Brain blood vessel segmentation method, device and storage medium based on black blood vessel wall curved surface reconstruction

Abstract

The present invention provides a cerebral blood vessel segmentation method, device and storage medium based on black blood vessel wall curved surface reconstruction, comprising: obtaining discrete points of the black blood vessel wall contour in an image, and obtaining a black blood vessel wall reconstructed curved surface based on the discrete points; obtaining the center line of the black blood vessel wall reconstructed curved surface, and performing linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively; establishing three imaging planes corresponding to the X-axis, Y-axis and Z-axis, and the three imaging planes respectively map the sampling data in the three vector directions of interest on the center line; constructing a segmentation subnet and a diagnosis subnet, and processing the sampling data based on the segmentation subnet and the diagnosis subnet, segmenting the lumen area and the outer wall area, and obtaining the morphological information and blood vessel morphology of the cerebral blood vessel.

Description

Technical Field

[0001] The present invention relates to blood vessel segmentation and artificial intelligence technology, and in particular to a cerebral blood vessel segmentation method, device and storage medium based on black blood vessel wall curved surface reconstruction. Background Art

[0002] In the field of medical imaging, there are many vascular imaging technologies: for example, the traditional common X-ray angiography technology, namely digital subtraction angiography (DSA), CT angiography technology (Computed Tomography Angiogram, CTA) and bright blood magnetic resonance angiography technology. The image information obtained by relying on traditional technology is generally the information of blood flow in the blood vessel cavity, rather than the information of the blood vessel wall. In order to compensate for this, in recent years, 3D black blood technology has been pointed out to be able to perform high-resolution imaging of the blood vessel wall.

[0003] In the planning of vascular surgeries, such as surgical resection or deployment of percutaneous stent implants, it depends heavily on the geometric characteristics of the vessels. Therefore, segmenting the vessels to assess their geometric characteristics plays an important role in medical treatment. Due to the tortuosity of the vessels, a complete image of the entire vessel wall cannot be displayed on a single imaging level of magnetic resonance imaging (MRI). In most cases, the visibility of blood vessels in traditional 3D MRI images is poor. Due to the common complex geometry of blood vessels and the importance of the old technology that cannot align the reading direction with the blood flow direction to achieve blood suppression, it is impossible to have a direction that can well depict the three planes of the vessel in a single acquisition.

[0004] At the same time, the traditional manual operation method of post-processing software is cumbersome, which makes the post-processing work of doctors operating the software to reconstruct images time-consuming. In addition, there are some disadvantages in using a manual segmentation method based on 2D slices to process 3D data sets. For example, the 2D slice-based method does not force contour continuity in the longitudinal dimension. Secondly, it does not allow efficient manual contour editing. In order to address these shortcomings, manual demarcation and segmentation of the aneurysm cavity and outer wall layer by layer from high-resolution MRI black blood images is very time-consuming and tedious, and it takes a skilled radiologist more than 30 minutes to make an effective distinction, which is very prone to fatigue errors. Therefore, there is an urgent need for a technical solution that can efficiently assist the aortic cavity and perform outer wall segmentation. Summary of the invention

[0005] The embodiments of the present invention provide a cerebral blood vessel segmentation method, device and storage medium based on black blood vessel wall curved surface reconstruction, which can efficiently assist the aortic cavity and perform outer wall segmentation.

[0006] A first aspect of an embodiment of the present invention provides a cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction, comprising:

[0007] Acquire discrete points of the black blood vessel wall contour in the image, and obtain a reconstructed curved surface of the black blood vessel wall based on the discrete points;

[0008] Acquire the center line of the black blood vessel wall reconstructed surface, and perform linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively;

[0009] Establishing three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line;

[0010] A segmentation subnet and a diagnosis subnet are constructed, and the sampled data are processed based on the segmentation subnet and the diagnosis subnet to segment the lumen area and the outer wall area to obtain the morphological information and vascular morphology of the cerebral blood vessels.

[0011] Optionally, in a possible implementation manner of the first aspect, acquiring discrete points of a black blood vessel wall contour in an image, and obtaining a black blood vessel wall reconstructed surface based on the discrete points includes:

[0012] After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve;

[0013] Projecting the contour line along the direction of interest to form a scanned surface;

[0014] The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

[0015] Optionally, in a possible implementation of the first aspect, the method further includes:

[0016] The centerline data is sampled, and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data;

[0017] Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data;

[0018] Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

[0019] Optionally, in a possible implementation manner of the first aspect, performing linear sampling on an X-axis, a Y-axis, and a Z-axis of the center line respectively includes:

[0020] At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously;

[0021] Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

[0022] Optionally, in a possible implementation manner of the first aspect, establishing three imaging planes corresponding to the X-axis, the Y-axis, and the Z-axis, wherein the three imaging planes respectively map sampling data in three vector directions of interest on the center line, comprises:

[0023] Three imaging planes Csox, Csoy, and Csoz are established to map the sampling data in the three vector directions of interest on the center line respectively, that is, the imaging plane Csox maps the volume data in the Lx vector direction, the imaging plane Csoy maps the volume data in the Ly vector direction, and the imaging plane Csoz maps the volume data in the Lz vector direction.

[0024] A second aspect of an embodiment of the present invention provides a cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction, comprising:

[0025] A reconstruction module, used to obtain discrete points of the black blood vessel wall contour in the image, and obtain a reconstructed surface of the black blood vessel wall based on the discrete points;

[0026] A sampling module, used for acquiring the center line of the black blood vessel wall reconstructed surface, and performing linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively;

[0027] A mapping module, used to establish three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line;

[0028] The processing module is used to construct a segmentation subnet and a diagnosis subnet, process the sampled data based on the segmentation subnet and the diagnosis subnet, segment the lumen area and the outer wall area, and obtain the morphological information and vascular morphology of the cerebral blood vessels.

[0029] Optionally, in a possible implementation manner of the second aspect, acquiring discrete points of a black blood vessel wall contour in an image, and obtaining a black blood vessel wall reconstructed surface based on the discrete points includes:

[0030] After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve;

[0031] Projecting the contour line along the direction of interest to form a scanned surface;

[0032] The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

[0033] Optionally, in a possible implementation manner of the second aspect, the method further includes:

[0034] The centerline data is sampled, and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data;

[0035] Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data;

[0036] Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

[0037] Optionally, in a possible implementation manner of the second aspect, performing linear sampling on an X-axis, a Y-axis, and a Z-axis of the center line respectively includes:

[0038] At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously;

[0039] Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

[0040] According to a third aspect of an embodiment of the present invention, a readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the method of the first aspect of the present invention and various possible designs of the first aspect.

[0041] The present invention provides a cerebral vascular segmentation method, device and storage medium based on black vessel wall curved surface reconstruction, which can solve the problem that conventional imaging outputs are all cross-sections and cannot accurately diagnose lesions. It can reconstruct patient-specific three-dimensional (3D) geometric shapes, which is helpful for measurement at various stages of treatment. In particular, these geometric shapes can facilitate the assessment of systemic vascular disease risk.

[0042] The present invention uses black blood technology to reconstruct blood vessel surfaces and proposes an improved blood vessel surface reconstruction algorithm. Based on a hybrid interpolation algorithm combining contour deformation and corresponding point matching, the contour shape and grayscale information of the target can be well maintained while performing interlayer interpolation of image sequences. This method not only shows the characteristics of tissue details, but also outlines the contours of human organs more clearly and meticulously. The reconstruction effect of the volume data processed by this method will be more vivid, and the positioning of tissue details will be more accurate. This technology develops a vessel wall imaging surface reconstruction algorithm. After the black blood high-resolution scan, the film is post-processed for vessel wall imaging surface reconstruction, so that the overall blood vessel situation can be displayed in a multi-directional, intuitive and clear manner, so that doctors have a qualitative and quantitative understanding of the thickness, shape and spatial position of the blood vessels. In the study of black blood vessel wall surface reconstruction, it solves the problem that traditional visualization technology cannot track and observe the vessel wall, and has comprehensive advantages in displaying the stenosis and stenosis position of the vascular lumen and judging the degree of arterial stenosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of a first implementation of a cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction;

[0044] Figure 2 A flowchart of a first implementation method of surface reconstruction;

[0045] Figure 3 A flowchart of a second implementation method of surface reconstruction;

[0046] Figure 4 A schematic diagram showing whether the profile can be edited;

[0047] Figure 5 The flowchart of the first embodiment of the cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in sequences other than those illustrated or described herein.

[0050] It should be understood that in various embodiments of the present invention, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0051] It should be understood that in the present invention, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0052] It should be understood that in the present invention, "plurality" refers to two or more than two. "And / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "Contains A, B and C", "Contains A, B, C" means that A, B, and C are all included, "Contains A, B or C" means that one of A, B, and C is included, and "Contains A, B and / or C" means that any one, any two, or any three of A, B, and C are included.

[0053] It should be understood that in the present invention, "B corresponding to A", "B corresponding to A", "A corresponds to B" or "B corresponds to A" means that B is associated with A and B can be determined based on A. Determining B based on A does not mean determining B based only on A, but B can also be determined based on A and / or other information. A and B match when the similarity between A and B is greater than or equal to a preset threshold.

[0054] Depending on the context, "if" as used herein may be interpreted as "when" or "when" or "in response to determining" or "in response to detecting."

[0055] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0056] The present invention provides a cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction, such as Figure 1 As shown, including:

[0057] Step S10: acquiring discrete points of the black blood vessel wall contour in the image, and obtaining a reconstructed black blood vessel wall surface based on the discrete points.

[0058] In step S10, it also includes:

[0059] After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve;

[0060] Projecting the contour line along the direction of interest to form a scanned surface;

[0061] The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

[0062] Multi-plane reconstruction is used to obtain discrete points of the black blood vessel wall contour; then, the discrete points are interpolated to obtain a smooth contour fitting curve; secondly, the contour line is projected along the direction of interest to form a scanning surface; finally, the vascular body data corresponding to the scanning surface is displayed to obtain the black blood vessel wall reconstructed surface. The basic process of surface reconstruction is as follows: Figure 2 shown.

[0063] Step S20, obtaining the center line of the black blood vessel wall reconstructed surface, and performing linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively.

[0064] In step S20, it also includes:

[0065] The centerline data is sampled, and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data;

[0066] Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data;

[0067] Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

[0068] First, the centerline data was sampled, and the coordinate values ​​in millimeters were converted into dimensional coordinate values ​​so that they were located in the original black vessel wall data. At the same time, the converted coordinate values ​​were differentially calculated to control the sampling step size of the centerline; then, the vector of interest was established at the sampled centerline data, and the surface grayscale value of the original black vessel wall data was sampled. The position of the sampling point was continuously updated during the sampling process; finally, the sampled data was surface reorganized and the reorganized data was grayscale transformed. Black blood MRI of the aneurysm wall and elliptical center contrast enhanced magnetic resonance angiography (CE-MRA) of the lumen were acquired continuously to minimize the influence of patient motion between the two acquisitions. The 3D sequence used slice-selective 90° excitation pulses and non-selective refocusing pulses. The refocusing flip angle was modulated to produce a constant transverse magnetization intensity on the echo train for a specific combination of 880ms and 40ms. The algorithm used eight start echoes, during which no data was collected, to avoid rapid changes in transverse magnetization at the beginning of the echo sequence and to remove the signal from the flowing blood.

[0069] Step S30: establishing three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line.

[0070] In step S30, it also includes:

[0071] The linear sampling on the X-axis, Y-axis and Z-axis of the center line includes:

[0072] At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously;

[0073] Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

[0074] Three imaging planes Csox, Csoy, and Csoz are established to map the sampling data in the three vector directions of interest on the center line respectively, that is, the imaging plane Csox maps the volume data in the Lx vector direction, the imaging plane Csoy maps the volume data in the Ly vector direction, and the imaging plane Csoz maps the volume data in the Lz vector direction.

[0075] In view of the problem that traditional surface reconstruction has a single vector of interest and cannot effectively maintain the shape of blood vessels, this algorithm can simultaneously observe the improved black blood vessel wall surface reconstruction in three specific vectors of interest, such as Figure 3 shown.

[0076] At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set at the same time, so that they are linearly sampled along the X, Y, and Z axis directions respectively, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions, that is, the integrity of the data sampling in the direction of the vector of interest at the center point is ensured, so that the tissue information around the center point is fully preserved, and the corresponding coordinate value of the center point is accurately recorded. After completing the above sampling steps, three imaging planes Csox, Csoy, and Csoz are established to map the sampling data in the three directions of the vector of interest on the center line, that is, the imaging plane Csox maps the volume data in the direction of the Lx vector, the imaging plane Csoy maps the volume data in the direction of the Ly vector, and the imaging plane Csoz maps the volume data in the direction of the Lz vector, and the stretching method of the stretched surface reconstruction is maintained during mapping. For the three imaging planes Csox, Csoy, and Csoz, for the convenience of description, the length corresponds to the center line length and the width corresponds to the sampling width of each vector of interest.

[0077] φ(x, y)>0,(x, y)∈I in

[0078] φ(z,y)<0,(x,y)∈I out

[0079] φ(x, y)=0,(x, y)∈C

[0080] According to the above formula, the algorithm will be defined as a spline with minimal energy guided by external constraints that pull it towards features such as edges. At the same time, it also defines an internal energy term to impose a smoothness constraint on the moving curve. The formula is based on the design of the velocity term, which allows the evolving front to gradually reach zero velocity as it approaches the object boundary and gradually stops. The velocity term may depend on the boundary of the front, and it can also use information inside the area surrounded by the evolving front to satisfy the zero-level set of the Lipschitz continuous function φ to achieve the subdivision of the global assembly cache.

[0081] Step S40: construct a segmentation subnet and a diagnosis subnet, process the sampled data based on the segmentation subnet and the diagnosis subnet, segment the lumen area and the outer wall area, and obtain the morphological information and vascular morphology of the cerebral blood vessels.

[0082] The present invention consists of a segmentation subnet and a diagnosis subnet for performing segmentation and diagnosis tasks on MRI images, wherein manually marked lumen area, manually marked outer wall area and manually marked lesion type are used as factual basis based on modified clinical criteria. Specifically, a deep U-shaped convolutional neural network (CNN) with a weighted fusion layer is designed as a segmentation subnet, which can simultaneously segment the lumen area and the outer wall area under triple supervision to provide a vascular wall map as morphological information. Then, images, image streams and morphological streams of the obtained vascular wall map are extracted from two deep convolutional neural networks (CNNs) and combined to obtain the vascular morphology in the diagnosis subnet.

[0083] Based on this formula, the present invention constrains the continuity of the boundary by checking whether the ratio between the cavity areas on adjacent slices is between 0.5 and 1.5. If the ratio is outside this range, the algorithm will pause to allow the user to edit the contour on the current image slice, and the user can determine whether to edit the contour before continuing the propagation, such as Figure 4 As shown. The orientation of the field of view determines the effectiveness of blood suppression, as the gradient is only executed in the readout direction during the start-up echo. As a result, for the coronal orientation read out in the rostral direction, the blood signal is well suppressed in the descending aorta, while residual signal is present in the aortic arch. Similarly, for the transverse orientation read out in the anteroposterior direction, the blood signal in the aortic arch is better suppressed, while a significant blood signal is visible in the descending aorta (arrow). The scan plan for the 3D images was performed using an excitation volume narrower than the encoding volume in the slice direction to spatially encode potential artifacts generated by the non-selective refocusing pulse. The resulting 3D image, displayed in the slice encoding direction, shows excellent suppression of the outer volume signal. Planar images through the inner and outer volumes show the effectiveness of outer volume suppression. The signal in the outer volume image is magnified 50 times to highlight the signal suppression in these vials, which range from 250 milliseconds to 1500 milliseconds in the first phase.

[0084] In extreme cases where blood signals are not properly suppressed or arteries are severely narrowed, the algorithm performs border editing. Since manual editing introduces observer variability and takes time, the automatic editing mechanism used by the algorithm should minimize user interaction while also providing flexibility. Its editing mechanism consists of two main components: (1) Curved Plannar Reformation (CPR) views and (2) construction of smooth contours in CPR views.

[0085] In the present algorithmic 3D segmentation method, the first vessel contour can be propagated to its neighboring image slices to be used as the initial contour, which is then refined using the deformable contour model described previously. Based on this technique, three innovations are made to the existing 2D slice-based segmentation method: (1) When segmenting the lumen, a registration module is used to convert the previous slice into the final contour before using it as the final contour. Before the next slice, the initial contour is improved. (2) A propagation mechanism is developed in which the user can edit the lumen contour to prevent the algorithm from straying from the direction. (3) In order to improve the accuracy of wall boundary segmentation, a two-pass process is introduced, in which the 2D slice-based segmentation is propagated forward and backward once, resulting in two contours on each axial image. Then, on each axial image, a gradient-based boundary decision process is used to determine the final contour based on the two contours.

[0086] Finally, the present technique used a set of distance- and area-based metrics to evaluate segmentation accuracy. First, the lumen and outer wall boundaries were evaluated separately. Second, the boundaries of the four semi-automatic segmentations were evaluated against the gold standard by comparing their associated mean, maximum thickness, and vessel wall area (i.e., the area enclosed by the lumen and outer wall boundaries).

[0087] Therefore, based on this conclusion, the black blood sequence using this algorithm can produce high-resolution images of the vessel wall and perform automatic vessel cutting on it according to the parameters and the actual situation. Direct comparison shows that compared with existing examinations, this image can show more vessel wall lesions and the contrast between the vessel wall and the surrounding tissue is also higher. The wall thickness measurement can be verified by comparing the MRI imaging of the ex vivo specimen with the histological measurement value.

[0088] The present invention has the following advantages. First, the algorithm can comprehensively, intuitively and delicately display the regions of interest in three-dimensional curved blood vessels at different levels, provide an overall analysis diagram, give people an overall global concept, and provide doctors with more accurate and objective imaging basis. It has irreplaceable auxiliary significance for the diagnosis of vessel wall tissue diseases using black blood technology.

[0089] Secondly, in clinical applications, it avoids the inefficiency caused by traditional manual operations, frees clinicians from massive tomographic images, and greatly improves the efficiency of doctors in diagnosing diseases. Vascular wall imaging is safe, reliable, and can produce excellent diagnostic images. In view of the problem that traditional surface reconstruction has a single vector of interest direction and cannot effectively maintain the trend of vascular shape, this algorithm uses a modified black vascular wall surface reconstruction that simultaneously observes three specific vectors of interest. It can comprehensively, intuitively and delicately display the three-dimensional curved coronary targets and other regions of interest at different levels, and provides an overall analysis diagram to give people an overall global concept. This segmentation program takes about 8-10 minutes to process the 300mm longitudinal coverage of the 3D femoral artery data set, which greatly reduces the analysis time compared to manual inspection. The tool can perform large cross-sectional analysis to assess the risks associated with increased peripheral arterial plaque load, while longitudinal studies can assess the correlation between plaque load size and PAD symptoms. It provides doctors with more accurate and objective imaging basis, which has irreplaceable auxiliary significance for the diagnosis of black vascular wall tissue diseases.

[0090] The present invention also provides a cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction, such as Figure 5 As shown, including:

[0091] A reconstruction module, used to obtain discrete points of the black blood vessel wall contour in the image, and obtain a reconstructed surface of the black blood vessel wall based on the discrete points;

[0092] A sampling module, used for acquiring the center line of the black blood vessel wall reconstructed surface, and performing linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively;

[0093] A mapping module, used to establish three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line;

[0094] The processing module is used to construct a segmentation subnet and a diagnosis subnet, process the sampled data based on the segmentation subnet and the diagnosis subnet, segment the lumen area and the outer wall area, and obtain the morphological information and vascular morphology of the cerebral blood vessels.

[0095] In one embodiment, the acquiring of discrete points of the black blood vessel wall contour in the image and obtaining the black blood vessel wall reconstructed surface based on the discrete points comprises:

[0096] After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve;

[0097] Projecting the contour line along the direction of interest to form a scanned surface;

[0098] The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

[0099] In one embodiment, it further includes:

[0100] The centerline data is sampled, and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data;

[0101] Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data;

[0102] Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

[0103] In one embodiment, performing linear sampling on the X-axis, Y-axis and Z-axis of the center line includes:

[0104] At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously;

[0105] Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

[0106] Among them, the readable storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application-specific integrated circuit (Application Specific Integrated Circuits, abbreviated as: ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the readable storage medium can also exist in a communication device as discrete components. The readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0107] The present invention also provides a program product, which includes an execution instruction, which is stored in a readable storage medium. At least one processor of a device can read the execution instruction from the readable storage medium, and at least one processor executes the execution instruction so that the device implements the methods provided in the above various embodiments.

[0108] In the above-mentioned terminal or server embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cerebral blood vessel segmentation method based on black blood vessel wall surface reconstruction, characterized in that: include: Acquire discrete points of the black blood vessel wall contour in the image, and obtain a reconstructed curved surface of the black blood vessel wall based on the discrete points; Acquire the center line of the black blood vessel wall reconstructed surface, and perform linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively; Establishing three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line; Constructing a segmentation subnet and a diagnosis subnet, processing the sampled data based on the segmentation subnet and the diagnosis subnet, segmenting the lumen area and the outer wall area, and obtaining morphological information and vascular morphology of the cerebral blood vessels; The step of constructing a segmentation subnet and a diagnosis subnet, processing the sampled data based on the segmentation subnet and the diagnosis subnet, segmenting the lumen area and the outer wall area, and obtaining the morphological information and vascular morphology of the cerebral blood vessels includes: A deep U-shaped convolutional neural network with a weighted fusion layer is designed as the segmentation subnet to simultaneously segment the lumen area and the outer wall area, providing a vascular wall map as morphological information; The image,image stream and the morphological stream of the obtained vessel wall map are extracted from two deep convolutional neural networks and combined to obtain the vessel morphology in the diagnosis subnet; The continuity of the boundary is constrained by checking whether the ratio between the cavity areas on adjacent slices is between 0.5 and 1.5, and if the ratio is out of range, the contour on the current image slice is edited; the editing mechanism includes curved plane reformatting of the curved reconstruction view and construction of smooth contours in the CPR view.

2. The cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction according to claim 1, characterized in that: The step of acquiring discrete points of the black blood vessel wall contour in the image and obtaining a reconstructed black blood vessel wall surface based on the discrete points comprises: After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve; Projecting the contour line along the direction of interest to form a scanned surface; The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

3. The cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction according to claim 2, characterized in that: Also includes: The centerline data is sampled and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data; Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data; Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

4. The cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction according to claim 1, characterized in that: The linear sampling on the X-axis, Y-axis and Z-axis of the center line includes: At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously; Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

5. The cerebral blood vessel segmentation method based on black blood vessel wall curved surface reconstruction according to claim 1, characterized in that: The establishing of three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line, comprises: Three imaging planes Csox, Csoy, and Csoz are established to map the sampling data in the three vector directions of interest on the center line respectively, that is, the imaging plane Csox maps the volume data in the Lx vector direction, the imaging plane Csoy maps the volume data in the Ly vector direction, and the imaging plane Csoz maps the volume data in the Lz vector direction.

6. A cerebral blood vessel segmentation device based on black blood vessel wall surface reconstruction, characterized in that: include: A reconstruction module, used to obtain discrete points of the black blood vessel wall contour in the image, and obtain a reconstructed surface of the black blood vessel wall based on the discrete points; A sampling module, used for acquiring the center line of the black blood vessel wall reconstructed surface, and performing linear sampling on the X-axis, Y-axis and Z-axis of the center line respectively; A mapping module, used to establish three imaging planes corresponding to the X-axis, the Y-axis and the Z-axis, wherein the three imaging planes respectively map the sampling data in the directions of three vectors of interest on the center line; A processing module, used for constructing a segmentation subnet and a diagnosis subnet, processing the sampled data based on the segmentation subnet and the diagnosis subnet, segmenting the lumen area and the outer wall area, and obtaining morphological information and vascular morphology of the cerebral blood vessels; Wherein, the processing module is also used for: A deep U-shaped convolutional neural network with a weighted fusion layer is designed as the segmentation subnet to simultaneously segment the lumen area and the outer wall area, providing a vascular wall map as morphological information; The image,image stream and the morphological stream of the obtained vessel wall map are extracted from two deep convolutional neural networks and combined to obtain the vessel morphology in the diagnosis subnet; The continuity of the boundary is constrained by checking whether the ratio between the cavity areas on adjacent slices is between 0.5 and 1.5, and if the ratio is out of range, the contour on the current image slice is edited; the editing mechanism includes curved plane reformatting of the curved reconstruction view and construction of smooth contours in the CPR view.

7. The cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction according to claim 6, characterized in that: The step of acquiring discrete points of the black blood vessel wall contour in the image and obtaining a reconstructed black blood vessel wall surface based on the discrete points comprises: After obtaining the discrete points, interpolating the discrete points to obtain a smooth contour fitting curve; Projecting the contour line along the direction of interest to form a scanned surface; The vascular body data corresponding to the scanned surface is displayed to obtain the black vascular wall reconstructed surface.

8. The cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction according to claim 7, characterized in that: Also includes: The centerline data is sampled and the coordinate values ​​in millimeters are converted into dimensional coordinate values ​​so that they are located in the original black blood vessel wall data; Perform differential calculation on the converted coordinate values, establish a vector of interest at the sampled centerline data, and sample the surface grayscale value of the original black blood vessel wall data; Perform surface reorganization on the sampled data and transform the grayscale value of the reorganized data.

9. The cerebral blood vessel segmentation device based on black blood vessel wall curved surface reconstruction according to claim 6, characterized in that: The linear sampling on the X-axis, Y-axis and Z-axis of the center line includes: At each center point Ci on the center line C, three vectors of interest Lx, Ly, and Lz are set simultaneously; Linear sampling is performed along the X, Y, and Z axis directions, and the sampling lengths correspond to the lengths of the black blood vessel wall data in three dimensions.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, it is used to implement any one of the methods of claims 1 to 5.

Images