Rapid marking method for cerebral vein collateral circulation in magnetic sensitive weighted sequence

Abstract

The invention relates to the technical field of image analysis, in particular to a cerebral vein collateral circulation rapid labeling method in a magnetic sensitive weighted sequence. A plurality of mIP images of the head of a patient are obtained and segmented to obtain image blocks. And acquiring connected domains based on the gray information of the pixel points in the image blocks, and determining the vascular structure region density according to the shape and number characteristics of the connected domains. And determining a cerebral vein compensation index to evaluate the compensation capability of the cerebral vein structure by analyzing the gray distribution, shape and position relationship between the connected domains. And determining a collateral circulation saliency value of each image block by adopting a saliency algorithm in combination with the image block position, the vascular structure region density and the cerebral vein compensation index, and quantifying the saliency degree of the collateral circulation. And finally, obtaining a saliency map through multi-scale fitting, analyzing the final saliency value of the pixel point, and marking the cerebral vein collateral region of interest. Various change characteristics of cerebral vein collateral circulation are combined, and the accuracy of cerebral vein structure labeling is improved.

Description

Technical Field

[0001] The invention relates to the technical field of image analysis, and in particular to a method for quickly marking cerebral venous collateral circulation in a magnetic susceptibility weighted sequence. Background Art

[0002] Susceptibility Weighted Imaging (SWI) can enhance the imaging ability of cerebral veins in the brain through the magnetic sensitivity effect, and can clearly show the changes in deoxyhemoglobin concentration. It has important clinical application value in diagnosing cerebral infarction and evaluating collateral circulation. Cerebral venous collateral circulation (or cerebral venous bypass circulation) mainly provides a blood flow path through the cerebral venous network, which is particularly important in certain brain lesions (such as cerebrovascular lesions, cerebral infarction, etc.).

[0003] The current cerebral venous collateral circulation labeling method is mainly based on the automatic labeling technology of machine vision. However, the existing technology often only relies on the grayscale features of the cerebral venous structure image in SWI imaging to mark the cerebral venous collateral interest area. However, cerebral venous collateral circulation is a complex physiological process. The morphological characteristics, grayscale characteristics and density of cerebral veins will change. The processing of a single feature will lead to the final labeling result being inaccurate and incomplete. Summary of the invention

[0004] In order to solve the technical problem that cerebral venous collateral circulation is a complex physiological process, the morphological characteristics, grayscale characteristics and density of cerebral venous vessels will change, and the processing of a single feature will lead to the final annotation result being inaccurate and incomplete, the purpose of the present invention is to provide a method for rapid annotation of cerebral venous collateral circulation in a susceptibility weighted sequence, and the technical scheme adopted is as follows: Acquire multiple mIP images of the patient's head; At different preset scales, each mIP image is segmented to obtain image blocks; in each image block, the connected domain is obtained based on the grayscale information of the pixels; in each image block, the vascular structure area density of each image block is determined based on the shape characteristics of the connected domain and the number characteristics of the pixels; At any preset scale, the grayscale distribution, shape characteristics, and relative position relationship between connected domains in each mIP image are analyzed to determine the cerebral venous compensation index of each connected domain; In each mIP image, the collateral circulation significance value of each image block is determined based on the positional relationship between image blocks, the regional density of vascular structures in the image blocks, and the cerebral venous compensation index of the connected domain in the image blocks; multi-scale fitting is performed on each mIP image based on the significance algorithm to obtain a saliency map and the final saliency value of each pixel in the saliency map; the final saliency value of the pixel in each saliency map is analyzed to mark the cerebral venous collateral interest area.

[0005] Furthermore, the method for obtaining the connected domain includes: A binary image of each image block is obtained based on the Otsu threshold segmentation algorithm, in which pixels with a gray value of 0 are regarded as pixels of the cerebral vein structure; In each image block, the connected domain analysis is performed on all the pixels of the cerebral vein structure to obtain the connected domain.

[0006] Furthermore, the method for obtaining the regional density of the vascular structure includes: In each image block, after performing convex hull detection on each connected domain, the ratio of the number of pixels in the convex hull to the number of pixels in the connected domain is used as the quantity weight; The number of pixels in the convex hull is weightedly fused using the number weights corresponding to the connected domain to obtain the density factor corresponding to each image block; The value obtained by normalizing the ratio of the density factor corresponding to each image block to the number of pixels in each image block is used as the vascular structure area density of each image block.

[0007] Furthermore, the method for obtaining the cerebral venous compensation index includes: In each connected domain, the grayscale fluctuation of the pixels and the shape characteristics of the connected domain are analyzed to obtain the cerebral vein structure index of each connected domain; A coordinate system is constructed in each mIP image and the symmetry axis is obtained. For any image block, the area symmetrical about the symmetry axis in the mIP image is used as the symmetric block of the image block. In the symmetric block corresponding to each image block, the connected domain analysis is performed on all the pixel points of the cerebral venous structure to obtain the comparative connected domain. In each image block and the corresponding symmetric block, the Hausdorff distance between any connected domain and any contrast connected domain is calculated, and the value after negative correlation mapping of the minimum value of all Hausdorff distances is used as the symmetry index of all connected domains in each image block; According to the cerebral vein structure index and symmetry index of each connected domain and the mean gray value of the pixels in each connected domain, the cerebral vein compensation factor of each connected domain is obtained, and the cerebral vein structure index is positively correlated with the cerebral vein compensation factor, and the symmetry index and the mean gray value are negatively correlated with the cerebral vein compensation factor; The principal component analysis is performed on the position coordinates of all pixel points in each connected domain to obtain the principal axis. In each image block, the angle between the principal axes corresponding to any two connected domains is used as the deviation factor, and the variance of all deviation factors corresponding to each connected domain is used as the eigenvalue of the cerebral venous brush. The normalized value of the product of the cerebral venous compensation factor corresponding to each connected domain and the cerebral venous brush characteristic value is used as the cerebral venous compensation index of each connected domain.

[0008] Furthermore, the method for obtaining the cerebral venous structure index includes: Analyze the shape characteristics of each connected domain to obtain the morphological change characteristic value; In each connected domain, the variance of the grayscale values ​​of all pixels is taken as the grayscale fluctuation eigenvalue; The product of the morphological change characteristic value and the grayscale fluctuation characteristic value corresponding to each connected domain is negatively correlated and normalized to obtain the value as the cerebral venous structure index of each connected domain.

[0009] Furthermore, the method for obtaining the morphological change characteristic value includes: Obtain the edge line of each connected domain and calculate the slope value of each edge pixel point, and take the average of the absolute values ​​of the slope values ​​of each edge pixel point and two adjacent edge pixel points as the shape fluctuation factor; take the average of the shape fluctuation factors between all edge pixel points on the edge line of each connected domain as the edge fluctuation characteristic value; For any edge pixel point on the edge line of each connected domain, obtain the normal line of the tangent line of the edge line at the edge pixel point, take another edge pixel point where the normal line intersects the edge line as the corresponding point of the edge pixel point, and take the average of the Euclidean distances between all edge pixels and the corresponding points as the shape feature value; The product of the shape eigenvalue of each connected domain and the edge fluctuation eigenvalue is taken as the morphological change eigenvalue.

[0010] Furthermore, the method for obtaining the collateral circulation significance value includes: In each mIP image, the Euclidean distance between the centers of any two image patches is used as the distance parameter; For any image block, other image blocks except the image block are used as reference blocks, and the formula model of the collateral circulation saliency value includes: in, Represents the collateral circulation significance value of each image block; represents the mean of the cerebral venous compensation index of all connected domains in each image block; Indicates the first The density of the vascular structure area of ​​the reference block; Represents each image block and the corresponding The distance parameter between reference blocks; Expressed as a natural constant An exponential function with base .

[0011] Furthermore, the multi-scale fitting is performed on each mIP image based on the saliency algorithm to obtain a saliency map and a final saliency value of each pixel in the saliency map, including: At all preset scales, multi-scale fitting is performed on each mIP image based on the saliency detection CA algorithm to obtain the saliency map corresponding to each mIP image and the final saliency value of each pixel in the saliency map.

[0012] Furthermore, the method for acquiring the cerebral venous collateral region of interest includes: In each saliency map, the final saliency values ​​of all pixels are arranged in descending order to obtain a sorted sequence; In the sorting sequence, the absolute value of the difference between each two adjacent final significant values ​​is calculated as the significant difference value, and the maximum value of the two final significant values ​​corresponding to the maximum significant difference value is used as the reference indicator; The pixel whose final significance value is greater than the reference index is taken as the target pixel; In each saliency map, all target pixels are morphologically processed to obtain the cerebral vein collateral region of interest.

[0013] Furthermore, the method for acquiring the image block includes: The superpixel segmentation method is used to segment each mIP image into regions to obtain several superpixel blocks, and each superpixel block is used as an image block.

[0014] The present invention has the following beneficial effects: Acquire multiple mIP images of the patient's head, perform regional segmentation on each mIP image at different preset scales, and obtain image blocks, which is helpful to capture the characteristics of the cerebral venous structure at different scales. Given that the cerebral venous structure usually presents uniformly distributed black and slender blood vessels, in each image block, the connected domain is obtained based on the grayscale information of the pixel points, and in order to improve the accuracy of subsequent local area comparison, the regional density of the vascular structure of each image block can be determined based on the shape characteristics of the connected domain and the number characteristics of the pixel points. This index can characterize the proportion of cerebral venous structure in each image block or the possibility of cerebral venous structure, which helps to quantify the distribution of vascular structure. Because cerebral venous vessels are mostly smooth line shapes and are affected by the magnetic sensitivity effect, the grayscale distribution will be relatively uniform. At the same time, when the cerebral venous collateral circulation is activated, the cerebral venous structure undergoes morphological changes, and the left-right symmetric features will be destroyed. Therefore, at any preset scale, the grayscale distribution, shape characteristics and relative position relationship between the connected domains in each mIP image are analyzed to determine the cerebral venous compensation index of each connected domain, which can evaluate the compensatory ability of the cerebral venous structure. In order to improve the labeling effect, the saliency algorithm is used in the present invention. Therefore, in each mIP image, the collateral circulation saliency value of each image block is determined based on the positional relationship between image blocks, the regional density of the vascular structure of the image block, and the cerebral venous compensation index of the connected domain in the image block. This step is conducive to quantifying the saliency of the collateral circulation and provides a basis for the subsequent labeling of the cerebral venous collateral interest region. Finally, multi-scale fitting is performed on each mIP image based on the saliency algorithm to obtain a saliency map and the final saliency value of each pixel in the saliency map, and the final saliency value of the pixel in each saliency map is analyzed to label the cerebral venous collateral interest region. In summary, the present invention combines the various change characteristics in the complex physiological process of cerebral venous collateral circulation, effectively improves the accuracy of segmenting the cerebral venous structure with collateral circulation, and further improves the efficiency of cerebral collateral circulation labeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A method flow chart of a method for rapid labeling of cerebral venous collateral circulation in a susceptibility-weighted sequence provided by one embodiment of the present invention; Figure 2 A mIP image of a patient's head provided by one embodiment of the present invention; Figure 3 A schematic diagram of a cerebral venous brush phenomenon provided by an embodiment of the present invention; Figure 4 A flow chart of a method for obtaining cerebral venous compensation indicators provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a method for rapid annotation of cerebral venous collateral circulation in a magnetic susceptibility weighted sequence proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The specific scheme of the method for rapid labeling of cerebral venous collateral circulation in a susceptibility-weighted sequence provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a method flow chart of a method for rapid annotation of cerebral venous collateral circulation in a susceptibility-weighted sequence provided by an embodiment of the present invention, the method comprising the following steps: Step S1: Acquire multiple mIP images of the patient's head.

[0021] Susceptibility Weighted Imaging (SWI) is an advanced magnetic resonance imaging technology that is highly sensitive to deoxyhemoglobin in cerebral venous blood and can clearly display the cerebral venous structure. Cerebral venous collateral circulation refers to the process in which the cerebral venous system assists blood return through alternative pathways in the case of insufficient arterial blood supply, maintains local brain tissue blood supply and oxygen supply, and allows ischemic tissue to be perfused and compensated. In clinical diagnosis, the evaluation of cerebral venous collateral circulation is crucial to understanding the pathophysiological process of cerebrovascular disease; and accurately identifying the cerebral venous collateral area of ​​interest in the image helps doctors more accurately evaluate the collateral circulation and judge the patient's condition, etc. Therefore, in this embodiment of the present invention, it is first necessary to obtain multiple MRI images of the patient's head.

[0022] Specifically, ensure that the patient is in an appropriate position, usually supine, fix the patient's head with a sponge pad to reduce motion artifacts, select an appropriate magnetic resonance imaging device (such as Discovery 750 3.0T MRI scanner), and adjust imaging parameters such as magnetic field strength, echo time, and repetition time according to clinical needs and patient conditions. For example, the imaging baseline is set to be parallel to the anterior and posterior commissure (AC-PC) line to ensure that the scan covers the entire brain; SWI sequence scanning parameters: TR28ms, TE20ms, scanning thickness 1mm, layer spacing 1mm, FOV220mm×193.6mm, matrix 352×352, flip angle 15°, and obtain SWI images. Since cerebral venous vessels are usually manifested as areas with low signal intensity, that is, low-density areas, in medical images, and in the annotation of cerebral venous collateral circulation, cerebral venous vessels and their branches need to be clearly identified, the use of minimum intensity projection images (minimum intensity projection, mIP) can effectively reduce the interference of brain tissue and other vascular structures on the display of cerebral venous vessels. Therefore, with a slice thickness of 20 mm and a slice spacing of 1 mm as the reconstruction parameters for volume imaging, a mIP image was reconstructed on each SWI image obtained, and in order to facilitate subsequent analysis, the mIP image was preprocessed: first, the mIP image was rotated to ensure that the left and right brains were in the same horizontal direction, and the mIP image in DICOM format was converted to JPG format. After the conversion, the mIP image in JPG format with the same number of slices as the SWI scan was obtained. Each image shows the complete image information of a slice and corresponds to the number of scanned slices. Please refer to Figure 2 , which shows an mIP image of a patient's head in one embodiment of the present invention.

[0023] It should be noted that the collection and acquisition of the mIP image of the patient's head in the embodiment of the present invention are all authorized by the relevant users, and the process does not violate relevant laws and regulations and does not violate public order and good customs.

[0024] Step S2: At different preset scales, each mIP image is segmented to obtain image blocks; in each image block, a connected domain is obtained based on the grayscale information of the pixels; in each image block, the vascular structure area density of each image block is determined based on the shape characteristics of the connected domain and the number characteristics of the pixels.

[0025] Saliency detection is a saliency detection method that can combine global and local information and is widely used in image segmentation, target detection and medical image analysis. In the mIP images of SWI sequences, the main characteristics of cerebral venous collateral circulation are the enhancement, dilation, morphological changes and symmetry changes of cerebral venous structures. These characteristics are in line with the core concept of saliency detection. Therefore, the cerebral venous structures that may have cerebral collateral circulation in SWI images can be annotated by optimizing the saliency detection algorithm.

[0026] The saliency detection algorithm identifies and highlights the region of interest in the image by combining local information and global information. Therefore, in the embodiment of the present invention, the mIP image is segmented at different preset scales to obtain multiple image blocks. This multi-scale analysis helps to capture the characteristics of the cerebral venous structure at different spatial resolutions, thereby improving the accuracy of recognition. Since the cerebral venous structure is usually manifested as a low signal (black) area in the mIP image, the connected domain can be obtained by analyzing the grayscale information of the pixel points. Because the cerebral venous structure usually presents evenly distributed black and slender blood vessels, it is usually difficult to segment the local area of ​​the cerebral venous structure into an image block alone. Therefore, in order to improve the accuracy of subsequent local area comparison, it is necessary to calculate the regional density of the vascular structure in each image block to quantify the distribution characteristics or richness of the cerebral venous structure in the image block.

[0027] First, each mIP image is segmented at different preset scales to obtain image blocks. In this embodiment of the present invention, the preset scale is The specific size of the scale can be adjusted according to the implementation scenario and is not limited here.

[0028] Preferably, in one embodiment of the present invention, the method for acquiring an image block includes: Superpixel segmentation can divide an image into a smaller number of regions (superpixel blocks) with similar attributes, and while reducing the complexity of the image, it can retain the key features of the image, such as edges, textures, and shapes, which has an important impact on the subsequent analysis process. Therefore, in this embodiment of the present invention, the superpixel segmentation method is used to perform regional segmentation on each mIP image to obtain a number of superpixel blocks, and each superpixel block is used as an image block.

[0029] It should be noted that the number of superpixel blocks is set to 32, and the specific number can be adjusted according to the implementation scenario and is not limited here; the superpixel segmentation method is a well-known technology and the specific process will not be repeated here.

[0030] At this point, multiple image blocks of each mIP image at each preset scale can be obtained. Here, a description of the subsequent analysis process of the mIP image at a single preset scale is given to facilitate illustration and explanation.

[0031] After obtaining multiple image blocks of each mIP image at each preset scale, the connected domain is obtained based on the grayscale information of the pixels.

[0032] Preferably, in one embodiment of the present invention, the method for acquiring the connected domain includes: The Otsu threshold segmentation algorithm is an adaptive threshold determination method that can automatically divide an image into background and target parts according to the grayscale characteristics of the image. Therefore, a binary image of each image block is obtained based on the Otsu threshold segmentation algorithm. Based on the above analysis, it can be seen that the cerebral venous structure is usually manifested as a low-intensity area, that is, a black area, so in the binary image, the pixel with a grayscale value of 0 is used as the pixel of the cerebral venous structure. And in each image block, all the pixels of the cerebral venous structure are analyzed for the connected domain, so as to obtain the connected domain. The connected domain can reflect the shape, size and position of the cerebral venous structure, which is crucial for the subsequent recognition and analysis processes.

[0033] It should be noted that Otsu threshold segmentation algorithm and connected domain analysis are both well-known technologies, and the specific process will not be described here.

[0034] Since bones contain a large amount of calcium in addition to blood vessels, they will also appear as low-intensity areas in the mIP image. However, the cerebral venous structure usually appears as evenly distributed slender blood vessels with a bifurcation structure, while the skull usually does not have a bifurcation structure. Therefore, in each image block, the vascular structure area density of each image block can be determined based on the shape characteristics of the connected domain and the number characteristics of the pixels, which is used to quantify the vascular structure density characteristics in the image block.

[0035] Preferably, in one embodiment of the present invention, the method for obtaining the regional density of the vascular structure includes: Convex hull detection can extract the outer boundary of the connected domain, which can well reflect the shape characteristics of the cerebral venous structure. Therefore, in each image block, convex hull detection is performed on each connected domain. The purpose is to distinguish different connected domains according to the structural characteristics of cerebral veins and reduce the interference of bones.

[0036] Since the cerebral venous structure may have bifurcations, the number of pixels in the convex hull will increase compared to the number of pixels in the connected domain. Therefore, the ratio of the number of pixels in the convex hull to the number of pixels in the connected domain is used as the quantity weight. The larger the quantity weight, the higher the possibility of cerebral venous structure.

[0037] Then, the number of pixels in the convex hull is weightedly fused using the number weights corresponding to the connected domain to obtain the density factor corresponding to each image block. That is, the number weight of each cerebral venous structure is multiplied by the number of pixels in the corresponding convex hull to obtain the weighted number factor, and the sum of the weighted number factors of all connected domains is used as the density factor corresponding to the image block. At this time, the larger the density factor, the higher the possibility that the connected domain in the image block represents the real cerebral venous structure, and the lower the possibility of bone interference.

[0038] Finally, the value after normalization of the ratio of the density factor corresponding to each image block to the number of pixels in each image block is used as the vascular structure area density of each image block. Normalization can eliminate the influence of the image block size, so that the vascular structure area density of all image blocks is finally compared and evaluated within a unified range. At this time, the greater the vascular structure area density, the denser the distribution of cerebral veins in the image block, and the higher the importance in the subsequent significance analysis. Normalization is a technical means well known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0039] It should be noted that convex hull detection is a well-known technology, and the specific process will not be described here in detail.

[0040] Step S3: Under any preset scale, the grayscale distribution, shape characteristics and relative position relationship between the connected domains in each mIP image are analyzed to determine the cerebral venous compensation index of each connected domain.

[0041] In the saliency detection process, the traditional calculation of saliency values ​​uses the comparison of color distances and spatial distances between different image blocks as a saliency analysis process, mainly focusing on high signal (high intensity) areas, and usually not involving the morphological characteristics of the structure within the image block, which may reduce the accuracy of saliency calculation of cerebral venous structures with collateral circulation. Therefore, in the embodiment of the present invention, the calculation process of the saliency value is optimized.

[0042] The main thing is to analyze the various features in the image. In the mIP images of the SWI sequence, the cerebral veins have unique shape and structural characteristics: the cerebral veins are in the shape of smooth lines, and the surface of the blood vessels does not have an uneven structure. Compared with other low-intensity areas (such as bones), the connected domain composed of cerebral veins is relatively narrow; and in normal brain tissue, the cerebral venous blood flow pattern is stable, and the flow rate of cerebral veins is slower than that of arteries, and the magnetic sensitivity effect is more significant. Therefore, the grayscale distribution of the cerebral veins in the mIP images of the SWI sequence is relatively uniform. In order to further improve the calculation accuracy and highlight the low-intensity areas that can represent the cerebral venous collateral circulation, it is also necessary to analyze the significant value of the cerebral venous compensation characteristics that can be reflected in each connected domain. The cerebral veins mainly appear as low intensity, regular morphology and symmetrical distribution on the mIP images of the SWI sequence; when the cerebral venous collateral circulation is activated, the cerebral venous structure may undergo morphological changes. Due to the increased consumption of cerebral venous oxygen and the increase in deoxyhemoglobin (dHb) concentration, the cerebral venous area becomes darker on the mIP image (the low signal is enhanced); in addition, due to the morphological changes of the cerebral veins caused by the activation of the cerebral venous collateral circulation, the left-right symmetry of the cerebral veins in the image is destroyed; and when the cerebral venous collateral circulation compensates, the density of the cerebral veins on the ischemic side will increase, forming a brush-like pattern, which is specifically manifested as: the cerebral venous brush sign is manifested as a more dense and disordered "brush"-like appearance of small cerebral veins than normal cerebral veins (such as Figure 3 Arrow), see Figure 3 , which shows a schematic diagram of the cerebral venous brush phenomenon in one embodiment of the present invention. In summary, the cerebral venous compensation characteristics generated by each connected domain can be analyzed through the changes in cerebral venous morphology, grayscale changes, and changes in symmetry caused by cerebral venous collateral circulation to obtain cerebral venous compensation indicators.

[0043] Preferably, in one embodiment of the present invention, the method for obtaining the cerebral venous compensation index includes: See also Figure 4 , which shows a method flow chart of a method for obtaining a cerebral venous compensation index in one embodiment of the present invention, the method comprising the following steps: Step S301: In each connected domain, the grayscale fluctuation of the pixels and the shape characteristics of the connected domain are analyzed to obtain the cerebral vein structure index of each connected domain.

[0044] Analyze the shape characteristics of each connected domain and obtain the morphological change characteristic value: Based on the Canny operator, the edge line of each connected domain is obtained, and then the least squares method is used to perform curve fitting on the edge line to obtain the fitting curve and calculate the slope value at each edge pixel on the fitting curve. The average of the absolute value of the difference between the slope values ​​of each edge pixel and the two adjacent edge pixels is used as the shape fluctuation factor. The larger the shape fluctuation factor, the more tortuous the shape of the edge line of the connected domain. The average of the shape fluctuation factors between all edge pixels on the edge line of each connected domain is used as the edge fluctuation eigenvalue. The smaller the edge fluctuation eigenvalue, the smoother the edge of the connected domain is, the less uneven the structure is, and the greater the possibility of it being a cerebral vein.

[0045] It should be noted that obtaining edge lines based on the Canny operator and performing curve fitting using the least squares method are both well-known technologies, and the specific process will not be described in detail here.

[0046] And because the cerebral veins are usually slender, that is, narrow in width, for any edge pixel point on the edge line of each connected domain, the normal of the tangent of the edge line is obtained at the edge pixel point, and another edge pixel point where the normal intersects with the edge line is used as the corresponding point of the edge pixel point. At this time, the distance between the edge pixel point and the corresponding point on the edge line of each connected domain can better measure the width characteristics of the blood vessel, and the smaller the distance, the narrower the width. The mean of the Euclidean distance between all edge pixels and the corresponding point is used as the shape feature value. The smaller the shape feature value, the smaller the width of the connected domain, and the more it conforms to the characteristics of the cerebral vein. It should be noted that each edge pixel point and the corresponding point exist on the edge line of the same connected domain. When calculating the Euclidean distance, a coordinate system can be constructed to calculate the Euclidean distance based on the position coordinates between the pixel points. The specific coordinate system construction process is not repeated here.

[0047] Then the product of the shape eigenvalue and the edge fluctuation eigenvalue of each connected domain is taken as the morphological change eigenvalue. Based on the above analysis, it can be seen that the smaller the morphological change eigenvalue is, the more the connected domain conforms to the characteristics of the cerebral venous structure.

[0048] Because the grayscale distribution inside the cerebral veins is relatively uniform under the influence of the magnetic susceptibility effect, the variance of the grayscale values ​​of all pixels in each connected domain is used as the grayscale fluctuation eigenvalue. At this time, the smaller the grayscale fluctuation eigenvalue is, the more uniform the grayscale distribution of the pixels is, and the more consistent it is with the characteristics of the cerebral vein structure.

[0049] The product of the morphological change characteristic value and the grayscale fluctuation characteristic value corresponding to each connected domain is negatively correlated and normalized to achieve logical relationship correction, thereby obtaining the cerebral venous structure index of each connected domain. That is, the smaller the morphological change characteristic value of each connected domain, the smaller the grayscale fluctuation characteristic value, the larger the cerebral venous structure index, and the more obvious the characteristics of the cerebral venous structure of the connected domain, and the more likely it is to represent the area where the cerebral venous structure is located. The negative correlation mapping and normalization processing here can be used with the formula ,in, It represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0050] Step S302: In each image block of each mIP image, the symmetry relationship between the image block and the symmetric region and the grayscale information in the connected domain of the image block are analyzed, and the cerebral venous structure index of the connected domain is integrated to obtain the cerebral venous compensation factor of each connected domain.

[0051] In order to highlight the areas that can represent cerebral venous collateral circulation, it is also necessary to analyze the cerebral venous compensation characteristics that each connected domain can reflect.

[0052] When the cerebral venous collateral circulation is activated, the morphological changes of the cerebral veins will occur, so the left-right symmetry of the cerebral veins in the image will be destroyed. Therefore, the symmetry relationship in the image can be analyzed as an indicator for calculating the cerebral venous compensation factor.

[0053] Construct a coordinate system in each mIP image and obtain the axis of symmetry: take the lower left corner of the mIP image as the starting point, the horizontal right direction as the positive direction of the x-axis, and construct a coordinate system vertically upward at the midpoint of the horizontal axis of the mIP image as the positive direction of the y-axis. At this time, the y-axis can be used as the axis of symmetry.

[0054] For any image block, the area symmetrical about the symmetry axis in the mIP image is taken as the symmetric block of the image block. In the symmetric block corresponding to each image block, the connected domain analysis is performed on all the pixels of the cerebral venous structure to obtain the comparative connected domain.

[0055] Hausdorff distance is a set metric that measures the similarity between two sets. In image processing, it can evaluate the similarity between two regions or shapes and then determine the symmetry relationship between them.

[0056] Therefore, in each image block and the corresponding symmetric block, the Hausdorff distance between any connected domain and any contrast connected domain is calculated. The larger the Hausdorff distance, the lower the similarity. The value after negative correlation mapping of the minimum value of all Hausdorff distances is used as the symmetry index of all connected domains in each image block. Negative correlation mapping of the minimum value of Hausdorff distance can realize logical relationship correction. At this time, the larger the symmetry index, the higher the similarity, which is regarded as the better the symmetry, and the smaller the degree of cerebral venous compensation; conversely, the smaller the symmetry index, the lower the similarity, which is regarded as the worse the symmetry, and the greater the degree of cerebral venous compensation.

[0057] When the cerebral venous collateral circulation is activated, the cerebral venous oxygen consumption increases, so the cerebral venous area will become darker on the mIP image, so the gray value can also be used as an indicator to measure cerebral venous compensation. Therefore, according to the cerebral venous structure index, symmetry index and the gray value average of the pixels in each connected domain, the cerebral venous compensation factor of each connected domain is obtained. The formula model of the cerebral venous compensation factor includes: in, represents the cerebral venous compensation factor for each connected domain; Indicators of cerebral venous structure representing each connected domain; Represents the mean gray value of pixels in each connected domain; Represents the symmetry index of each connected domain; Indicates the preset first parameter; Represents the normalization function.

[0058] In the formula model of the cerebral venous compensation factor, based on the above analysis, it can be seen that the smaller the symmetry index is, the more the left-right symmetry of the veins in the image is destroyed, and the greater the degree of cerebral venous compensation; the smaller the gray value mean is, the more the cerebral venous collateral circulation is activated, which is also considered to be the greater the degree of cerebral venous compensation; and the larger the cerebral venous structure index is, the more obvious the characteristics of the cerebral venous structure of the connected domain are. Therefore, the cerebral venous structure index is used as a weight to adjust the prominence between other connected domains. Therefore, the cerebral venous structure index is positively correlated with the cerebral venous compensation factor, and the symmetry index and the gray value mean are negatively correlated with the cerebral venous compensation factor. Therefore, based on this logic, the above formula model is constructed to calculate the cerebral venous compensation factor. At this time, the larger the cerebral venous compensation factor is, the more significant the cerebral venous compensation characteristics of the connected domain are.

[0059] It should be noted that the first parameter is preset The function of is to prevent the denominator from being 0, and the value can be 0.001. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0060] Step S303: In each image block, the positional relationship between connected domains is analyzed to determine the cerebral vein brush feature value of each connected domain.

[0061] When the cerebral venous collateral circulation compensates, more dense and disordered "brush-like" small veins will appear. Therefore, the morphological changes caused by the cerebral venous collateral circulation can also be analyzed to obtain the cerebral venous brush characteristic value, which is used to reflect the compensation status of the cerebral veins.

[0062] The principal component analysis is performed on the position coordinates of all pixels in each connected domain to obtain the principal axis, which represents the main extension direction of the connected domain and is an important feature for analyzing the positional relationship between connected domains. In each image block, the angle between the principal axes corresponding to any two connected domains is used as the deviation factor. The larger the deviation factor, the larger the positional deviation between the two connected domains. Finally, the variance of all deviation factors corresponding to each connected domain is used as the cerebral venous brush characteristic value. At this time, the larger the cerebral venous brush characteristic value, the greater the degree of disorder in the positional distribution between the connected domains, and it can be regarded as the more significant characteristics of the cerebral venous collateral circulation compensation.

[0063] Step S304: The cerebral venous compensation factor and the cerebral venous brush characteristic value of each connected domain are integrated to obtain the cerebral venous compensation index of each connected domain.

[0064] Based on the analysis of the above steps, it can be known that when the cerebral venous compensation factor of a connected domain is larger, it means that the significance of the cerebral venous compensation characteristics of the connected domain is higher; similarly, when the cerebral venous brush characteristic value of a connected domain is larger, it means that the degree of positional distribution disorder between the connected domains is greater, and it can be regarded that the characteristics of its cerebral venous collateral circulation compensation are more significant, so the product of the cerebral venous compensation factor corresponding to each connected domain and the cerebral venous brush characteristic value is normalized and used as the cerebral venous compensation index of each connected domain. At this time, the larger the cerebral venous compensation index, the more significant the cerebral venous compensation characteristics of the connected domain, and the higher the attention. Normalization is a technical means well known to those skilled in the art, and the selection of the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0065] Step S4: In each mIP image, the collateral circulation significance value of each image block is determined based on the positional relationship between image blocks, the regional density of the vascular structure of the image block, and the cerebral venous compensation index of the connected domain in the image block; multi-scale fitting is performed on each mIP image based on the significance algorithm to obtain a significance map and the final significance value of each pixel in the significance map; the final significance value of the pixel in each significance map is analyzed to mark the cerebral venous collateral interest area.

[0066] In the saliency detection process, the traditional calculation of the saliency value is by comparing the color distance and position distance between different image blocks. However, in the embodiment of the present invention, in order to improve the subsequent cerebral venous collateral circulation labeling effect, the calculation process of the saliency value is adjusted by analyzing the positional relationship between image blocks, the vascular structure area density of the image blocks, and the cerebral venous compensation index of the connected domain in the image blocks, so as to obtain the collateral circulation saliency value of each image block.

[0067] Preferably, in one embodiment of the present invention, the method for obtaining the collateral circulation significance value includes: In each mIP image, the Euclidean distance between the centers of any two image patches is used as the distance parameter; For any image block, other image blocks except the image block are used as reference blocks, and the formula model of the collateral circulation saliency value includes: in, Represents the collateral circulation significance value of each image block; represents the mean of the cerebral venous compensation index of all connected domains in each image block; Indicates the first The density of the vascular structure area of ​​the reference block; Represents each image block and the corresponding The distance parameter between reference blocks; Expressed as a natural constant An exponential function with base .

[0068] In the formula model of the collateral circulation significance value, for each image block, the distance parameter between it and the reference block is calculated, and the weight of the reference block with higher vascular structure density when performing significance comparison with the image block is increased by using the indicator of vascular structure area density, thereby improving the contrast effect; and the contrast effect is adjusted using the mean of the cerebral venous compensation index of all connected domains in each image block, which can further amplify the significance calculation results of the venous structure with cerebral venous collateral circulation characteristics, thereby obtaining the collateral circulation significance value of each image block, and thus obtaining the collateral circulation significance value of all pixels in each image block, and the collateral circulation significance value of the pixels in each graphic block is the same.

[0069] It should be noted that the formula model for the collateral circulation significance value refers to the formula model for the significance value in the significance detection CA algorithm, so the construction process of the formula model will not be described in detail here.

[0070] At this point, based on the above process, the collateral circulation saliency value of each image block at each preset scale can be obtained. In order to enhance the contrast between salient areas and non-salient areas, multi-scale fitting can be performed on each mIP image based on the saliency algorithm to obtain a saliency map and the final saliency value of each pixel in the saliency map.

[0071] Preferably, in one embodiment of the present invention, multi-scale fitting is performed on each mIP image based on a saliency algorithm to obtain a saliency map and a final saliency value of each pixel in the saliency map, including: At all preset scales, multi-scale fitting is performed on each mIP image based on the saliency detection CA algorithm to obtain the saliency map corresponding to each mIP image and the final saliency value of each pixel in the saliency map.

[0072] It should be noted that the process of obtaining the saliency map and the final saliency value by the saliency detection CA algorithm is a well-known technology, and the specific process will not be described here.

[0073] After obtaining the saliency map, the region of interest of the cerebral venous collaterals can be determined based on the final saliency value of each pixel in the saliency map, so as to mark it.

[0074] Preferably, in one embodiment of the present invention, the method for acquiring the cerebral venous collateral region of interest includes: In each saliency map, the final saliency values ​​of all pixels are arranged in descending order to obtain a sorted sequence.

[0075] In the sorting sequence, the absolute value of the difference between each two adjacent final significant values ​​is calculated as the significant difference value. This process can identify the mutation points between the final significant values. Then the maximum value of the two final significant values ​​corresponding to the maximum significant difference value is used as the reference indicator. The reference indicator provides a threshold that can be used to distinguish between significant and insignificant pixels.

[0076] The pixels whose final saliency values ​​were greater than the reference index were taken as target pixels, and in each saliency map, all target pixels were morphologically processed to eliminate internal gaps, and the connected domain obtained after processing was taken as the cerebral venous collateral interest region.

[0077] Finally, in each mIP image on the SWI sequence, the cerebral venous collateral circulation vein structure is marked according to the marked cerebral venous collateral region of interest to complete the labeling process.

[0078] In summary, obtaining multiple mIP images of the patient's head and performing regional segmentation on each mIP image at different preset scales to obtain image blocks is helpful to capture the characteristics of the cerebral venous structure at different scales. Given that the cerebral venous structure usually presents uniformly distributed black and slender blood vessels, in each image block, the cerebral venous structure pixels are marked based on the grayscale information of the pixels and the connected domain is obtained. In addition, in order to improve the accuracy of subsequent local area comparison, the vascular structure regional density of each image block can be determined based on the shape characteristics of the connected domain and the number characteristics of the pixels. This index can characterize the proportion of cerebral venous structure in each image block or the possibility of cerebral venous structure, which helps to quantify the distribution of vascular structure. Because cerebral venous vessels are mostly smooth line shapes and are affected by the magnetic sensitivity effect, the grayscale distribution will be relatively uniform. At the same time, when the cerebral venous collateral circulation is activated, the cerebral venous structure undergoes morphological changes, and the left and right symmetric features will be destroyed. Therefore, at any preset scale, the grayscale distribution, shape characteristics and relative position relationship between the connected domains in each mIP image are analyzed to determine the cerebral venous compensation index of each connected domain, which can evaluate the compensatory ability of the cerebral venous structure. In order to improve the labeling effect, a saliency algorithm is used in the embodiment of the present invention. Therefore, in each mIP image, the collateral circulation saliency value of each image block is determined based on the positional relationship between image blocks, the regional density of the vascular structure of the image block, and the cerebral venous compensation index of the connected domain in the image block. This step is conducive to quantifying the saliency of the collateral circulation and provides a basis for the subsequent labeling of the cerebral venous collateral interest region. Finally, multi-scale fitting is performed on each mIP image based on the saliency algorithm to obtain a saliency map and the final saliency value of each pixel in the saliency map, and the final saliency value of the pixel in each saliency map is analyzed to label the cerebral venous collateral interest region. In summary, the embodiment of the present invention combines the various change characteristics in the complex physiological process of cerebral venous collateral circulation, effectively improves the accuracy of segmenting the cerebral venous structure with collateral circulation, and further improves the efficiency of cerebral collateral circulation labeling.

[0079] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences, characterized in that: The method comprises: Acquire multiple mIP images of the patient's head; At different preset scales, each mIP image is segmented to obtain image blocks; in each image block, the connected domain is obtained based on the grayscale information of the pixels; in each image block, the vascular structure area density of each image block is determined based on the shape characteristics of the connected domain and the number characteristics of the pixels; At any preset scale, the grayscale distribution, shape characteristics, and relative position relationship between connected domains in each mIP image are analyzed to determine the cerebral venous compensation index of each connected domain; In each mIP image, the collateral circulation significance value of each image block is determined based on the positional relationship between image blocks, the regional density of vascular structures in the image blocks, and the cerebral venous compensation index of the connected domain in the image blocks; multi-scale fitting is performed on each mIP image based on the significance algorithm to obtain a saliency map and the final saliency value of each pixel in the saliency map; the final saliency value of the pixel in each saliency map is analyzed to mark the cerebral venous collateral interest area.

2. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method for obtaining the connected domain includes: A binary image of each image block is obtained based on the Otsu threshold segmentation algorithm, in which pixels with a gray value of 0 are regarded as pixels of the cerebral vein structure; In each image block, the connected domain analysis is performed on all the pixels of the cerebral vein structure to obtain the connected domain.

3. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method for obtaining the regional density of the vascular structure includes: In each image block, after performing convex hull detection on each connected domain, the ratio of the number of pixels in the convex hull to the number of pixels in the connected domain is used as the quantity weight; The number of pixels in the convex hull is weightedly fused using the number weights corresponding to the connected domain to obtain the density factor corresponding to each image block; The value obtained by normalizing the ratio of the density factor corresponding to each image block to the number of pixels in each image block is used as the vascular structure area density of each image block.

4. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 2, characterized in that: The method for obtaining the cerebral venous compensation index includes: In each connected domain, the grayscale fluctuation of the pixels and the shape characteristics of the connected domain are analyzed to obtain the cerebral vein structure index of each connected domain; A coordinate system is constructed in each mIP image and the symmetry axis is obtained. For any image block, the area symmetrical about the symmetry axis in the mIP image is used as the symmetric block of the image block. In the symmetric block corresponding to each image block, the connected domain analysis is performed on all the pixel points of the cerebral venous structure to obtain the comparative connected domain. In each image block and the corresponding symmetric block, the Hausdorff distance between any connected domain and any contrast connected domain is calculated, and the value after negative correlation mapping of the minimum value of all Hausdorff distances is used as the symmetry index of all connected domains in each image block; According to the cerebral vein structure index and symmetry index of each connected domain and the mean gray value of the pixels in each connected domain, the cerebral vein compensation factor of each connected domain is obtained, and the cerebral vein structure index is positively correlated with the cerebral vein compensation factor, and the symmetry index and the mean gray value are negatively correlated with the cerebral vein compensation factor; The principal component analysis is performed on the position coordinates of all pixel points in each connected domain to obtain the principal axis. In each image block, the angle between the principal axes corresponding to any two connected domains is used as the deviation factor, and the variance of all deviation factors corresponding to each connected domain is used as the eigenvalue of the cerebral venous brush. The normalized value of the product of the cerebral venous compensation factor corresponding to each connected domain and the cerebral venous brush characteristic value is used as the cerebral venous compensation index of each connected domain.

5. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 4, characterized in that: The method for obtaining the cerebral venous structure index comprises: Analyze the shape characteristics of each connected domain to obtain the morphological change characteristic value; In each connected domain, the variance of the grayscale values ​​of all pixels is taken as the grayscale fluctuation eigenvalue; The product of the morphological change characteristic value and the grayscale fluctuation characteristic value corresponding to each connected domain is negatively correlated and normalized to obtain the value as the cerebral venous structure index of each connected domain.

6. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 5, characterized in that: The method for obtaining the morphological change characteristic value comprises: Obtain the edge line of each connected domain and calculate the slope value of each edge pixel point, and take the average of the absolute values ​​of the slope values ​​of each edge pixel point and two adjacent edge pixel points as the shape fluctuation factor; take the average of the shape fluctuation factors between all edge pixel points on the edge line of each connected domain as the edge fluctuation characteristic value; For any edge pixel point on the edge line of each connected domain, obtain the normal line of the tangent line of the edge line at the edge pixel point, take another edge pixel point where the normal line intersects the edge line as the corresponding point of the edge pixel point, and take the average of the Euclidean distances between all edge pixels and the corresponding points as the shape feature value; The product of the shape eigenvalue of each connected domain and the edge fluctuation eigenvalue is taken as the morphological change eigenvalue.

7. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method for obtaining the collateral circulation significance value comprises: In each mIP image, the Euclidean distance between the centers of any two image patches is used as the distance parameter; For any image block, other image blocks except the image block are used as reference blocks, and the formula model of the collateral circulation saliency value includes: in, Represents the collateral circulation significance value of each image block; represents the mean of the cerebral venous compensation index of all connected domains in each image block; Indicates the first The density of the vascular structure area of ​​the reference block; Represents each image block and the corresponding The distance parameter between reference blocks; Expressed as a natural constant An exponential function with base .

8. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method of performing multi-scale fitting on each mIP image based on the saliency algorithm to obtain a saliency map and a final saliency value of each pixel in the saliency map includes: At all preset scales, multi-scale fitting is performed on each mIP image based on the saliency detection CA algorithm to obtain the saliency map corresponding to each mIP image and the final saliency value of each pixel in the saliency map.

9. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method for acquiring the cerebral venous collateral region of interest comprises: In each saliency map, the final saliency values ​​of all pixels are arranged in descending order to obtain a sorted sequence; In the sorting sequence, the absolute value of the difference between each two adjacent final significant values ​​is calculated as the significant difference value, and the maximum value of the two final significant values ​​corresponding to the maximum significant difference value is used as the reference indicator; The pixel whose final significance value is greater than the reference index is taken as the target pixel; In each saliency map, all target pixels are morphologically processed to obtain the cerebral vein collateral region of interest.

10. The method for rapid labeling of cerebral venous collateral circulation in susceptibility-weighted sequences according to claim 1, characterized in that: The method for obtaining the image block includes: The superpixel segmentation method is used to segment each mIP image into regions to obtain several superpixel blocks, and each superpixel block is used as an image block.

Images