions of the remaining individual pixels along the segmentation direction can be directly calculated by interpolation or fitting to obtain the contour of the object to be analyzed, which is much simpler in calculation and has little impact on accuracy. The final contour information of the object to be analyzed is the complete contour of the object to be analyzed.

[0141] Step S106: Determine the volume of the object to be analyzed based on its contour information.

[0142] Since the contour information of the object to be analyzed is obtained, that is, the complete contour of the object to be analyzed, the volume of the object to be analyzed can be easily determined.

[0143] For example, one can iterate through the contours of the object to be analyzed in each cross-sectional image of the contour information and fill the interior of the contours, specifically by filling the pixel values ​​inside the contours with 255 to achieve pixel marking inside the contours. See also... Figure 3 This is a schematic diagram showing the pixels inside the outline of the object to be analyzed in a cross-sectional image after they have been marked. Then, by mapping the marked pixels in each cross-sectional image to the voxels in the volume data, the marked volume data mask can be obtained.

[0144] In the volume data mask, unmarked voxels are 0, and marked voxels are 255. The volume corresponding to a single voxel can be known in advance. Therefore, by counting the total number of voxels marked as 255 in the volume data mask and multiplying the total number by the volume corresponding to a single voxel, the volume of the object to be analyzed can be obtained.

[0145] Furthermore, in some implementations, after obtaining the volume data mask, the grayscale values ​​of the corresponding pixels in the volume data can be obtained for these voxels marked as 255, thereby obtaining a grayscale histogram, which can be used for projects such as blood flow analysis.

[0146] In one specific embodiment of the present invention, it may further include:

[0147] After determining the contour of the object to be analyzed in N cross-sectional images, for each of the N cross-sectional images, K connection points are set on the contour of the cross-sectional image.

[0148] For any two adjacent cross-sectional images in N cross-sectional images, the connection points of the two adjacent cross-sectional images are connected according to the connection rule that all connection points must participate in the connection, so as to establish multiple non-overlapping triangular patches between the two adjacent cross-sectional images, and to construct the patch model of the object to be analyzed.

[0149] In this implementation method, a patch model of the object to be analyzed can be obtained. For further understanding, please refer to [reference needed]. Figure 4 , Figure 4 The diagram shows the contours of the object to be analyzed in three adjacent cross-sectional images out of N cross-sectional images, labeled as contour O, contour P, and contour Q, respectively. K connection points need to be set on the contour of each cross-sectional image.

[0150] Next, the connection points of two adjacent cross-sectional images need to be connected. The specific content of the connection rules can be set and adjusted according to actual needs, but each connection point needs to participate in the connection, and the result is multiple non-overlapping triangular facets.

[0151] For example, for the outline of the object to be analyzed in each cross-sectional image, the connection point directly above the center point is recorded as connection point 1, and then numbered sequentially in counterclockwise order as connection point 2, connection point 3, and so on until connection point K.

[0152] During connection, with Figure 4 Taking contours O and P as examples, for instance, selecting connection point P... i and connection point P i+1 and connection point O i+1 This allows them to be connected to form a triangular facet. Similarly, select connection point O. i+1 and connection point O i+2 and connection point P i+1 Then they can be connected to form the next triangular facet. Select connection point P. i+1 and connection point P i+2 and connection point O i+2Then, they can be connected to form the next triangular facet, and so on. Ultimately, multiple non-overlapping triangular facets are established between contours O and P, with each connection point between them participating in the connection. Similarly, multiple non-overlapping triangular facets can also be established between contours P and Q.

[0153] After all connections are completed, the patch model of the object to be analyzed is obtained. It can be understood that, compared with the contour information of the object to be analyzed obtained in the above implementation, the patch model of the object to be analyzed is also another form of contour reflection of the object to be analyzed. In some implementations, the obtained patch model needs to be rendered and displayed.

[0154] By applying the technical solution provided in the embodiments of the present invention, after obtaining the three-dimensional volume data of the object to be analyzed, it is necessary to determine the segmentation direction of the three-dimensional volume data. Subsequently, along the segmentation direction, the three-dimensional volume data will be segmented N times by N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction.

[0155] For these N cross-sectional images, the user only needs to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images. Subsequently, the system can automatically perform feature analysis on the contours of each annotated cross-sectional image. Since the image features of the contours in different cross-sectional images are similar, the contours of the object to be analyzed in the remaining cross-sectional images can be determined based on the obtained feature analysis results. Finally, because the contours in different cross-sectional images are similar, based on the contours of the object to be analyzed in the N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction can be obtained, serving as the contour information of the object to be analyzed. Having obtained the contour information of the object to be analyzed, the volume of the object to be analyzed can be determined.

[0156] As can be seen, the solution in this application only requires the user to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images, making the operation simple. Furthermore, it automatically determines the volume of the object to be analyzed, eliminating the accuracy issues caused by inappropriate rotation angle settings found in traditional solutions. Therefore, in summary, the solution in this application can conveniently, effectively, and relatively accurately determine the volume of the object to be analyzed.

[0157] Corresponding to the above method embodiments, this invention also provides an object volume determination system, which can be referred to in conjunction with the above.

[0158] See also Figure 5 A schematic diagram of a system for determining the volume of an object, including:

[0159] The segmentation direction determination module 501 is used to determine the segmentation direction of the three-dimensional volume data after obtaining the three-dimensional volume data of the object to be analyzed.

[0160] The segmentation module 502 is used to segment the three-dimensional volume data N times along the segmentation direction using N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction.

[0161] The contour annotation module 503 is used to perform contour annotation of the object to be analyzed on at least one of the N cross-sectional images based on the received annotation instructions; where N is a positive integer greater than 1.

[0162] The contour prediction module 504 is used to perform feature analysis on the contours of each of the cross-sectional images that have been contour-annotated, and to determine the contours of the object to be analyzed in the remaining cross-sectional images based on the obtained feature analysis results.

[0163] The contour information determination module 505 is used to expand the contour of the object to be analyzed in the cross-sectional images along the segmentation direction based on the contour of the object to be analyzed in N cross-sectional images, and use the obtained contour information of the object to be analyzed.

[0164] The volume calculation module 506 is used to determine the volume of the object to be analyzed based on the contour information of the object to be analyzed.

[0165] In one specific embodiment of the present invention, the contour annotation module 503 is specifically used for:

[0166] Based on the received annotation instructions, at least 3 of the N cross-sectional images are used to annotate the contours of the object to be analyzed.

[0167] The cross-sectional image with contour annotation includes at least a first cross-sectional image, an m-th cross-sectional image, and an N-th cross-sectional image; m is a positive integer; the first cross-sectional image represents the cross-sectional image segmented at the first segmentation point in the segmentation direction; the N-th cross-sectional image represents the cross-sectional image segmented at the last segmentation point in the segmentation direction; and the m-th cross-sectional image represents the cross-sectional image segmented at the middle segmentation point in the segmentation direction.

[0168] In one specific embodiment of the present invention, the contour prediction module 504 includes:

[0169] The feature analysis unit is used to perform feature analysis on the contours of each of the cross-sectional images that have been contour-annotated, and to obtain the feature analysis results of each of the cross-sectional images that have been contour-annotated.

[0170] The contour prediction unit is used to determine the contour of the object to be analyzed in each of the cross-sectional images without contour annotation by using the feature analysis results of the two adjacent cross-sectional images with contour annotation, based on the similarity of the contour regions.

[0171] In one specific embodiment of the present invention, the contour prediction unit is specifically used for:

[0172] For each of the cross-sectional images that have not been contour-annotated, the initial contour of the object to be analyzed in the current cross-sectional image is obtained by interpolation calculation based on the contours of the two adjacent cross-sectional images that have been contour-annotated.

[0173] The reference feature analysis results of this cross-sectional image are obtained by analyzing the features of two adjacent cross-sectional images with contour annotations.

[0174] The initial contour is dynamically adjusted. When, during the adjustment process, it is determined that the feature analysis result obtained under the current contour is similar to the reference feature analysis result, the current contour is taken as the contour of the object to be analyzed in the current cross-sectional image.

[0175] In one specific embodiment of the present invention, the contour prediction unit is further configured to:

[0176] After obtaining the initial contour, a contour analysis region is established based on the initial contour;

[0177] Wherein, the outer contour of the contour analysis region encloses the initial contour, and the initial contour encloses the inner contour of the contour analysis region;

[0178] Accordingly, the initial contour is dynamically adjusted, including:

[0179] The initial contour is dynamically adjusted according to the principle that no pixel in the adjusted contour exceeds the contour analysis area.

[0180] In one specific embodiment of the present invention, the contour information determination module 505 is specifically used for:

[0181] Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded by interpolation or fitting, and are used as the contour information of the object to be analyzed.

[0182] In one specific embodiment of the present invention, a patch model generation module is further included, used for:

[0183] After determining the outline of the object to be analyzed in N cross-sectional images, K connection points are set on the outline of each of the N cross-sectional images.

[0184] For any two adjacent cross-sectional images among the N cross-sectional images, the connection points of the two adjacent cross-sectional images are connected according to the connection rule that each connection point must participate in the connection, so as to establish multiple non-overlapping triangular patches between the two adjacent cross-sectional images, thereby constructing the patch model of the object to be analyzed.

[0185] In one specific embodiment of the present invention, the segmentation direction determination module 501 is specifically used for:

[0186] After obtaining the three-dimensional volume data of the object to be analyzed, the three-dimensional volume data of the object to be analyzed is displayed on the display interface;

[0187] When a dividing line drawn by the user on the display interface is detected, the dividing direction of the three-dimensional volume data is determined based on the starting and ending positions of the dividing line.

[0188] The three-dimensional volume data of the object to be analyzed displayed on the display interface is rotated and / or translated so that the segmentation direction is parallel to the specified axis of the display interface, and the midpoint of the segmentation line coincides with the center point of the display interface.

[0189] Corresponding to the above methods and system embodiments, this invention also provides an object volume determination device and a computer-readable storage medium, which can be referred to in conjunction with the above.

[0190] See also Figure 6 The device for determining the volume of the object may include:

[0191] Memory 601 is used to store computer programs;

[0192] Processor 602 is configured to execute the computer program to implement the steps of the object volume determination method as described in any of the above embodiments.

[0193] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the object volume determination method as described in any of the above embodiments. The computer-readable storage medium referred to herein includes random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

[0194] It should also be noted that, in this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0195] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0196] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. Specific examples have been used in this application to illustrate the principles and implementation methods of the invention. The description of the above embodiments is only for the purpose of helping to understand the technical solution and core ideas of the invention. It should be noted that those skilled in the art can make several improvements and modifications to the invention without departing from the principles of the invention, and these improvements and modifications also fall within the protection scope of the invention.

Claims

1. A method for determining the volume of an object, characterized in that, include: After obtaining the three-dimensional volume data of the object to be analyzed, the segmentation direction of the three-dimensional volume data is determined; Along the segmentation direction, the three-dimensional volume data is segmented N times by N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction; Based on the received annotation instructions, at least one of the N cross-sectional images is used to annotate the contour of the object to be analyzed; where N is a positive integer greater than 1. Feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated, and the contours of the object to be analyzed in the remaining cross-sectional images are determined based on the obtained feature analysis results. Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded to obtain the contour information of the object to be analyzed. Based on the contour information of the object to be analyzed, the volume of the object to be analyzed is determined.

2. The method for determining the volume of an object according to claim 1, characterized in that, Based on the received annotation instructions, at least one of the N cross-sectional images is used to annotate the contour of the object to be analyzed, including: Based on the received annotation instructions, at least 3 of the N cross-sectional images are used to annotate the contours of the object to be analyzed. Among them, the cross-sectional images with contour annotations include at least the first cross-sectional image, the m-th cross-sectional image, and the N-th cross-sectional image; m is a positive integer; The first cross-sectional image represents the cross-sectional image segmented at the first segmentation point in the segmentation direction. The Nth cross-sectional image represents the cross-sectional image segmented at the last segmentation point in the segmentation direction; The m-th cross-sectional image represents the cross-sectional image segmented at the middle segmentation point in the segmentation direction.

3. The method for determining the volume of an object according to claim 1, characterized in that, Feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated, and the contours of the object to be analyzed in the remaining cross-sectional images are determined based on the obtained feature analysis results, including: Feature analysis is performed on the contours of each of the cross-sectional images that have been contour-annotated to obtain the feature analysis results of each of the cross-sectional images that have been contour-annotated. For each cross-sectional image without contour annotation, the contour of the object to be analyzed in the current cross-sectional image is determined based on the feature analysis results of the two adjacent cross-sectional images with contour annotation, and the similarity of the contour regions.

4. The method for determining the volume of an object according to claim 3, characterized in that, For each of the aforementioned cross-sectional images without contour annotation, the contour of the object to be analyzed in the current cross-sectional image is determined based on the feature analysis results of the two adjacent cross-sectional images with contour annotation, and on the similarity of the contour regions. This includes: For each of the cross-sectional images that have not been contour-annotated, the initial contour of the object to be analyzed in the current cross-sectional image is obtained by interpolation calculation based on the contours of the two adjacent cross-sectional images that have been contour-annotated. The reference feature analysis results of this cross-sectional image are obtained by analyzing the features of two adjacent cross-sectional images with contour annotations. The initial contour is dynamically adjusted. When, during the adjustment process, it is determined that the feature analysis result obtained under the current contour is similar to the reference feature analysis result, the current contour is taken as the contour of the object to be analyzed in the current cross-sectional image.

5. The method for determining the volume of an object according to claim 4, characterized in that, After obtaining the initial contour, the process also includes: Based on the initial contour, a contour analysis region is established; Wherein, the outer contour of the contour analysis region encloses the initial contour, and the initial contour encloses the inner contour of the contour analysis region; Accordingly, the initial contour is dynamically adjusted, including: The initial contour is dynamically adjusted according to the principle that no pixel in the adjusted contour exceeds the contour analysis area.

6. The method for determining the volume of an object according to claim 1, characterized in that, Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded, and this is used as the obtained contour information of the object to be analyzed, including: Based on the contours of the object to be analyzed in N cross-sectional images, the contours of the object to be analyzed in the cross-sectional images at each pixel position along the segmentation direction are expanded by interpolation or fitting, and are used as the contour information of the object to be analyzed.

7. The method for determining the volume of an object according to claim 1, characterized in that, Also includes: After determining the outline of the object to be analyzed in N cross-sectional images, K connection points are set on the outline of each of the N cross-sectional images. For any two adjacent cross-sectional images among the N cross-sectional images, the connection points of the two adjacent cross-sectional images are connected according to the connection rule that each connection point must participate in the connection, so as to establish multiple non-overlapping triangular patches between the two adjacent cross-sectional images, thereby constructing the patch model of the object to be analyzed.

8. The method for determining the volume of an object according to any one of claims 1 to 7, characterized in that, Determining the segmentation direction of the three-dimensional volume data includes: The three-dimensional volume data of the object to be analyzed is displayed on the display interface; When a dividing line drawn by the user on the display interface is detected, the dividing direction of the three-dimensional volume data is determined based on the starting and ending positions of the dividing line. The three-dimensional volume data of the object to be analyzed displayed on the display interface is rotated and / or translated so that the segmentation direction is parallel to the specified axis of the display interface, and the midpoint of the segmentation line coincides with the center point of the display interface.

9. A system for determining the volume of an object, characterized in that, include: The segmentation direction determination module is used to determine the segmentation direction of the three-dimensional volume data after obtaining the three-dimensional volume data of the object to be analyzed. The segmentation module is used to segment the three-dimensional volume data N times along the segmentation direction using N perpendicular planes to obtain N cross-sectional images perpendicular to the segmentation direction. The contour annotation module is used to annotate the contour of the object to be analyzed in at least one of the N cross-sectional images based on the received annotation instructions; where N is a positive integer greater than 1. The contour prediction module is used to perform feature analysis on the contours of each of the cross-sectional images that have been contour-annotated, and to determine the contours of the object to be analyzed in the remaining cross-sectional images based on the obtained feature analysis results. The contour information determination module is used to obtain the contour of the object to be analyzed in the cross-sectional image at each pixel position along the segmentation direction based on the contour of the object to be analyzed in N cross-sectional images, and use it as the obtained contour information of the object to be analyzed. The volume calculation module is used to determine the volume of the object to be analyzed based on the contour information of the object.

10. A device for determining the volume of an object, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the object volume determination method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the object volume determination method as described in any one of claims 1 to 8.

Images