A method and apparatus for stent identification and reconstruction

Abstract

This application provides a method and apparatus for stent identification and reconstruction, relating to the field of image processing technology. The stent identification method includes: acquiring a raw intravascular ultrasound (IVUS) image; performing anisotropic diffusion filtering on the raw IVUS image to obtain a stent signal-enhanced IVUS image; processing the stent signal-enhanced IVUS image into a binary mask image containing only the stent signal; and obtaining the stent identification result in the raw IVUS image based on the binary mask image and the raw IVUS image. Because intravascular ultrasound has a deeper imaging depth, even if the stent's depth within the blood vessel increases, intravascular ultrasound can still detect the stent signal, thus avoiding situations where the stent cannot be identified.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method and apparatus for scaffold recognition and reconstruction. Background Technology

[0002] Intravascular coronary stenting is a technique used during percutaneous coronary intervention (PCI) to place a metal mesh catheter within a narrowed artery. This catheter serves to restore blood flow to the vessel wall after balloon angioplasty and prevent acute vascular occlusion. After intravascular stenting, incomplete stent expansion (the stent is correctly aligned with the vessel wall but not fully expanded) or misalignment (the stent is not in complete contact with the vessel wall) may occur. Incomplete stent deployment due to insufficient expansion or misalignment can lead to restenosis and thrombosis. Therefore, stent identification is necessary after placement to determine the stent's location and shape.

[0003] Traditional stent identification methods rely on intravascular optical coherence tomography (OCT). When the stent is first placed in the vessel, it is close to the vessel wall, making it easily identifiable by OCT. However, restenosis can occur some time after placement, increasing the depth of the previously placed stent within the vessel. Because OCT is an optical imaging technique with a shallow imaging depth, it may fail to identify the stent in such cases. Summary of the Invention

[0004] In view of this, this application provides a method and apparatus for stent identification and reconstruction to solve the problem that the shallow depth of OCT imaging in the prior art leads to the inability to identify stents. The technical solution is as follows:

[0005] A method for identifying a stent, comprising:

[0006] Acquire raw intravascular ultrasound (IVUS) images, wherein the raw IVUS images are obtained by performing IVUS imaging on the stent within the blood vessel;

[0007] Anisotropic diffusion filtering is applied to the original IVUS image to obtain an IVUS image with enhanced stent signal;

[0008] The IVUS image with enhanced stent signal is processed into a binary mask image containing only the stent signal;

[0009] Based on the binarized mask image and the original IVUS image, the scaffold recognition result in the original IVUS image is obtained.

[0010] Optionally, anisotropic diffusion filtering is applied to the original IVUS image to obtain an IVUS image with enhanced stent signal, including:

[0011] Anisotropic diffusion filtering with a preset number of iterations is applied to the original IVUS image, and the filtered image after the last iteration is determined as the IVUS image with enhanced stent signal.

[0012] The anisotropic diffusion filtering process in one iteration includes:

[0013] The four-way nearest neighbor difference of the target image is calculated. Based on the four-way nearest neighbor difference of the target image, the four-way transmission coefficient of the current iteration is calculated. Based on the four-way nearest neighbor difference of the target image and the four-way transmission coefficient of the current iteration, the pixel values ​​of the filtered image of the current iteration are calculated to obtain the filtered image of the current iteration. In the case that the current iteration is the first iteration, the target image is the original IVUS image. In the case that the current iteration is not the first iteration, the target image is the filtered image of the previous iteration.

[0014] Optionally, the four-directional nearest neighbor differences include the east-directional nearest neighbor difference, the west-directional nearest neighbor difference, the south-directional nearest neighbor difference, and the north-directional nearest neighbor difference;

[0015] Calculate the four-directional nearest neighbor difference of the target image, including:

[0016] For each pixel in the target image, calculate the first pixel difference between the pixel corresponding to the previous row and the pixel. The calculated first pixel difference is determined as the north nearest neighbor difference of the pixel. If the pixel is the first row of pixels in the target image, the pixel value of the pixel corresponding to the previous row is 0.

[0017] For each pixel in the target image, calculate the second pixel difference between the pixel and the corresponding pixel in the next row. The calculated second pixel difference is determined as the south nearest neighbor difference of the pixel. If the pixel is the last row of pixels in the target image, the pixel value of the corresponding pixel in the next row is 0.

[0018] For each pixel in the target image, calculate the difference between the pixel in the previous column and the third pixel of the pixel. The calculated third pixel difference is determined as the western nearest neighbor difference of the pixel. If the pixel is in the first column of the target image, the pixel value of the pixel in the previous column is 0.

[0019] For each pixel in the target image, calculate the difference between the fourth pixel and the pixel in the next column. The calculated fourth pixel difference is determined as the nearest neighbor difference in the east direction of the pixel. If the pixel is the last column of pixels in the target image, the pixel value of the pixel in the next column is 0.

[0020] Optionally, the four-directional conduction coefficients include the east-direction conduction coefficient, the west-direction conduction coefficient, the south-direction conduction coefficient, and the north-direction conduction coefficient;

[0021] The four-directional transmission coefficients for this iteration are calculated based on the four-directional nearest neighbor differences of the target image, including:

[0022] Based on the difference between the nearest neighbors in the east direction of the target image and the preset edge sensitivity, the east direction transmission coefficient is calculated in this iteration.

[0023] The westward transmission coefficient is calculated based on the westward nearest neighbor difference and edge sensitivity of the target image in this iteration.

[0024] The southward transmission coefficient is calculated based on the southward nearest neighbor difference and edge sensitivity of the target image in this iteration.

[0025] The northward transmission coefficient is calculated based on the northward nearest neighbor difference and edge sensitivity of the target image in this iteration.

[0026] Optionally, the IVUS image with enhanced stent signal is processed into a binary mask image containing only the stent signal, including:

[0027] The IVUS image with enhanced stent signal is binarized by a preset threshold to obtain the binarized image.

[0028] Based on the shape of the support frame and the size of the connected components, connected components are removed from the binarized image to obtain a binarized mask image containing only the support frame signal.

[0029] Optionally, based on the binarized mask image and the original IVUS image, the scaffold recognition result in the original IVUS image is obtained, including:

[0030] The binarized mask image and the original IVUS image are multiplied by a dot to obtain the scaffold recognition result in the original IVUS image.

[0031] A stent reconstruction method, comprising:

[0032] Acquire multiple raw IVUS images, and use any of the above-mentioned stent identification methods to obtain stent identification results in the multiple raw IVUS images;

[0033] Obtain coronary angiography DSA images from two different angiography angles, and perform three-dimensional reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in three-dimensional space.

[0034] The stent identification results from multiple raw IVUS images are located and fused on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0035] Optionally, multiple raw IVUS images can be acquired at equal intervals;

[0036] The stent identification results from multiple raw IVUS images are located and fused on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent, including:

[0037] The stent identification results from multiple raw IVUS images are arranged at equal intervals into the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0038] A bracket identification device, comprising:

[0039] The IVUS image acquisition module is used to acquire raw intravascular ultrasound IVUS images, wherein the raw IVUS images are obtained by performing IVUS imaging on the stent in the blood vessel.

[0040] The stent signal enhancement module is used to perform anisotropic diffusion filtering on the original IVUS image to obtain an IVUS image with stent signal enhancement.

[0041] The mask image determination module is used to process the IVUS image with enhanced stent signal into a binary mask image containing only the stent signal;

[0042] The first stent recognition module is used to obtain the stent recognition result in the original IVUS image based on the binarized mask image and the original IVUS image.

[0043] A stent reconstruction device, comprising:

[0044] The second stent recognition module is used to acquire multiple raw IVUS images and obtain stent recognition results in the multiple raw IVUS images using any of the stent recognition methods mentioned above.

[0045] The 3D reconstruction module for blood vessels is used to acquire coronary angiography DSA images from two different angiography angles, and to perform 3D reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in 3D space.

[0046] The stent 3D reconstruction module is used to locate and fuse the stent identification results from multiple original IVUS images on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0047] As described above, the stent identification method provided in this application uses intravascular ultrasound (IVUS) technology to obtain a raw IVUS image. Then, anisotropic diffusion filtering is applied to the raw IVUS image to enhance the stent signal, resulting in a stent-enhanced IVUS image. This enhanced IVUS image is then processed into a binary mask image containing only the stent signal. Finally, based on the binary mask image and the raw IVUS image, the stent identification result in the raw IVUS image is obtained. Because intravascular ultrasound has a deeper imaging depth, even if the stent's depth within the blood vessel increases, intravascular ultrasound can still detect the stent signal, thus avoiding situations where the stent cannot be identified. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application 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 embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0049] Figure 1 A schematic flowchart illustrating a stent identification method provided in an embodiment of this application;

[0050] Figure 2 A schematic diagram illustrating the positional relationship between pixels and the four-directional transmission coefficients provided in the embodiments of this application;

[0051] Figure 3(a) is a schematic diagram of an original IVUS image with a support provided in an embodiment of this application;

[0052] Figure 3(b) is a schematic diagram of an IVUS image with enhanced stent signal provided in an embodiment of this application;

[0053] Figure 3(c) is a two-dimensional intensity map of the original IVUS image provided in the embodiment of this application;

[0054] Figure 3(d) is a two-dimensional intensity map of the IVUS image with stent signal enhancement provided in the embodiment of this application;

[0055] Figure 4 A schematic flowchart of a stent reconstruction method provided in an embodiment of this application;

[0056] Figure 5 A schematic diagram of discrete point coordinate mapping provided in an embodiment of this application;

[0057] Figure 6 A schematic diagram of a three-dimensional angiography mapping provided in an embodiment of this application;

[0058] Figure 7This is a schematic diagram of stents arranged equidistantly in a three-dimensional blood vessel according to an embodiment of this application;

[0059] Figure 8 This is a schematic diagram of the structure of a bracket identification device provided in an embodiment of this application;

[0060] Figure 9 This is a schematic diagram of a support reconstruction device provided in an embodiment of this application;

[0061] Figure 10 A hardware structure block diagram of a bracket recognition device provided in an embodiment of this application;

[0062] Figure 11 This is a hardware structure block diagram of a scaffold reconstruction device provided in an embodiment of this application. Detailed Implementation

[0063] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0064] To facilitate the explanation of the stent identification and reconstruction method and apparatus provided in this application, the relevant terms used in this application are explained below.

[0065] Intravenous ultrasound (IVUS) is a medical imaging technique that combines non-invasive ultrasound technology with invasive catheter technology. It uses a special catheter with an ultrasound probe attached to its end to perform cross-sectional morphological imaging of the cardiovascular system. As the gold standard for coronary artery examination, IVUS can visualize the morphology inside the blood vessels through cross-sectional morphological imaging.

[0066] Digital subtraction angiography (DSA) is a technique that visualizes blood vessels by digitizing angiography images; DSA can image the overall morphology of blood vessels, but it cannot visualize the morphology inside the blood vessels.

[0067] This application provides a stent identification method, which will be described in detail below through the following embodiments.

[0068] Please see Figure 1 The diagram illustrates a flowchart of a stent identification method provided in an embodiment of this application. The stent identification method may include:

[0069] Step S101: Obtain raw intravascular ultrasound (IVUS) images.

[0070] The original IVUS image is obtained by performing IVUS imaging on the stent within the blood vessel.

[0071] Given that IVUS has a deeper penetration depth than OCT (OCT is 1.5 mm, IVUS is 5 mm), in order to still be able to identify the stent location after it is encased in vascular tissue (i.e., restenosis), intravascular ultrasound technology can be used to perform IVUS imaging on the stent in the blood vessel to obtain the original IVUS image.

[0072] Because the stent is placed close to the inner wall of the blood vessel in the early stages, and the acoustic impedance of the stent is significantly greater than that of the vascular tissue, the stent appears as a speckle pattern with a strong signal on the original IVUS image. This application can determine the stent location by the position of the specific speckle pattern formed by the stent strut on the original IVUS image.

[0073] Step S102: Perform anisotropic diffusion filtering on the original IVUS image to obtain an IVUS image with enhanced stent signal.

[0074] This application can enhance the signal at the location of the support by using anisotropic diffusion filtering. Here, anisotropic diffusion filtering treats the image as a force field or thermal flow field in physics, where image pixels always flow or move towards areas whose values ​​are not significantly different from their own, thus preserving areas with large differences (such as edges).

[0075] Step S103: Process the IVUS image with enhanced stent signal into a binarized mask image containing only the stent signal.

[0076] Since the pixel difference between the stent position and other positions in the IVUS image with stent signal enhancement is large, by selecting an appropriate threshold, the IVUS image with stent signal enhancement can be processed into a binary mask image containing only the stent signal.

[0077] In this application, the pixel value at the bracket position in the binarized mask image is different from the pixel value at other positions. For example, optionally, the pixel value at the bracket position is 1, and the pixel value at other positions is 0.

[0078] Step S104: Based on the binarized mask image and the original IVUS image, obtain the scaffold recognition result in the original IVUS image.

[0079] This application can mask the pixel values ​​at other locations in the original IVUS image using a binary mask image, leaving only the pixels at the stent location, to obtain the stent identification result in the original IVUS image. This application can determine the location of the stent in the blood vessel using the stent identification result.

[0080] The stent identification method provided in this application uses intravascular ultrasound (IVUS) technology to obtain a raw IVUS image. Then, anisotropic diffusion filtering is applied to the raw IVUS image to enhance the stent signal, resulting in a stent-enhanced IVUS image. This enhanced IVUS image is then processed into a binary mask image containing only the stent signal. Finally, the stent identification result in the raw IVUS image is obtained based on the binary mask image and the raw IVUS image. Because intravascular ultrasound has a deeper imaging depth, it can still detect stent signals even as the stent's depth within the blood vessel increases, thus avoiding situations where the stent cannot be identified.

[0081] The following describes "step S102, performing anisotropic diffusion filtering on the original IVUS image to obtain an IVUS image with enhanced stent signal" in the above embodiment.

[0082] In this embodiment, in order to better highlight the stent signal on the IVUS image, the original IVUS image can be subjected to multiple anisotropic diffusion filters. That is, this application can perform anisotropic diffusion filters on the original IVUS image for a preset number of iterations, and determine the filtered image under the last iteration as the IVUS image with enhanced stent signal.

[0083] Optionally, the process of anisotropic diffusion filtering in one iteration may include: calculating the four-directional nearest neighbor difference of the target image, calculating the four-directional transmission coefficient in this iteration based on the four-directional nearest neighbor difference of the target image, and calculating the pixel value of the filtered image in this iteration based on the four-directional nearest neighbor difference of the target image and the four-directional transmission coefficient in this iteration, so as to obtain the filtered image in this iteration.

[0084] In the case that this iteration is the first iteration, the target image is the original IVUS image; in the case that this iteration is not the first iteration, the target image is the filtered image from the previous iteration.

[0085] Specifically, the gray-level diffusion equation of the filtered image in the t-th iteration is:

[0086]

[0087] In the formula, I tLet represent the filtered image at the t-th iteration, div represent the divergence operator, ▽ represent the gradient operator, Δ represent the Laplacian operator, and c(x,y,t) represent the conduction coefficient (also known as the diffusion coefficient).

[0088] In this application, c(x,y,t) is a constant. When c(x,y,t) is a constant, the filtered image I under the t-th iteration is... t The gray diffusion equation can be equivalent to the isotropic thermal diffusion equation shown in the following formula (2).

[0089] I t =cΔI Formula (2)

[0090] Suppose that in the t-th iteration, we want to smooth a certain region of the filtered image from the previous iteration, but we don't want to smooth the boundary of that region. We can set the thermal diffusivity coefficient to 1 inside each region and set the thermal diffusivity coefficient to 0 at the boundary. The blurring of the filtered image from the previous iteration will be completed within that region, but the boundary of the region will remain sharp.

[0091] In short, anisotropic diffusion filters treat the original IVUS image as a thermal field, with image pixels representing the diffused heat within this field. During anisotropic diffusion filtering, the change in a pixel depends on its relationship to surrounding pixels within the current region. Based on the thermal diffusivity coefficient set above, when a pixel has no surrounding edge pixels, its thermal diffusivity coefficient is larger, resulting in a smoother diffused area; conversely, if the pixel has surrounding edge pixels, its thermal diffusivity coefficient is smaller, weakening or halting diffusion. By applying multiple anisotropic diffusion filters to the original IVUS image with the scaffold, tissue noise can be smoothed while preserving the scaffold.

[0092] The following section describes the process of performing anisotropic diffusion filtering on the target image.

[0093] Anisotropic diffusion, scale space, and edge detection of the target image can be described by a formula, formula (1), which can be discretized on a square grid. The square grid is equivalent to the matrix of the target image, and the pixels of the target image are located at the vertices of the grid. The conduction coefficients between pixels are associated with the connecting lines between the vertices of the grid. Taking the anisotropic diffusion filtering in the (t+1)th iteration as an example, as follows... Figure 2 We can use the four nearest neighbors discretization of the Laplace operator to rearrange equation (2) into the following equation (3):

[0094]

[0095] In the formula, λ represents the weighting coefficient (0≤λ≤1 / 4), c N cS c E and c W This represents the four-directional transmission coefficient (which is the four components of the constant term c in formula (2) above). and The nearest neighbor difference represents the difference in four directions (which is also the gradient operator mentioned above), where N, S, E, and W represent the four directions: north, south, east, and west, respectively. Let represent the pixel value at position (i, j) in the filtered image after the t-th iteration (in formula (3), this image is the target image). This represents the pixel value at position (i, j) in the filtered image after the (t+1)th iteration.

[0096] As can be seen from the above formula (3), the four-directional nearest neighbor differences in this application include: the nearest neighbor difference in the east direction. Westward nearest neighbor difference South nearest neighbor difference Differences from the nearest neighbor to the north The four-directional transmission coefficients include: the eastward transmission coefficient c. E , Westward transmission coefficient c W Southward transmission coefficient c S And the northward transmission coefficient c N As long as the nearest neighbor difference and the transmission coefficient in the four directions are obtained, they can be substituted into formula (3) to obtain the pixel value of each pixel in the filtered image under the (t+1)th iteration, thus obtaining the filtered image under the (t+1)th iteration.

[0097] Optionally, for each pixel in the target image, use pixel I i，j (equivalent to formula (3)) For ease of subsequent description, the "t" will be omitted. Taking this example, let's calculate the pixel I... i，j The process of determining the nearest neighbor difference in four directions includes the following steps S11 to S14:

[0098] Step S11: Calculate the difference between the first pixel of the pixel and the corresponding pixel in the previous row. Determine the calculated first pixel difference as the nearest neighbor difference in the north direction of the pixel. If the pixel is the first row of pixels in the target image, the pixel value of the corresponding pixel in the previous row is 0.

[0099] Specifically, pixel I can be calculated using formula (4). i，j North nearest neighbor difference

[0100]

[0101] In the formula, Ii-1，j Indicates pixel I i，j The previous row corresponds to the pixels.

[0102] Step S12: Calculate the second pixel difference between the pixel and the pixel in the next row. Determine the calculated second pixel difference as the nearest neighbor difference in the south direction of the pixel. If the pixel is the last row of pixels in the target image, the pixel value of the pixel in the next row is 0.

[0103] Specifically, pixel I can be calculated using formula (5). i，j South nearest neighbor difference

[0104]

[0105] In the formula, I i+1，j Indicates pixel I i，j The next row corresponds to the pixels.

[0106] Step S13: Calculate the difference between the third pixel of the pixel and the previous column corresponding pixel. Determine the calculated third pixel difference as the western nearest neighbor difference of the pixel. If the pixel is the first column of pixels in the target image, the pixel value of the previous column corresponding pixel is 0.

[0107] Specifically, pixel I can be calculated using formula (6). i，j The difference between the nearest neighbors in the west direction

[0108]

[0109] In the formula, I i，j-1 Indicates pixel I i，j The corresponding pixel in the previous column.

[0110] Step S14: Calculate the difference between the fourth pixel of the pixel and the next corresponding pixel in the next column. Determine the calculated fourth pixel difference as the nearest neighbor difference in the east direction of the pixel. If the pixel is the last column of pixels in the target image, the pixel value of the next corresponding pixel in the next column is 0.

[0111] Specifically, pixel I can be calculated using formula (7). i，j Eastward nearest neighbor difference

[0112]

[0113] In the formula, I i，j+1 Indicates pixel I i，j The next column corresponds to the pixels.

[0114] In an optional embodiment, the process of “calculating the four-directional transmission coefficients in this iteration based on the four-directional nearest neighbor differences of the target image” is described.

[0115] It should be noted that although the four-directional transmission coefficients of each pixel do not change during a single iteration, they are updated for each subsequent iteration. Therefore, the updated four-directional transmission coefficients need to be calculated for each iteration.

[0116] Alternatively, the four-directional transmission coefficient can be calculated using the following formula (8):

[0117]

[0118] In the formula, the constant term K is used to control the sensitivity to the edge, and is usually selected based on the strength of the scaffold and the noise of the tissue.

[0119] As previously mentioned, c has four components, namely c0, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10, c11, c2N c S c E and c W The above formula (8) can be rewritten as the formulas for the four components as follows:

[0120]

[0121]

[0122]

[0123]

[0124] In the formula, Indicates pixel I i，j The northward transmission coefficient, Indicates pixel I i，j The southward transmission coefficient, Indicates pixel I i，j The westward transmission coefficient, Indicates pixel I i，j The eastward transmission coefficient.

[0125] Based on the above formula (9), this application can determine the target image based on the nearest neighbor difference in the north direction. Given the edge sensitivity K, calculate the northward transmission coefficient c in this iteration. N That is, for each pixel in the target image (denoted by I) i，j For example, based on the difference between the nearest neighbor in the north direction of the pixel. And K, calculate the northward transmission coefficient of the pixel in this iteration.

[0126] Accordingly, based on the above formula (10), this application can determine the target image based on the nearest neighbor difference in the south direction. Given the edge sensitivity K, calculate the southward transmission coefficient c in this iteration. S That is, for each pixel in the target image (denoted by I) i，j (For example), based on the difference of the nearest neighbor in the south direction of this pixel) And K, calculate the southward transmission coefficient of the pixel in this iteration.

[0127] Based on the above formula (11), this application can determine the target image based on the westward nearest neighbor difference. Given the edge sensitivity K, calculate the westward transmission coefficient c in this iteration. W That is, for each pixel in the target image (denoted by I) i，j For example, based on the difference between the pixel's western nearest neighbors... And K, calculate the westward transmission coefficient of the pixel in this iteration.

[0128] Based on the above formula (12), this application can determine the nearest neighbor difference in the east direction of the target image. Given the preset edge sensitivity K, calculate the eastward transmission coefficient c for this iteration. E That is, for each pixel in the target image (denoted by I) i，j For example, based on the difference between the nearest neighbors in the east direction of this pixel. And K, calculate the eastward transmission coefficient of the pixel in this iteration.

[0129] Substituting the results obtained from the above formulas (4) to (7) and formulas (9) to (12) into formula (3), we can obtain the results of the original IVUS image after anisotropic diffusion, where the stent signal is enhanced and the tissue and noise signals are suppressed.

[0130] Taking the original IVUS image with a scaffold shown in Figure 3(a) as an example, the IVUS image with enhanced scaffold signal obtained after multiple anisotropic diffusion filters is shown in Figure 3(b). Compared to Figure 3(a), the scaffold signal in Figure 3(b) is more obvious. To more clearly show the difference between the two images, this application also provides two-dimensional intensity maps of the two images, where Figure 3(c) is the two-dimensional intensity map of the original IVUS image shown in Figure 3(a), and Figure 3(d) is the two-dimensional intensity map of the IVUS image with enhanced scaffold signal shown in Figure 3(b). Compared to Figure 3(c), Figure 3(d) effectively filters out noise pixels other than the scaffold signal, thus highlighting the scaffold signal much better.

[0131] In summary, this application takes into account that the stent is close to the inner wall of the blood vessel during early placement, and the acoustic impedance of the stent is significantly greater than that of the blood vessel tissue. Therefore, the stent appears as a speckle pattern with a strong signal in the original IVUS image. The signal at the location of the stent is enhanced by anisotropic diffusion filtering, making it easier to distinguish the stent location in the IVUS image with enhanced stent signal.

[0132] An embodiment of this application describes the aforementioned processes of "step S103, processing the IVUS image with enhanced stent signal into a binary mask image containing only stent signal" and "step S104, obtaining the stent identification result in the original IVUS image based on the binary mask image and the original IVUS image".

[0133] Specifically, the process of "step S103, processing the IVUS image with enhanced stent signal into a binary mask image containing only the stent signal" may include:

[0134] Step S21: Binarize the IVUS image with enhanced stent signal using a preset threshold to obtain the binarized image.

[0135] Here, all the support signals are included in the binarized image.

[0136] Specifically, because this application pre-enhances the signals of the pixels representing the scaffold in the original IVUS image through anisotropic diffusion filtering, the pixel values ​​of those pixels representing the scaffold are larger in the IVUS image with enhanced scaffold signals, while the pixel values ​​of the remaining pixels outside the scaffold are relatively smaller. Therefore, by setting an appropriate threshold, the pixels representing the scaffold can be easily distinguished from the other pixels, resulting in a binarized image.

[0137] For example, optionally, the binarization process of this application includes: adjusting the pixel values ​​in the IVUS image with enhanced stent signal that are greater than or equal to a preset threshold to 1, and adjusting the pixel values ​​that are less than the preset threshold to 0, to obtain the binarized image.

[0138] Step S22: Based on the shape of the support and the size of the connected components, perform connected component deletion on the binarized image to obtain a binarized mask image containing only the support signal.

[0139] Specifically, this step removes noise by removing smaller connected components from the binarized image, resulting in a binarized mask image containing only the support signal.

[0140] The method for removing connected components is an existing technology and will not be described in detail here.

[0141] In this application, the IVUS image with stent signal enhancement is denoted as I. S , to I S Binarization processing yields the binarized image I. BS By removing I BS By using smaller connected components to remove noise, a binary mask image of the support structure is finally obtained. Mask .

[0142] Because of I Mask The image contains only the support signal. In step S104, the binarized mask image I can be used. Mask Perform a dot product with the original IVUS image I to obtain the scaffold recognition result I in the original IVUS image. Stent That is, in step S104, the stent identification result I in the original IVUS image can be obtained using formula (13). Stent .

[0143] I Stent =I.*I Mask Formula (13)

[0144] In the formula, ".*" represents the dot product symbol.

[0145] After the stent identification method provided in the foregoing embodiments has segmented the stent, it can also perform three-dimensional reconstruction of the stent according to the embodiments of this application, so that relevant personnel can understand the three-dimensional morphology of the stent in the blood vessel in a timely manner.

[0146] Therefore, this application also provides a stent reconstruction method, see [link to relevant documentation]. Figure 4 This is a schematic flowchart of the stent reconstruction method provided in the embodiments of this application.

[0147] like Figure 4 As shown, the stent reconstruction method provided in this application may include:

[0148] Step S401: Acquire multiple raw IVUS images and use a stent recognition method to obtain stent recognition results in the multiple raw IVUS images.

[0149] Specifically, IVUS can only visualize the morphology inside blood vessels through cross-sectional morphology imaging. However, the stent identification result of a single original IVUS image only reflects one cross-sectional morphology of the stent. If three-dimensional reconstruction of the stent is required, multiple original IVUS images need to be acquired, and then the stent identification results in multiple original IVUS images can be obtained using the stent identification method provided in any of the above embodiments.

[0150] Step S402: Obtain coronary angiography DSA images from two different angiography angles, and perform three-dimensional reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in three-dimensional space.

[0151] This step, which involves obtaining blood vessels in three-dimensional space from DSA images at two different angiography angles, is an existing technology. To make the solution of this application more complete, it is briefly described below.

[0152] Specifically, the three-dimensional reconstruction of vascular structures can be transformed into a problem of three-dimensional reconstruction of discrete points under two different angiographic angles. (Combined with...) Figure 5 To explain the principle of 3D reconstruction, assume S1 and S2 are X-ray sources, U1V1O1 and U2V2O2 are the two acquisition planes for angiography (i.e., DSA images at two different angiography angles), S1O1 and S2O2 are the axes from the centers of the two acquisition planes to the centers of the projection coordinate systems X1Y1Z1S1 and X2Y2Z2S2, α1 and α2 represent the left and right angles of the two angiography images relative to the coordinate axes ZYZO, and β1 and β2 represent the front and back angles of the angiography. Here, α1 refers to the angle formed by A1OZ, α2 refers to the angle formed by A2OZ, β1 refers to the angle formed by A2OO2, and β2 refers to the angle formed by A1OO1. Take a point on each of the two acquisition planes, for example, P. 1j (u 1j v 1j ) and p 2j (u 2j v 2j Assume p 1j and p 2j These are two points on the same blood vessel at the same location on the angiographic image. Using two points as an example can illustrate the three-dimensional reconstruction of blood vessels. For two X-ray source coordinate systems X... i Y i Z i S i (i.e., X1Y1Z1S1 and X2Y2Z2S2 mentioned above), p 1j and p 2j The coordinates within it are (x i y i , z i (Specifically, in X1Y1Z1S1, the coordinates are (x1, y1, z1), and in X2Y2Z2S2, they are (x2, y2, z2).) Based on geometric relationships, the geometric relationship between coordinate system X1Y1Z1S1 and X2Y2Z2S2 is as follows:

[0153]

[0154] Where R is the rotation matrix. As a translation matrix, the coordinate system XYZO can also be transformed to X using translation and rotation. i Y i Z i S i The transformation, namely:

[0155] [x i y i , z i ,1]=[X,Y,Z,1]·R Y (α i )·R X (β i )·T i Formula (15)

[0156] Among them, T i Represents translation along the Z-axis, matrix R X (β i ) indicates a clockwise rotation β around the X-axis i Angle, R Y (α i ) indicates a clockwise rotation α around the Y-axis i Angle, where β i Including β1 and β2 mentioned above, α i Including α1 and α2. In practice, it is known that (u 1j v 1j ) and (u 2j v 2j p can be solved using the translation and rotation matrices. j The three-dimensional coordinates, p j It refers to p 1j S1 and p 2j The intersection of S2.

[0157] Similarly, by selecting several pairs of matching points on two DSA images, the two-dimensional blood vessels can be mapped as... Figure 6 The vascular structure shown is in three-dimensional space.

[0158] Step S403: The stent identification results from multiple original IVUS images are located and fused on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0159] In an optional embodiment, when acquiring multiple raw IVUS images in step S401, multiple raw IVUS images acquired at equal intervals can be acquired. When acquiring multiple raw IVUS images at equal intervals, it can be assumed that the acquired images are also equally spaced in the reconstructed vascular spatial structure. Therefore, step S403 may include: arranging the stent identification results from the multiple raw IVUS images at equal intervals into the blood vessels in three-dimensional space to obtain the three-dimensional structure of the stent, such as... Figure 7As shown.

[0160] In summary, this embodiment of the application performs IVUS imaging on the target in the early stage of stent placement, supplemented by DSA imaging. Because the stent is close to the vessel wall in the early placement stage, and the acoustic impedance of the stent is significantly greater than that of the vessel tissue, the stent appears as a speckle pattern with a strong signal in the original IVUS image. Anisotropic diffusion filtering is used to enhance the signal at the stent location. A threshold is set for measurement, and the enhanced image is binarized. At this point, all stent signals are contained in this binary image. Then, based on the shape of the stent and the size of its connected components, the binary image is further processed. Connected component deletion is performed, resulting in a binary image containing only the stent signal. This binary image is used as a mask, and matrix multiplication is performed on the IVUS image with stent signal enhancement to obtain a matrix of stent position and intensity. By measuring the point spread function (PSF) of the stent, a normalized two-dimensional cross-correlation is performed with the matrix obtained in the previous step. An appropriate threshold is set to remove unnecessary noise, and the spatial coordinates of the centroid of the cross-correlated image are extracted one by one to locate individual stent points. Finally, combined with the spatial position and structural information of the blood vessel obtained from DSA, the stent is accumulated frame by frame to the corresponding position of the blood vessel to obtain the final three-dimensional stent image.

[0161] This application provides a method for intravascular ultrasound image stent identification and localization (3D reconstruction), which provides an effective method for stent placement and immediate post-stent placement effect evaluation and analysis. Combining DSA for 3D stent reconstruction can visualize the structure and spatial location of the stent, providing more information in the early stage of stent placement.

[0162] This application also provides a bracket identification device, which is described below.

[0163] Please see Figure 8 The diagram shows a structural schematic of the bracket identification device provided in an embodiment of this application, such as... Figure 8 As shown, the bracket recognition device may include: an IVUS image acquisition module 801, a bracket signal enhancement module 802, a mask image determination module 803, and a first bracket recognition module 804.

[0164] IVUS image acquisition module 801 is used to acquire raw intravascular ultrasound IVUS images, wherein the raw IVUS images are obtained by performing IVUS imaging on a stent within the blood vessel.

[0165] The stent signal enhancement module 802 is used to perform anisotropic diffusion filtering on the original IVUS image to obtain an IVUS image with stent signal enhancement.

[0166] The mask image determination module 803 is used to process the IVUS image with enhanced stent signal into a binary mask image containing only the stent signal.

[0167] The first support recognition module 804 is used to obtain the support recognition result in the original IVUS image based on the binarized mask image and the original IVUS image.

[0168] The working principle of the stent identification device provided in this application is the same as that of the stent identification method described above. For details, please refer to the above description, which will not be repeated here.

[0169] This application also provides a stent reconstruction device, which is described below.

[0170] Please see Figure 9 The diagram shows a structural schematic of the scaffold reconstruction device provided in an embodiment of this application, as shown below. Figure 9 As shown, the stent reconstruction device may include: a second stent identification module 901, a vascular three-dimensional reconstruction module 902, and a stent three-dimensional reconstruction module 903.

[0171] The second stent recognition module 901 is used to acquire multiple raw IVUS images and use a stent recognition method to obtain stent recognition results in the multiple raw IVUS images.

[0172] The 3D reconstruction module 902 is used to acquire coronary angiography DSA images from two different angiography angles, and to perform 3D reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in 3D space.

[0173] The stent 3D reconstruction module 903 is used to locate and fuse the stent identification results from multiple original IVUS images on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0174] The working principle of the stent reconstruction device provided in this application is the same as that of the stent reconstruction method described above. For details, please refer to the above description, which will not be repeated here.

[0175] This application also provides a bracket identification device. Optionally, Figure 10 The hardware structure block diagram of the bracket recognition device is shown below. Figure 10 The hardware structure of the bracket identification device may include: at least one processor 1001, at least one communication interface 1002, at least one memory 1003 and at least one communication bus 1004.

[0176] In this embodiment of the application, the number of processor 1001, communication interface 1002, memory 1003 and communication bus 1004 is at least one, and processor 1001, communication interface 1002 and memory 1003 communicate with each other through communication bus 1004.

[0177] The processor 1001 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0178] The memory 1003 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device;

[0179] The memory 1003 stores a program, and the processor 1001 can call the program stored in the memory 1003. The program is used for:

[0180] Acquire raw intravascular ultrasound (IVUS) images, wherein the raw IVUS images are obtained by performing IVUS imaging on the stent within the blood vessel;

[0181] Anisotropic diffusion filtering is applied to the original IVUS image to obtain an IVUS image with enhanced stent signal;

[0182] The IVUS image with enhanced stent signal is processed into a binary mask image containing only the stent signal;

[0183] Based on the binarized mask image and the original IVUS image, the scaffold recognition result in the original IVUS image is obtained.

[0184] Optionally, the refined and extended functions of the program can be found in the description above.

[0185] This application also provides a scaffold reconstruction device. Optionally, Figure 11 The hardware structure block diagram of the scaffold reconstruction device is shown, with reference to... Figure 11 The hardware structure of the support reconstruction device may include: at least one processor 1101, at least one communication interface 1102, at least one memory 1103 and at least one communication bus 1104;

[0186] In this embodiment of the application, the number of processor 1101, communication interface 1102, memory 1103 and communication bus 1104 is at least one, and processor 1101, communication interface 1102 and memory 1103 communicate with each other through communication bus 1104.

[0187] The processor 1101 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0188] The memory 1103 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device;

[0189] The memory 1103 stores a program, and the processor 1101 can call the program stored in the memory 1103. The program is used for:

[0190] Multiple raw IVUS images were acquired, and the stent recognition method was used to obtain the stent recognition results in the multiple raw IVUS images.

[0191] Obtain coronary angiography DSA images from two different angiography angles, and perform three-dimensional reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in three-dimensional space.

[0192] The stent identification results from multiple raw IVUS images are located and fused on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent.

[0193] Optionally, the refined and extended functions of the program can be found in the description above.

[0194] This application also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the bracket identification method described above.

[0195] Optionally, the refined and extended functions of the program can be found in the description above.

[0196] This application also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the scaffold reconstruction method described above.

[0197] Optionally, the refined and extended functions of the program can be found in the description above.

[0198] Finally, it should be noted that in this document, relational terms such as "second" and "etc." are used merely 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.

[0199] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0200] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for stent reconstruction, characterized in that, include: Multiple raw IVUS images are acquired, and a stent identification method is used to obtain stent identification results in the multiple raw IVUS images. Obtain coronary angiography DSA images from two different angiography angles, and perform three-dimensional reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in three-dimensional space. The stent identification results from multiple original IVUS images are located and fused on the blood vessel in the three-dimensional space to obtain the three-dimensional structure of the stent; wherein: The plurality of original IVUS images are images acquired at equal intervals; The step of locating and fusing the stent identification results from multiple original IVUS images on the blood vessel in three-dimensional space to obtain the three-dimensional structure of the stent includes: The stent identification results from multiple original IVUS images are arranged at equal intervals into the blood vessels in the three-dimensional space to obtain the three-dimensional structure of the stent; The bracket identification method includes: Acquire raw intravascular ultrasound (IVUS) images, wherein the raw IVUS images are obtained by performing IVUS imaging on the stent within the blood vessel; Anisotropic diffusion filtering is applied to the original IVUS image to obtain an IVUS image with enhanced stent signal; The IVUS image with enhanced stent signal is processed into a binary mask image containing only the stent signal; Based on the binarized mask image and the original IVUS image, the scaffold identification result in the original IVUS image is obtained.

2. The stent reconstruction method according to claim 1, characterized in that, The process of performing anisotropic diffusion filtering on the original IVUS image to obtain an IVUS image with enhanced stent signal includes: The original IVUS image is subjected to anisotropic diffusion filtering for a preset number of iterations, and the filtered image after the last iteration is determined as the IVUS image with enhanced stent signal. The anisotropic diffusion filtering process in one iteration includes: The four-way nearest neighbor difference of the target image is calculated. Based on the four-way nearest neighbor difference of the target image, the four-way transmission coefficient of the current iteration is calculated. Based on the four-way nearest neighbor difference of the target image and the four-way transmission coefficient of the current iteration, the pixel values ​​of the filtered image of the current iteration are calculated to obtain the filtered image of the current iteration. Wherein, if the current iteration is the first iteration, the target image is the original IVUS image. If the current iteration is not the first iteration, the target image is the filtered image of the previous iteration.

3. The stent reconstruction method according to claim 2, characterized in that, The four-directional nearest neighbor differences include the east-directional nearest neighbor difference, the west-directional nearest neighbor difference, the south-directional nearest neighbor difference, and the north-directional nearest neighbor difference; The calculation of the four-directional nearest neighbor difference of the target image includes: For each pixel in the target image, calculate the first pixel difference between the pixel and the corresponding pixel in the previous row. The calculated first pixel difference is determined as the north nearest neighbor difference of the pixel. If the pixel is a pixel in the first row of the target image, the pixel value of the corresponding pixel in the previous row is 0. For each pixel in the target image, calculate the second pixel difference between the pixel and the pixel in the next row. The calculated second pixel difference is determined as the south nearest neighbor difference of the pixel. If the pixel is the last row of pixels in the target image, the pixel value of the pixel in the next row is 0. For each pixel in the target image, calculate the difference between the pixel in the previous column and the third pixel of the pixel. The calculated third pixel difference is determined as the western nearest neighbor difference of the pixel. If the pixel is a pixel in the first column of the target image, the pixel value of the pixel in the previous column is 0. For each pixel in the target image, calculate the difference between the fourth pixel and the pixel in the next column. The calculated fourth pixel difference is determined as the nearest neighbor difference in the east direction of the pixel. If the pixel is the last column of pixels in the target image, the pixel value of the pixel in the next column is 0.

4. The stent reconstruction method according to claim 3, characterized in that, The four-directional conduction coefficients include the east-direction conduction coefficient, the west-direction conduction coefficient, the south-direction conduction coefficient, and the north-direction conduction coefficient; The step of calculating the four-directional transmission coefficients for this iteration based on the four-directional nearest neighbor differences of the target image includes: Based on the eastward nearest neighbor difference of the target image and the preset edge sensitivity, the eastward transmission coefficient in this iteration is calculated; The westward transmission coefficient is calculated in this iteration based on the westward nearest neighbor difference of the target image and the edge sensitivity. The southward transmission coefficient is calculated in this iteration based on the southward nearest neighbor difference of the target image and the edge sensitivity. The northward transmission coefficient is calculated based on the northward nearest neighbor difference of the target image and the edge sensitivity in this iteration.

5. The stent reconstruction method according to claim 1, characterized in that, The process of converting the IVUS image enhanced with the stent signal into a binary mask image containing only the stent signal includes: The IVUS image with enhanced stent signal is binarized by a preset threshold to obtain a binarized image. Based on the shape of the support and the size of the connected components, the connected components of the binarized image are deleted to obtain the binarized mask image containing only the support signal.

6. The stent reconstruction method according to claim 1, characterized in that, Based on the binarized mask image and the original IVUS image, the scaffold identification result in the original IVUS image is obtained, including: The binarized mask image and the original IVUS image are multiplied by a dot product to obtain the scaffold recognition result in the original IVUS image.

7. A scaffold reconstruction device, characterized in that, include: The second stent recognition module is used to acquire multiple original IVUS images and use the stent reconstruction method described in claim 1 to obtain stent recognition results in the multiple original IVUS images. The 3D reconstruction module for blood vessels is used to acquire coronary angiography DSA images from two different angiography angles, and to perform 3D reconstruction of the blood vessel where the stent is located based on the two DSA images to obtain the blood vessel in 3D space. The stent 3D reconstruction module is used to locate and fuse the stent identification results from multiple original IVUS images on the blood vessel in the three-dimensional space to obtain the three-dimensional structure of the stent.

Images