Rapid curve evolution morphological medical image segmentation method

Abstract

The invention relates to the field of image segmentation, and discloses a fast curve evolution morphological medical image segmentation method, which can interactively select a segmentation region and improve the segmentation precision. The method comprises the following steps: converting a three-dimensional medical image into a multi-layer two-dimensional slice image; at least two user interaction seed points selected by a user in multiple layers are calibrated for the first time, all pixels in a target area of the user interaction seed points are added into a distance queue, the pixels are popped up one by one, whether the path distance between the pixel and each adjacent pixel is smaller than the value of the adjacent pixel in the distance transformation diagram or not is calculated and judged, and if yes, the user interaction seed points are calibrated; if not, updating the distance transformation diagram and continuously adding the adjacent pixels into the distance queue until the queue is empty; dynamically calculating user interaction seed points of each layer and carrying out secondary calibration; preprocessing the calibrated two-dimensional slice image; carrying out parallel calculation on curve evolution morphology based on curvature on the calibrated two-dimensional slice image; and converting the two-dimensional slice image into a segmented three-dimensional medical image.

Description

Technical Field

[0001] The present application relates to the field of image segmentation, and in particular to a fast curve evolution morphology medical image segmentation method. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the present application as recited in the claims. The description herein is not admitted to be prior art as disclosed by virtue of its inclusion in this section.

[0003] In the current medical image segmentation methods, there are several problems to be solved:

[0004] 1. Since the grayscale difference between the muscles and other organs around the organs or tumors to be segmented in medical images and the target area is small, the edges of the organs or tumors to be segmented are blurred. This will cause great trouble for the segmentation methods based on grayscale, region and edge;

[0005] 2. Although the model-based image segmentation method has better segmentation effect than other traditional methods, it consumes expensive computational costs when solving time-related partial differential equations and has numerical stability issues;

[0006] 3. There are certain deviations in the morphology, position and volume of organs and tumors of different individuals. At the same time, the organs to be segmented run through multiple slices of the image, and the morphological structure between different slices also varies greatly, which puts forward higher generalization requirements for medical image segmentation methods based on deep learning;

[0007] 4. Segmentation methods based on deep learning have achieved remarkable results in recent years. However, the training of neural networks requires manual annotation of a large amount of data, which is undoubtedly boring and time-consuming. At the same time, the interpretability of deep learning methods is insufficient, especially in medical applications, which still needs to be solved;

[0008] 5. Traditional 3D segmentation algorithms consume expensive time in the calculation process, and real-time performance is crucial in medical applications. Long calculation time may seriously affect the judgment and decision of radiologists during surgery.

[0009] 6. Based on the semi-automatic interactive segmentation method, the user is required to select the foreground or background area to be segmented. It is often necessary to select multiple points in the target area or to outline or select the target range. The interaction with 3D images is more complicated. These increase the user's interaction time and reduce the user's experience. Summary of the invention

[0010] The purpose of this application is to provide a fast curve evolution morphology medical image segmentation method, which can interactively select the segmentation area and improve the segmentation accuracy.

[0011] The present application discloses a fast curve evolution morphology medical image segmentation method, comprising:

[0012] Convert three-dimensional medical images into multi-layer two-dimensional slice images;

[0013] Performing an initial calibration on at least two user interaction seed points selected by the user in the multiple layers, the initial calibration comprising: adding all pixels in the target area of ​​the user interaction seed point to the distance queue, popping out the pixels one by one, calculating and determining whether the path distance between the pixel and each of its adjacent pixels is less than the value of the adjacent pixel in the distance transformation image, and if so, updating the distance transformation image and continuing to add the adjacent pixels to the distance queue until the queue is empty;

[0014] Dynamically calculate the user interaction seed points of each layer, and perform secondary calibration on the user interaction seed points;

[0015] Calculating the gradient magnitude map of the calibrated two-dimensional slice image for preprocessing;

[0016] For the calibrated two-dimensional slice image, calculating in parallel the curvature-based curve evolution morphology;

[0017] The morphological closing operation post-processing converts the two-dimensional slice image into a segmented three-dimensional medical image.

[0018] In a preferred embodiment, the initial calibration further comprises:

[0019] Calculate the grayscale weighted distance change map of the layer where the user interaction seed point is located:

[0020] D(x,y)=min{D(x,y),min[D(u,v)+w(u,v)*d(x,y,u,v)]}

[0021] Where D(x,y) represents the distance from pixel (x,y) to the target area, (u,v) represents the neighboring pixels of (x,y), w(u,v) represents the grayscale weight of pixel (u,v), i.e., the pixel value, and d(x,y,u,v) is the distance function between pixel (x,y) and pixel (u,v);

[0022] With the user interaction seed point as the center of the square diagonal, a square area with a side length of β pixels is constructed, and the mean pixel value in the square area is calculated. If it is greater than a threshold, the seed point is retained. If it is less than the threshold, the maximum pixel values ​​of the x-axis and y-axis of the user interaction seed point are calculated respectively, and the position of the largest one of the maximum pixel values ​​of the x-axis and y-axis is selected as the corrected coordinates of the user interaction seed point.

[0023] Select two adjacent user interaction seed points in order and calculate:

[0024] width=abs(seed_user[k][z-axis]-seed_user[k+1][z-axis]),

[0025] height=abs(seed_user[k][y-axis]-seed_user[k+1][y-axis]),

[0026] length=abs(seed_user[k][x-axis]-seed_user[k+1][x-axis]);

[0027] For each integer value i in the range (0, width-1), compute:

[0028] height_change=(i+1)*(height / width),

[0029] length_change=(i+1)*(length / width);

[0030] If seed_user[k][yaxis]≤seed_user[k+1][yaxis],

[0031] Then y=seed_user[k][y-axis]+height_change,

[0032] Otherwise y = seed_user[k][y axis]-height_change;

[0033] If seed_user[k][x-axis]≤seed_user[k+1][x-axis],

[0034] Then x = seed_user[k][x axis] + length_change,

[0035] Otherwise x = seed_user[k][x-axis]-length_change;

[0036] Where seed_user is the seed point array selected by user interaction, abs() represents the absolute value operation, and k represents the index of the current user interaction seed point;

[0037] Add the dynamically generated seed point (x, y, seed_user[k][z axis]+1) to the seed point array seed_final;

[0038] The secondary calibration is repeated until all the user interaction seed points are traversed.

[0039] In a preferred embodiment, the step of calculating the calibrated gradient amplitude map of the two-dimensional slice image for preprocessing further comprises:

[0040] Calculate the two-dimensional Gaussian function derivative of the two-dimensional slice image

[0041]

[0042] Where σ is set to 2 as the standard deviation of the Gaussian function, and the first-order partial derivative of the two-dimensional Gaussian function is

[0043]

[0044]

[0045] The gradient amplitude of the two-dimensional slice image is

[0046]

[0047] In a preferred embodiment, the parallel calculation of the curvature-based curve evolution morphology for the calibrated two-dimensional slice image includes:

[0048] The partial differential equation will be solved

[0049]

[0050] Approximate conversion to morphological operations:

[0051]

[0052]

[0053]

[0054] Among them, D d and E d are the discrete representations of morphological dilation and erosion, n refers to the number of iterations in the morphological implementation of the partial differential equation for calculating geodesic active contours, μ refers to the number of consecutive applications of the morphological smoothing operator, θ is the preset threshold used to limit g(I), q represents the level set function that needs to be evolved, p is the preset balloon force parameter, and α represents an arbitrary point when calculating on a two-dimensional slice image. represents the supremum-infimum morphological operator given by a discrete basis consisting of a portion of an interval of a specified length, Represents the infimum-supremum morphological operator given by the discrete basis consisting of the specified length interval. g(I) is used to detect the boundary of the image target and to solve the geodesic distance. δ is the preset value:

[0055]

[0056] A multi-thread strategy is adopted to calculate the curve evolution in the two-dimensional slice image in parallel.

[0057] In a preferred embodiment, before converting the three-dimensional medical image into a multi-layer two-dimensional slice image, the method further includes:

[0058] The window width and window level of the medical image to be segmented are adjusted, wherein the window width is 150 and the window level is 175.

[0059] In a preferred embodiment, it further comprises:

[0060] The user selects at least one user interaction seed point in the starting layer and the last layer respectively, and selects any number of user interaction seed points in the middle layer between the starting layer and the last layer, and sorts the user interaction seed points according to the z-axis coordinates;

[0061] The maximum value and the minimum value sorted along the z-axis are determined as the starting layer and the ending layer, and the three-dimensional medical images in the starting layer and the ending layer are converted into multi-layer two-dimensional slice images.

[0062] In a preferred example, 5 layers below the starting layer and 5 layers above the ending layer are removed in the initial calibration.

[0063] The present application also discloses a medical image segmentation device, comprising:

[0064] a memory for storing computer executable instructions; and,

[0065] A processor, coupled to the memory, is configured to implement the steps in the method described above when executing the computer executable instructions.

[0066] The present application also discloses a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, the steps in the method described above are implemented.

[0067] In the implementation manner of the present application, at least two user interaction seed points need to be selected, and the target area segmentation can be calculated on the two-dimensional slice image based on the initial calibration and secondary calibration of the seed points. The initial calibration is performed by adding all pixels in the target area of ​​the user interaction seed point to the distance queue, popping up the pixels one by one, calculating and judging whether the path distance between the pixel and each of its adjacent pixels is less than the value of the adjacent pixel in the distance transformation map, and if so, updating the distance transformation map and continuing to add the adjacent pixels to the distance queue until the queue is empty, so that the selection of the user interaction seed point located at the edge can be automatically calibrated based on the grayscale weighted distance change map, and the user interaction seed point after the initial calibration is then calibrated secondary in each layer of the multi-layer two-dimensional slice, so that the image segmentation can be more accurate;

[0068] Furthermore, a method is adopted to transform the problem of solving partial differential equations into evolving contours using morphological operators. Since morphological operators do not encounter the common numerical stability problems in solving partial differential equations, this method is stable and improves the speed of the algorithm.

[0069] Furthermore, in the embodiments of the present application, good segmentation effects can be obtained without using any manually annotated data sets. At the same time, multi-threading methods are used to perform parallel segmentation on two-dimensional slices, and the segmentation time is greatly reduced in combination with a dynamic round iteration strategy. The segmentation time is at the same order of magnitude as that of the deep learning method. At the same time, user interaction can make up for the shortcomings of poor interpretability of deep learning, and prevent errors in the learning model used in deep learning from causing errors in the output results. Users can interactively select the segmentation area, further improving the accuracy and reliability of medical image segmentation.

[0070] Furthermore, a method based on curve evolution is adopted. The process of image segmentation is the process of solving the energy functional, where the energy function includes internal energy terms, namely the bending force and rigidity force of the evolution curve, and external energy terms, which drive the evolution curve to move toward the target boundary. Therefore, it can avoid the segmentation errors caused by image segmentation methods based entirely on grayscale.

[0071] Each technical feature disclosed in the above invention content, each technical feature disclosed in each implementation mode and example below, and each technical feature disclosed in the accompanying drawings can be freely combined with each other to form various new technical solutions (these technical solutions should be deemed to have been recorded in this specification), unless such combination of technical features is technically infeasible. For example, in one example, feature A+B+C is disclosed, and in another example, feature A+B+D+E is disclosed, and features C and D are equivalent technical means that play the same role. Technically, only one of them can be used, and it is impossible to use them at the same time. Feature E can be combined with feature C technically. Then, the solution of A+B+C+D should not be deemed to have been recorded because it is technically infeasible, while the solution of A+B+C+E should be deemed to have been recorded. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a flowchart of a fast curve evolution morphology medical image segmentation method according to one embodiment of the present application;

[0073] Figure 2 It is a flowchart of a fast curve evolution morphology medical image segmentation method according to one embodiment of the present application;

[0074] Figure 3 It is a flowchart of a fast curve evolution morphology medical image segmentation method according to one embodiment of the present application;

[0075] Figure 4 It is a flowchart of a fast curve evolution morphology medical image segmentation method according to one embodiment of the present application;

[0076] Figure 5 It is a flowchart of a fast curve evolution morphology medical image segmentation method according to one embodiment of the present application;

[0077] Fig. 6A This is a user interactive seed point calibration effect diagram of a fast curve evolution morphology medical image segmentation method according to an embodiment of the present application;

[0078] Figure 6B This is a user interactive seed point calibration effect diagram of a fast curve evolution morphology medical image segmentation method according to an embodiment of the present application;

[0079] Fig. 7A is a schematic diagram of segmentation results in a CT image of a fast curve evolution morphology medical image segmentation method according to an embodiment of the present application;

[0080] Figure 7BIt is a schematic diagram of the segmentation results in MR images of the fast curve evolution morphology medical image segmentation method according to one embodiment of the present application. DETAILED DESCRIPTION

[0081] In the following description, many technical details are provided to help readers better understand the present application. However, those skilled in the art can understand that the technical solution claimed in the present application can be implemented even without these technical details and various changes and modifications based on the following embodiments.

[0082] Description of some concepts:

[0083] Curve evolution: describes the process of deformation of the curve under the time variable, that is, the process of observing the change of the curve over time under the influence of geometric parameters such as unit normal vector and curvature.

[0084] Morphology: Use structural elements with certain shapes to measure and extract corresponding shapes in images to achieve the purpose of image analysis and recognition. Its application can simplify or remove irrelevant structures while maintaining the basic shape structure of the image. The basic operations of morphology are: dilation, erosion, opening, and closing.

[0085] Image segmentation: It is the process of dividing an image into several regions with similar properties. From a mathematical point of view, image segmentation is the process of dividing an image into non-intersecting regions. There are mainly threshold-based, region-based, edge-based, clustering-based, model-based and deep learning-based image segmentation methods.

[0086] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below in conjunction with the accompanying drawings.

[0087] The first embodiment of the present application relates to a fast curve evolution morphology medical image segmentation method, the process of which is as follows: Figure 1 As shown, the method comprises the following steps:

[0088] In step S101 , a three-dimensional medical image is converted into multi-layer two-dimensional slice images.

[0089] In step S102, at least two user interaction seed points selected by the user in multiple layers are initially calibrated.

[0090] Optionally, in one embodiment, the initial calibration steps are as follows:

[0091] In step S1021, all pixels in the target area of ​​the user interaction seed point are added to the distance queue.

[0092] In step S1022, pixels are popped out one by one.

[0093] In step S1023, it is calculated and determined whether the path distance between the pixel and each of its adjacent pixels is less than the value of the adjacent pixel in the distance transformation map.

[0094] In step S1024, if it is less than, the distance transformation map is updated and the adjacent pixels continue to be added to the distance queue until the queue is empty.

[0095] In step S103, the user interaction seed points of each layer are dynamically calculated, and the user interaction seed points are recalibrated.

[0096] In step S104, the gradient magnitude map of the calibrated two-dimensional slice image is calculated for preprocessing.

[0097] In step S105 , the curvature-based curve evolution morphology is calculated in parallel for the calibrated two-dimensional slice image.

[0098] In step S106, a morphological closing operation is performed to convert the two-dimensional slice image into a segmented three-dimensional medical image.

[0099] In an optional embodiment, the initial calibration may further include:

[0100] Calculate the grayscale weighted distance change map of the layer where the user interaction seed point is located:

[0101] D(x,y)=min{D(x,y),min[D(u,v)+w(u,v)*d(x,y,u,v)]}

[0102] Among them, D(x,y) represents the distance from pixel (x,y) to the target area, (u,v) represents the neighboring pixels of (x,y), w(u,v) represents the grayscale weight of pixel (u,v), and d(x,y,u,v) is the distance function between pixel (x,y) and pixel (u,v).

[0103] With the user interaction seed point as the center of the square diagonal, a square area with a side length of β pixels (optionally, the side length is 10-60 pixels, preferably 20 pixels) is constructed, and the mean pixel value in the square area is calculated. If it is greater than the threshold, the seed point is retained. If it is less than the threshold, the maximum pixel values ​​of the x-axis and y-axis of the user interaction seed point are calculated respectively, and the position of the largest one of the maximum pixel values ​​of the x-axis and y-axis is selected as the corrected user interaction seed point coordinates. Optionally, as Fig. 6A and Figure 6B As shown, Fig. 6A and Figure 6BThis is a calibration diagram of user interaction seed points in different layers. After the user selects a user interaction seed point in the two-dimensional slice image, the user interaction seed point constructs a 20-pixel square area with itself as the center, and continuously calculates and calibrates the position within the square area, and finally automatically moves to the target position along the x-axis or y-axis.

[0104] In an optional embodiment, if Figure 2 As shown, the secondary calibration may include:

[0105] In step S1031, two adjacent user interaction seed points are selected in order and the following is calculated:

[0106] width=abs(seed_user[k][z-axis]-seed_user[k+1][z-axis]),

[0107] height=abs(seed_user[k][y-axis]-seed_user[k+1][y-axis]),

[0108] length=abs(seed_user[k][x-axis]-seed_user[k+1][x-axis]);

[0109] In step S1032, for each integer value i in the range (0, width-1), calculate:

[0110] height_change=(i+1)*(height / width),

[0111] length_change=(i+1)*(length / width);

[0112] In step S1033, it is determined whether seed_user[k][y axis] is less than or equal to seed_user[k+1][y axis].

[0113] If it is less than, then y = seed_user[k][y axis] + height_change,

[0114] Otherwise y = seed_user[k][y axis]-height_change;

[0115] In step S1034, it is determined whether seed_user[k][x axis] is less than or equal to seed_user[k+1][x axis].

[0116] If it is less than, x=seed_user[k][x-axis]+length_change,

[0117] Otherwise x = seed_user[k][x-axis]-length_change;

[0118] Where seed_user is the seed point array selected by user interaction, abs() represents the absolute value operation, and k represents the index of the current user interaction seed point;

[0119] In step S1035, the dynamically generated seed point (x, y, seed_user[k][z axis]+1) is added to the seed point array seed_final;

[0120] In step S1036 , the calibration is repeated twice until all user interaction seed points are traversed.

[0121] In an optional embodiment, if Figure 3 As shown, calculating the gradient magnitude map of the calibrated two-dimensional slice image for preprocessing may further include:

[0122] In step S1041, a two-dimensional Gaussian function derivative of the two-dimensional slice image is calculated and a two-dimensional Gaussian function derivative is calculated;

[0123] The two-dimensional Gaussian function of the two-dimensional slice image is:

[0124]

[0125] Where σ is set to the standard deviation of the two-dimensional Gaussian function, and the first-order partial derivative of the two-dimensional Gaussian function is:

[0126]

[0127]

[0128] In step S1042, the gradient amplitude of the two-dimensional slice image is calculated:

[0129]

[0130] In an optional embodiment, if Figure 4 As shown, for the calibrated two-dimensional slice image, parallel calculation of curvature-based curve evolution morphology can further include:

[0131] In S1051, the partial differential equation

[0132]

[0133] Approximate conversion to morphological operations:

[0134]

[0135]

[0136]

[0137] Among them, D d and E d are the discrete representations of morphological dilation and erosion, n refers to the number of iterations in the morphological implementation of the partial differential equation for calculating geodesic active contours, μ refers to the number of consecutive applications of the morphological smoothing operator, θ is the preset threshold used to limit g(I), q represents the level set function that needs to be evolved, p is the preset balloon force parameter, and α represents an arbitrary point when calculating on a two-dimensional slice image. represents the supremum-infimum morphological operator given by a discrete basis consisting of a portion of an interval of a specified length, Represents the infimum-supremum morphological operator given by the discrete basis consisting of the specified length interval. g(I) is used to detect the boundary of the image target and to solve the geodesic distance. δ is the preset value:

[0138]

[0139] In S1052, a multi-thread strategy is adopted to calculate the curve evolution in the two-dimensional slice image in parallel.

[0140] In an optional embodiment, before converting the three-dimensional medical image into a multi-layer two-dimensional slice image, the process further includes:

[0141] Adjust the window width and window level of the medical image to be segmented, with the window width being 150 and the window level being 175.

[0142] In an optional embodiment, if Figure 5 As shown, it may further include:

[0143] In step S1011, the user selects at least one user interaction seed point in the starting layer and the last layer respectively, and selects any number of user interaction seed points in the middle layer between the starting layer and the last layer (optionally, 0 user interaction seed points may also be selected in the middle layer), and sorts the user interaction seed points according to the z-axis coordinates;

[0144] In step S1012, the maximum and minimum values ​​sorted along the z-axis are determined as the starting layer and the ending layer, and the three-dimensional medical images in the starting layer and the ending layer are converted into multi-layer two-dimensional slice images.

[0145] In an optional embodiment, since the anatomical structures at the beginning and the end are relatively small, the five layers below the starting layer and the five layers above the end layer may be removed from the automatic calibration during the initial calibration.

[0146] In an optional embodiment, when processing the situation where the user interaction seed point is close to the edge of the target area, a grayscale weighted Euclidean distance transformation map is used as a reference. Optionally, other methods may also be used as a reference, such as calculating image gradients, using edge operator filtering to obtain an edge probability map, etc.

[0147] In an optional embodiment, when the morphological curve is evolved in two-dimensional space, the number of iteration rounds is dynamically changed by linear change. Optionally, a nonlinear change strategy may also be adopted.

[0148] The advantages of the embodiments of the present application are briefly described below:

[0149] like Fig. 7A and 7B As shown, Fig. 7A The image segmentation result of the CT image after adopting the implementation mode of the present application is shown in FIG. Figure 7B The image segmentation result of the MR image after adopting the implementation method of the present application can be segmented to obtain a very clear edge image of each layer of the target organ (in this embodiment, the target organ is the liver, and the target organ may also be other internal organs or a tumor), thereby converting the 2D image into a very accurate 3D model, which greatly reduces the workload of doctors and can also accurately quantify the volume, shape, and position of the target organ or tumor, assisting doctors in specifying surgical plans.

[0150] A second embodiment of the present application relates to a medical image segmentation device, the medical image segmentation device comprising:

[0151] a memory for storing computer executable instructions; and,

[0152] A processor, coupled to the memory, is configured to implement any steps in the method described above when executing the computer executable instructions.

[0153] The first implementation manner is a method implementation manner corresponding to the present implementation manner. The technical details in the first implementation manner can be applied to the present implementation manner, and the technical details in the present implementation manner can also be applied to the first implementation manner.

[0154] It should be noted that those skilled in the art should understand that the implementation functions of each module shown in the above-mentioned medical image segmentation device can be understood by referring to the relevant description of the above-mentioned fast curve evolution morphology medical image segmentation method. The functions of each module shown in the above-mentioned medical image segmentation device can be implemented by a program (executable instruction) running on a processor, or by a specific logic circuit. If the above-mentioned medical image segmentation device of the embodiment of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read Only Memory), disk or optical disk and other media that can store program codes. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0155] Accordingly, the embodiments of the present application also provide a computer-readable storage medium, in which computer executable instructions are stored, and when the computer executable instructions are executed by the processor, the various method embodiments of the present application are implemented. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be a computer-readable instruction, a data structure, a module of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable storage media does not include temporary computer-readable media (transitory media), such as modulated data signals and carriers.

[0156] It should be noted that, in the present application, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "includes one" does not exclude the presence of other identical elements in the process, method, article or device including the element. In the present application, if it is mentioned that an action is performed according to an element, it means at least the meaning of performing the action according to the element, which includes two situations: performing the action only according to the element, and performing the action according to the element and other elements. Expressions such as multiple, multiple, and multiple include 2, 2 times, 2 kinds, and more than 2, more than 2 times, and more than 2 kinds.

[0157] The serial numbers used in describing the steps of the method do not themselves constitute any limitation on the order of these steps. For example, the step with a larger serial number does not necessarily have to be executed after the step with a smaller serial number. The step with a larger serial number may be executed first and then the step with a smaller serial number. They may also be executed in parallel, as long as this execution order is reasonable for those skilled in the art. For another example, multiple steps with consecutive serial numbers (e.g., step 101, step 102, step 103, etc.) do not limit other steps that can be executed in between. For example, there may be other steps between step 101 and step 102.

[0158] This specification includes combinations of the various embodiments described herein. Individual references to embodiments (e.g., "one embodiment" or "some embodiments" or "preferred embodiments"); however, these embodiments are not mutually exclusive unless indicated as mutually exclusive or it is clear to a person skilled in the art that they are mutually exclusive. It should be noted that the word "or" is used in this specification in a non-exclusive sense unless the context clearly indicates or requires otherwise.

[0159] In addition, it should be understood that the above description is only a preferred embodiment of this specification and is not intended to limit the scope of protection of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification should be included in the scope of protection of one or more embodiments of this specification.

[0160] In some cases, the actions or steps described in the claims may be performed in a different order than in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A fast curve evolution morphology medical image segmentation method, characterized in that: include: Convert three-dimensional medical images into multi-layer two-dimensional slice images; Performing an initial calibration on at least two user interaction seed points selected by the user in the multiple layers, the initial calibration comprising: adding all pixels in the target area of ​​the user interaction seed point to the distance queue, popping out the pixels one by one, calculating and determining whether the path distance between the pixel and each of its adjacent pixels is less than the value of the adjacent pixel in the distance transformation map, and if so, updating the distance transformation map and continuing to add the adjacent pixels to the distance queue until the queue is empty; Dynamically calculate the user interaction seed points of each layer, and perform secondary calibration on the user interaction seed points; Calculating the gradient magnitude map of the calibrated two-dimensional slice image for preprocessing; For the calibrated two-dimensional slice image, calculating in parallel the curvature-based curve evolution morphology; The morphological closing operation post-processing converts the two-dimensional slice image into a segmented three-dimensional medical image.

2. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: The initial calibration further comprises: Calculate the grayscale weighted distance change map of the layer where the user interaction seed point is located: D(x,y)=min{D(x,y),min[D(u,v)+w(u,v)*d(x,y,u,v)]} Where D(x,y) represents the distance from pixel (x,y) to the target area, (u,v) represents the neighboring pixels of (x,y), w(u,v) represents the grayscale weight of pixel (u,v), and d(x,y,u,v) is the distance function between the pixel (x,y) and the pixel (u,v); With the user interaction seed point as the center of the square diagonal, a square area with a side length of β pixels is constructed, and the mean pixel value in the square area is calculated. If it is greater than a threshold, the seed point is retained. If it is less than the threshold, the maximum pixel values ​​of the x-axis and y-axis of the user interaction seed point are calculated respectively, and the position of the largest one of the maximum pixel values ​​of the x-axis and y-axis is selected as the corrected coordinates of the user interaction seed point.

3. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: The secondary calibration includes: Select two adjacent user interaction seed points in order and calculate: width=abs(seed_user[k][z-axis]-seed_user[k+1][z-axis]), height=abs(seed_user[k][y-axis]-seed_user[k+1][y-axis]), length=abs(seed_user[k][x-axis]-seed_user[k+1][x-axis]); For each integer value i in the range (0, width-1), compute: height_change=(i+1)*(height / width), length_change=(i+1)*(length / width); If seed_user[k][yaxis]≤seed_user[k+1][yaxis], Then y=seed_user[k][y-axis]+height_change, Otherwise y = seed_user[k][y axis]-height_change; If seed_user[k][x-axis]≤seed_user[k+1][x-axis], Then x = seed_user[k][x axis] + length_change, Otherwise x = seed_user[k][x-axis]-length_change; Where seed_user is the seed point array selected by user interaction, abs() represents the absolute value operation, and k represents the index of the current user interaction seed point; Add the dynamically generated seed point (x, y, seed_user[k][z axis]+1) to the seed point array seed_final; The secondary calibration is repeated until all the user interaction seed points are traversed.

4. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: The calculating of the calibrated gradient magnitude map of the two-dimensional slice image for preprocessing further comprises: Calculate the two-dimensional Gaussian function derivative of the two-dimensional slice image Where σ is set to the standard deviation of the two-dimensional Gaussian function, and the first-order partial derivative of the two-dimensional Gaussian function is The gradient amplitude of the two-dimensional slice image is 5. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: The step of calculating the curvature-based curve evolution morphology in parallel for the calibrated two-dimensional slice image comprises: The partial differential equation will be solved Approximate conversion to morphological operations: Among them, D d and E d are the discrete representations of morphological dilation and erosion, n refers to the number of iterations in the morphological implementation of the partial differential equation for calculating geodesic active contours, μ refers to the number of consecutive applications of the morphological smoothing operator, θ is the preset threshold used to limit g(I), q represents the level set function that needs to be evolved, p is the preset balloon force parameter, and α represents an arbitrary point when calculating on a two-dimensional slice image. represents the supremum-infimum morphological operator given by a discrete basis consisting of a portion of an interval of a specified length, Represents the infimum-supremum morphological operator given by the discrete basis consisting of the specified length interval. g(I) is used to detect the boundary of the image target and to solve the geodesic distance. δ is the preset value: A multi-thread strategy is adopted to calculate the curve evolution in the two-dimensional slice image in parallel.

6. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: The step of converting the three-dimensional medical image into a multi-layer two-dimensional slice image also includes: The window width and window level of the medical image to be segmented are adjusted, wherein the window width is 150 and the window level is 175.

7. The fast curve evolution morphology medical image segmentation method according to claim 1, characterized in that: Further including: The user selects at least one user interaction seed point in the starting layer and the last layer respectively, and selects any number of user interaction seed points in the middle layer between the starting layer and the last layer, and sorts the user interaction seed points according to the z-axis coordinates; The maximum value and the minimum value sorted along the z-axis are determined as the starting layer and the ending layer, and the three-dimensional medical images in the starting layer and the ending layer are converted into multi-layer two-dimensional slice images.

8. The fast curve evolution morphology medical image segmentation method according to claim 7, characterized in that: In the initial calibration, 5 layers below the starting layer and 5 layers above the ending layer are removed.

9. A medical image segmentation device, characterized in that: include: A memory for storing computer executable instructions; as well as, A processor, coupled to the memory, configured to implement the steps of the method according to any one of claims 1 to 8 when executing the computer executable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, the steps in the method according to any one of claims 1 to 8 are implemented.

Images