A method for processing image data of a knee osteotomy surgery

Abstract

The present application relates to the medical image processing technical field, specifically to a kind of knee arthrotomy surgery image data processing method.The method includes: obtaining the highlight area and bone tissue area in the gray image of knee arthrotomy surgery image;The distribution of highlight area and bone tissue area is analyzed to obtain the degree of dependence of highlight area;Obtain the edge gradient intensity index of highlight area;The gradient direction angle of edge pixel point is obtained, according to the edge gradient intensity index of a highlight area, the gradient direction angle of each edge pixel point and the pixel point gray value to obtain the reliability index of the highlight area;Further, the reliability of highlight area is obtained, and the calcification point area is obtained by screening;Finally, according to the reliability of calcification point area, the initial color gaussian function parameter is corrected, and the calcification point area is filtered based on the initial color gaussian function parameter of the corrected using bilateral filtering.The present application can more accurately filter knee arthrotomy surgery image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical image processing, in particular to a knee osteotomy surgery image data processing method. BACKGROUND

[0002] Knee osteotomy surgery is a knee-preserving surgery for correcting abnormal knee force line, delaying knee degeneration or replacing joint replacement. Knee osteotomy surgery image has important clinical significance in the postoperative stage. According to the bone healing progress and structural stability shown in the postoperative image, the doctor can scientifically formulate or adjust the patient's rehabilitation plan, control the weight-bearing rhythm and avoid secondary injury caused by too fast or too slow recovery. Therefore, the knee osteotomy surgery image is not only an important basis for efficacy evaluation, but also an important link to ensure rehabilitation safety and improve treatment quality.

[0003] After knee osteotomy, bone tissue needs to regrow and heal, and stably fuse with the implant or osteotomy surface. In this process, bone cells actively proliferate and begin to deposit calcium salt in the fracture or osteotomy area to form calcification points. Calcification is an early manifestation of osteogenesis, reflecting the local structure tending to be stable, which is of great significance for evaluating postoperative recovery progress and formulating rehabilitation plans. However, in order to enhance the readability of postoperative images, it is often necessary to perform denoising processing on CT images through filters (usually bilateral filters, which can remove image noise while preserving edge information, which is very important for identifying key anatomical features such as bone structure, fissure and screw plate edge in knee osteotomy surgery images). However, in knee CT scanning, due to the significant difference in tissue absorption and scattering when X-rays penetrate different density tissues (such as high-density cortical bone, lower-density cancellous bone and soft tissue), local radiation intensity may fluctuate, which may cause abnormal concentration of gray values in the image reconstruction process, especially at the junction of cortical bone and soft tissue, forming a cluster-like structural noise. Due to the physical properties of calcified substances, the density of these substances is much higher than that of the surrounding soft tissue or tissue in the bone marrow cavity, so the absorption rate of X-rays is very high, and the image usually shows a local high gray area with high brightness. At the same time, the above-mentioned structural noise also appears as a local bright spot in the image, both of which have characteristics such as blurred boundary and spatial isolation, which may cause the brightness of calcification points to be easily "diluted" or "blended" with the gray values of the surrounding pixels during denoising by the filter, resulting in a decrease or even disappearance of contrast, which is mistaken for noise and smoothed out, ultimately preventing the doctor from accurately monitoring the postoperative effect and affecting the formulation of the rehabilitation plan. SUMMARY

[0004] To solve the above technical problems, the purpose of the present application is to provide a knee osteotomy surgery image data processing method, and the technical solution adopted is as follows:

[0005] One embodiment of the present application provides a knee osteotomy surgery image data processing method, which comprises the following steps:

[0006] gray-scale the knee osteotomy surgery image to obtain a gray-scale image, acquire a highlight region and a bone tissue region in the gray-scale image, and set initial color Gaussian function parameters of bilateral filtering;

[0007] acquire a degree of attachment of a highlight region according to a minimum distance between a centroid of the highlight region and each edge pixel point of the bone tissue region and an overlapping region area of the highlight region and the bone tissue region;

[0008] acquire an edge gradient intensity index of the highlight region according to gradient vectors in a horizontal direction and a vertical direction of each edge pixel point of the highlight region;

[0009] acquire a gradient direction angle of the edge pixel point based on the gradient vectors in the horizontal direction and the vertical direction, and acquire a credibility index of the highlight region according to the edge gradient intensity index of the highlight region, the gradient direction angle of each edge pixel point and a pixel gray value;

[0010] acquire a credibility degree of the highlight region based on the degree of attachment and the credibility index of the highlight region, screen a calcification point region from the highlight region according to the credibility degree, correct the initial color Gaussian function parameters based on the credibility degree of a calcification point region, and filter the calcification point region based on the corrected initial color Gaussian function parameters by using bilateral filtering.

[0011] Preferably, the acquiring of the highlight region and the bone tissue region in the gray-scale image comprises the following steps:

[0012] acquire three gray-scale images, binarize a first gray-scale image by using global threshold segmentation, and mark a highlight region in the first gray-scale image by using connected domain analysis and morphological processing;

[0013] determine a segmentation threshold by using Otsu, acquire an initial bone tissue region by using the segmentation threshold to perform threshold segmentation on a second gray-scale image, and acquire the bone tissue region by performing morphological processing on the initial bone tissue region;

[0014] map the highlight region on the first gray-scale image and the bone tissue region on the second gray-scale image to a third gray-scale image to obtain the highlight region and the bone tissue region in the gray-scale image.

[0015] Preferably, the acquiring of the degree of attachment of the highlight region according to the minimum distance between the centroid of the highlight region and each edge pixel point of the bone tissue region and the overlapping region area of the highlight region and the bone tissue region comprises the following steps:

[0016] The distance feature term is obtained by subtracting a normalized value of a minimum distance between a centroid of the highlight region and distances between the edge pixel points of the bone tissue region from the first preset value; the area feature term is obtained by comparing an area of an overlapping region of the highlight region and the bone tissue region with an area of the highlight region; and the adhesion degree of the highlight region is obtained by adding the distance feature term and the area feature term.

[0017] Preferably, the edge gradient intensity index of a highlight region is obtained according to gradient vectors in horizontal and vertical directions of the edge pixel points of the highlight region, and the edge gradient intensity index comprises:

[0018] The gradient vector amplitude of an edge pixel point of a highlight region is obtained according to gradient vectors in horizontal and vertical directions of the edge pixel point; and the edge gradient intensity index of the highlight region is obtained by averaging and normalizing absolute values of the gradient vector amplitudes of all the edge pixel points of the highlight region.

[0019] Preferably, the credibility index of a highlight region is obtained according to the edge gradient intensity index, the gradient direction angle of the edge pixel points and the pixel grayscale value of the highlight region, and the credibility index comprises:

[0020] The gradient direction angle cosine value and the gradient direction angle sine value of an edge pixel point of a highlight region are combined to form a two-dimensional unit vector; the gradient direction feature term is obtained by averaging and normalizing lengths of the two-dimensional unit vectors corresponding to the edge pixel points in the highlight region; the difference between a first preset value and a normalized value of an entropy of the grayscale values of all the pixel points in the highlight region is obtained and recorded as the grayscale feature term; and the credibility index of the highlight region is obtained by adding the edge gradient intensity index, the gradient direction feature term and the grayscale feature term.

[0021] Preferably, the credibility degree of a highlight region is obtained by using the adhesion degree and the credibility index of the highlight region, and the credibility degree comprises:

[0022] The adhesion degree of a highlight region is used as a base, and the credibility index of the highlight region is used as an index to obtain the credibility degree of the highlight region.

[0023] Preferably, the calcification point region is obtained by screening the highlight region according to the credibility degree, and the screening comprises:

[0024] When the credibility degree of a highlight region is greater than a judgment threshold, the highlight region is the calcification point region.

[0025] Preferably, the initial color Gaussian function parameter is corrected based on the credibility degree of a calcification point region, and the correcting comprises:

[0026] The initial color Gaussian function parameter is multiplied by a difference between a first preset value and the credibility degree of the calcification point region to obtain the corrected color Gaussian function parameter of the calcification point region.

[0027] The application has at least the following beneficial effects: the application obtains the attachment degree of the highlight region by obtaining the highlight region and the bone tissue region in the gray image of the knee arthrotomy surgery image and analyzing the distribution attachment of the highlight region and the bone tissue region; then the edge pixel points of the highlight region are analyzed to obtain the edge gradient intensity index; then the edge gradient intensity index of a highlight region, the gradient direction angle of each edge pixel point and the pixel gray value are used to obtain the reliability index of the highlight region, and the reliability degree of the highlight region is obtained by comprehensively considering the attachment degree and the reliability index, and the highlight region with higher reliability degree is considered to be more likely to be a real calcification point, that is, a calcification point region, and the initial color Gaussian function parameters of the bilateral filtering should be reduced according to the reliability degree of the calcification point region in subsequent filtering processing, and the smoothing strength is reduced to retain the real pathological information of the calcification point region, effectively prevent the calcification point from being mistaken for noise and weaken the image features, and effectively improve the reliability of the image. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0029] Figure 1 A method flowchart of a knee arthrotomy surgery image data processing method provided by the application. DETAILED DESCRIPTION

[0030] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the knee arthrotomy surgery image data processing method according to the present application are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0032] The specific scheme of the knee arthrotomy surgery image data processing method provided by the present application is specifically described below in combination with the drawings.

[0033] Embodiment:

[0034] The main application scenario of the present application is: by solving the noise reduction of the knee osteotomy surgery image, improving the accuracy of the image, and then improving the reliability of the knee osteotomy surgery image.

[0035] Please refer to Figure 1 , which shows a method flowchart of a knee osteotomy surgery image data processing method provided by an embodiment of the present application, which comprises the following steps:

[0036] Step S1, the knee osteotomy surgery image is grayed to obtain a gray image; the highlight area and the bone tissue area in the gray image are obtained; and the initial color Gaussian function parameters of bilateral filtering are set.

[0037] Generally, the knee osteotomy surgery image is obtained by CT and X-ray, and the format is generally DICOM or TIFF / JPEG / PNG format exported by the hospital. The knee osteotomy surgery image is converted to a gray image (.png / .jpg / .tiff, etc.), and gray scale normalization is performed to limit the gray value range to [0, 1], thereby obtaining the gray image of the knee osteotomy surgery image.

[0038] Further, three gray images are obtained. For the first gray image, global threshold segmentation is used to binarize the first gray image, and the highlight area P i in the first gray image is marked through connected domain analysis and morphological processing. i P bone represents the i-th highlight area in the first gray image.

[0039] For the second gray image, because the bone tissue generally has a higher gray value, the segmentation threshold can be determined by Otsu, and then the initial bone tissue area is locked through threshold segmentation, and the small holes in the initial bone tissue area are closed through morphological processing. The small noise points in the initial bone tissue area are removed through opening operation, and the complete and continuous bone tissue area M bone is obtained, wherein the bone tissue area is 1 and the rest is 0.

[0040] The highlight area on the first gray image and the bone tissue area on the second gray image are mapped to the third gray image to obtain the highlight area and the bone tissue area in the gray image, wherein the highlight area and the bone tissue area are displayed in different colors, and subsequent analysis is based on the third gray image.

[0041] Finally, according to actual requirements and business experience, the distance weight parameter σ s and the initial color Gaussian function parameter σ r of bilateral filtering are set, and the filter window is determined. The commonly used window size of medical images is generally 5*5.

[0042] In step S2, the adhesion degree of the highlighted region is obtained according to the minimum distance between the center of the highlighted region and each edge pixel point of the bone tissue region and the area of the overlapping region of the highlighted region and the bone tissue region.

[0043] The calcification point is an insoluble substance formed by abnormal deposition in the tissue during the growth and healing process of the bone tissue, which comes from the fracture repair or bone healing process, and the calcium salt deposition can only occur at the bone fracture end or the adjacent region because that is the region where the osteoblasts are active, thus the calcification point can only adhere or deposit on the bone tissue or the surrounding structure, which is determined by the physiological and pathological processes;

[0044] And due to the significant difference in tissue absorption and scattering, the similar cluster-like structural noise eventually formed does not have strong adhesion to the bone tissue in the spatial position, and more shows uncertainty and randomness in the spatial position;

[0045] Therefore, based on the strong adhesion of the calcification point to the bone tissue, first, the center position of any region is obtained by finding the center of the extracted multiple highlighted regions, and then the shortest Euclidean distance between the center of the highlighted region and the edge of the bone tissue region is calculated to reflect the spatial distance between the current highlighted region and the bone tissue region;

[0046] At the same time, due to the irregular shape of the highlighted region, even if the edge is close to the bone tissue, the center of the highlighted region may be far away, or the center of the bright spot is far away, but the edge of the bright spot has actually overlapped with the bone tissue region, so in order to further describe the real dependence degree of the highlighted region and the bone tissue, the proportion of the area of the overlapping region of the highlighted region and the bone tissue region to the area of the highlighted region is calculated to make up for the limitations of the single distance index, and the adhesion degree of the highlighted region and the bone tissue region is obtained.

[0047] The adhesion degree of the highlighted region is obtained according to the minimum distance between the center of the highlighted region and each edge pixel point of the bone tissue region and the area of the overlapping region of the highlighted region and the bone tissue region. Specifically, the distance feature item is obtained by subtracting the normalized value of the minimum distance between the center of the highlighted region and each edge pixel point of the bone tissue region from the first preset value; the area feature item is obtained by comparing the area of the overlapping region of the highlighted region and the bone tissue region with the area of the highlighted region; and the adhesion degree of the highlighted region is obtained by adding the distance feature item and the area feature item.

[0048] The specific calculation model of the adhesion degree is:

[0049]

[0050] wherein, The attachment degree of the i-th highlight region represents its attachment degree to the bone tissue region, Norm represents normalization, and the result is limited to (0, 1). The greater the attachment degree result, the closer the i-th highlight region center position is to the bone tissue, and more spatial overlap occurs, and it is more likely that the calcification is actually attached to the edge or inside of the bone tissue;

[0051] d min The minimum distance between the i-th highlight region centroid and the edge pixel points of the bone tissue region is calculated by the Euclidean distance calculation method. The smaller the distance feature item 1-Norm(d min ) is, that is, the greater the distance feature item 1-Norm(d min ) is, the closer the i-th highlight region is to the bone tissue region, and 1 is a first preset value;

[0052] S(P i ∩M bone ) is the intersection area of the i-th highlight region and the bone tissue region, is the area of the i-th highlight region, is the area feature item, which represents the proportion of the intersection area to the highlight area. The greater the value is, the more the highlight region falls in the bone tissue region, and there is a high degree of coincidence. Thus, the attachment degree of each highlight region can be obtained.

[0053] In step S3, the edge gradient strength index of the highlight region is obtained according to the horizontal and vertical gradient vectors of the edge pixel points of the highlight region.

[0054] In the previous step, based on the strong attachment of the calcification point to the bone tissue, the minimum Euclidean distance from the centroid of the highlight region to the edge of the bone tissue and the proportion of the intersection area are comprehensively considered in the attachment degree of the highlight region. Since the structural noise in the cluster state exhibits uncertainty and randomness in space, the attachment degree to the bone tissue cannot completely determine whether the highlight region belongs to the calcification point or the cluster noise. Therefore, further analysis is needed according to the structural features of the two.

[0055] First, the calcification point is generally formed by high-density calcium salt deposition, and the tissue density is greatly different from the surrounding soft tissue, causing a sudden change in X-ray penetration rate. Therefore, the edge region gray scale changes steeply, and the pixel gray scale change gradient is large. The structural noise is mainly caused by the X-ray penetration of the high and low density tissue interface (such as cortical bone-cancellous bone, cartilage-synovial membrane) during CT reconstruction. Due to uneven tissue thickness and scattering, the gray value is abnormal. These fluctuations are not sudden, but gradual, and the edge region pixel gray scale change gradient is small.

[0056] Therefore, for the highlight region, the gradient vector of each edge pixel point in the horizontal direction and the vertical direction (X direction and Y direction) is calculated by the Sobel operator, which represents the direction in which the gray value of an edge pixel point changes most quickly from dark to light. Since the calcification point usually has a higher gray value than the surrounding tissue, the gradient change amplitude is obtained by the gradient vector, which represents the degree of change in the gray value in the outward direction of the edge of the highlight region. The gradient vector amplitude of the edge pixel point of each highlight region is calculated, and the edge gradient intensity index is obtained by summing and averaging, which is used to reflect the intensity of the gray value change of the edge of the entire highlight region.

[0057] Specifically, the gradient vector amplitude of an edge pixel point of a highlight region is obtained according to the gradient vector of the edge pixel point in the horizontal direction and the vertical direction; the edge gradient intensity index of the highlight region is obtained by averaging and normalizing the absolute values of the gradient vector amplitudes of all edge pixel points of the highlight region.

[0058] The calculation model of the edge gradient intensity index is specifically:

[0059]

[0060] wherein, is the edge gradient intensity index of the i-th highlight region, and Norm represents normalization, which limits the result to (0, 1), The greater the edge gradient intensity index, the more intense the gray change of the edge of the i-th highlight region along the normal direction, the steeper the edge, and the more likely the region is a real calcification point.

[0061] represents the number of edge pixel points of the i-th highlight region, represents the gradient vector amplitude (obtained by the Sobel operator) of the j-th edge pixel point of the i-th highlight region, G x represents the gradient vector of the edge pixel point in the horizontal direction (X direction), G y represents the gradient vector of the edge pixel point in the horizontal direction (Y direction), the amplitude of which is obtained by the gradient vectors of the edge pixel point in the X and Y directions, which is used to reflect the degree of change in the gray value in the outward direction of the edge of the point. The absolute value is accumulated and averaged to obtain the average change intensity of the edge pixel point of the highlight region, that is, the edge gradient intensity index.

[0062] Thus, the edge gradient intensity index of each highlight region can be obtained.

[0063] In step S4, the gradient direction angle of the edge pixel point is obtained based on the gradient vectors in the horizontal direction and the vertical direction; and the reliability index of the highlight region is obtained according to the edge gradient intensity index of the highlight region, the gradient direction angle of each edge pixel point, and the gray value of the pixel point.

[0064] Calcification points are usually local solid areas formed due to pathological processes, and are formed by physical changes in the structure of the tissue, so the gradient direction of the edge tends to be radially directed to the center, and the texture is relatively concentrated, while the structural noise does not have a clear structure, so the edge gradient direction lacks integrity, is distributed in disorder and divergence, and at the same time, because of such organizational properties, the gray value of the calcification point area is relatively consistent, and the entropy is low, while the structural noise has obvious gray fluctuation, and the entropy is high.

[0065] Thus, the gradient direction angle of each edge pixel point of the highlight area can be obtained according to the horizontal direction gradient and the vertical direction gradient of each edge pixel point of the highlight area, and then the cosine value and the sine value of the gradient direction angle are used to form a two-dimensional unit vector, so that the two-dimensional unit vector corresponding to each edge pixel point can be obtained (cosθ j ,sinθ j ), θ j represents the gradient direction angle of the jth edge pixel point of a highlight area, cosθ j and sinθ j represent the cosine value and the sine value of the gradient direction angle. The acquisition of the gradient direction angle is a known technology, and will not be described in detail here.

[0066] The mean value of the lengths of the two-dimensional unit vectors corresponding to the edge pixel points in a highlight area is calculated and normalized to obtain a gradient direction feature item; the difference between a first preset value and the normalized value of the entropy of the gray values of all pixel points in the highlight area is calculated, and is recorded as a gray feature item; the edge gradient intensity index, the gradient direction feature item and the gray feature item of the highlight area are added to obtain the credibility index of the highlight area.

[0067] The calculation model of the credibility index of the highlight area is specifically:

[0068]

[0069] wherein, is the credibility index of the i th highlight area, and the larger the result is, the higher the gradient direction gray value change intensity of the edge pixel points of the highlight area is, and at the same time, the direction angles of most edge pixel points are more similar, the vector directions are more consistent, the internal gray change is not intense, the structure is clear, and it is more likely to be a real calcification point;

[0070] M j represents the length of the two-dimensional unit vector corresponding to the jth edge pixel point, and the larger the value is, the more consistent the direction vectors of the edge pixel points are, while the smaller the value is, the more dispersed the direction angles are, and the more offset each other is, and Norm represents normalization, represents the number of edge pixel points of the i th highlight area, represents the gradient direction feature item;

[0071] H(P i ) is the entropy of the gray value of all pixels in the i-th highlight region. The gray value of the calcification point region is relatively consistent, and its entropy is low. The structural noise has obvious gray fluctuation, and the entropy is high. 1-Norm[H(P i )] represents the gray feature term. Thus, the reliability index of each highlight region can be obtained.

[0072] In step S5, the reliability degree of the highlight region is obtained using the attachment degree and the reliability index of the highlight region. The calcification point region is obtained by screening the highlight region according to the reliability degree. The initial color Gaussian function parameters are corrected based on the reliability degree of a calcification point region, and the calcification point region is filtered based on the corrected initial color Gaussian function parameters using bilateral filtering.

[0073] The attachment degree and the reliability index of the highlight region obtained by the above steps are determined by pathology. The calcification point is only possible to attach or deposit on the bone tissue or its surrounding structure, which is the basic threshold for judging whether the analyzed region is a calcification point. Therefore, the attachment degree is used as the base, and the reliability index is used as the index to obtain the reliability degree of the highlight region. If the highlight region also has the structural characteristics of the calcification point, it means that the highlight region may belong to the calcification point, and the reliability index is used to significantly amplify the final output result.

[0074] Specifically, the attachment degree of a highlight region is used as the base, and the reliability index of the highlight region is used as the index to obtain the reliability degree of the highlight region.

[0075] The calculation model of the reliability degree is specifically:

[0076]

[0077] wherein, represents the reliability degree of the i-th highlight region, Norm represents the normalization operation, represents the attachment degree of the i-th highlight region, represents the reliability index of the i-th highlight region. The greater the reliability degree indicates that the highlight region has strong attachment to the bone tissue, and the degree of change of the gray value in the outward direction of the edge of the highlight region is high, and the direction of the edge pixel points is relatively consistent, the internal gray change is not intense, and it is more likely to be a real calcification point.

[0078] Further, a judgment threshold is set, wherein the reference value of the judgment threshold is 0.7, and the value of the judgment threshold can be adjusted according to different scenes and actual business requirements. When the reliability of a highlighted region is greater than the judgment threshold, the highlighted region is a calcification point region.

[0079] By analyzing each highlighted region affected by the knee arthrotomy surgery, the reliability of each highlighted region is obtained. When the reliability is greater than the judgment threshold, it is considered that the current highlighted region is more likely to be a real calcification point region. When the image is denoised by bilateral filtering, the initial color Gaussian function parameter σ r of the bilateral filtering should be reduced to reduce the smoothing strength and retain the real pathological information of the calcification point region, effectively preventing the calcification point from being mistaken for noise and weakening its image features, and effectively improving the diagnostic reliability of the image.

[0080] The initial color Gaussian function parameter is corrected based on the reliability of a calcification point region. Specifically, the initial color Gaussian function parameter is multiplied by the difference between the first preset value and the reliability of the calcification point region to obtain the corrected color Gaussian function parameter of the calcification point region.

[0081] The calculation model of the corrected color Gaussian function parameter is specifically:

[0082]

[0083] wherein, σ r,a represents the corrected color Gaussian function parameter corresponding to the a th calcification point region; σ r represents the initially set global color Gaussian function parameter, that is, the initial color Gaussian function parameter; represents the reliability of the a th calcification point region.

[0084] Finally, the pixel points in each calcification point region are smoothed and filtered by bilateral filtering based on the corrected initial color Gaussian function parameter corresponding to each calcification point region, so as to retain the real pathological information of the calcification point region and improve the reliability of the knee arthrotomy surgery image.

[0085] In summary, the calcification points can only attach or deposit on the bone tissue or its surrounding structure, which is determined by physiological and pathological processes, so the attachment relationship of the highlighted area is determined by obtaining the attachment degree of the highlighted area and the bone tissue area, then since the edge gray scale of the calcification point changes steeply, the edge pixel gray scale changes gradient, and the gradient direction often points to the center in a radial manner, and the texture is more concentrated, so the attachment degree and the reliability index of the highlighted area are obtained by obtaining the change characteristics of the edge gradient of the highlighted area, the reliability degree of the highlighted area is obtained by comprehensively obtaining the attachment degree and the reliability index, the highlighted area with higher reliability degree is considered to be a real calcification point area, and in subsequent filtering processing, the initial color Gaussian function parameters of the bilateral filtering should be reduced according to the reliability degree, and the smoothing strength is reduced, so as to retain the real pathological information of the calcification point area, effectively prevent the calcification point from being mistaken for noise and weaken the image features, and effectively improve the reliability of the image.

[0086] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0087] Each embodiment in the specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0088] The above only describes the preferred embodiments of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for processing image data from knee osteotomy surgery, characterized in that, The method comprises: The knee osteotomy surgery image is grayed to obtain a gray image; a highlight region and a bone tissue region in the gray image are obtained; initial color Gaussian function parameters of bilateral filtering are set; A minimum distance in distances between a centroid of a highlight region and edge pixel points of the bone tissue region and an overlapping region area of the highlight region and the bone tissue region are obtained to obtain a degree of attachment of the highlight region; Edge gradient intensity indicators of the highlight region are obtained according to gradient vectors in horizontal and vertical directions of the edge pixel points of the highlight region; Gradient direction angles of the edge pixel points are obtained based on the gradient vectors in the horizontal and vertical directions; a credibility index of the highlight region is obtained according to the edge gradient intensity indicators, the gradient direction angles of the edge pixel points and pixel grayscale values; The degree of attachment and the credibility index of the highlight region are used to obtain a credibility of the highlight region; a calcification point region is obtained by screening the highlight region according to the credibility; the initial color Gaussian function parameters are corrected based on the credibility of the calcification point region, and the calcification point region is filtered based on the corrected initial color Gaussian function parameters by using bilateral filtering.

2. The method of claim 1, wherein the method further comprises: The highlight region and the bone tissue region in the gray image are obtained, comprising: Three gray images are obtained; a first gray image is binarized by using global threshold segmentation; a highlight region in the first gray image is marked by connected domain analysis and morphological processing; A segmentation threshold is determined by Otsu; an initial bone tissue region is obtained by threshold segmentation of a second gray image using the segmentation threshold; the bone tissue region is obtained by morphological processing of the initial bone tissue region; The highlight region on the first gray image and the bone tissue region on the second gray image are mapped to a third gray image to obtain the highlight region and the bone tissue region in the gray image.

3. The method of claim 1, wherein the method further comprises: The degree of attachment of the highlight region is obtained according to a minimum distance in distances between a centroid of the highlight region and edge pixel points of the bone tissue region and an overlapping region area of the highlight region and the bone tissue region, comprising: A distance characteristic item is obtained by subtracting a normalized value of the minimum distance in the distances between the centroid of the highlight region and the edge pixel points of the bone tissue region from a first preset value; an area characteristic item is obtained by comparing the overlapping region area of the highlight region and the bone tissue region with an area of the highlight region; and the degree of attachment of the highlight region is obtained by adding the distance characteristic item and the area characteristic item.

4. The method of claim 1, wherein the method further comprises: The edge gradient intensity indicators of the highlight region are obtained according to gradient vectors in horizontal and vertical directions of the edge pixel points of the highlight region, comprising: Gradient vector amplitudes of an edge pixel point of a highlight region are obtained according to gradient vectors in horizontal and vertical directions of the edge pixel point; the edge gradient intensity indicators of the highlight region are obtained by averaging and normalizing absolute values of the gradient vector amplitudes of all edge pixel points of the highlight region.

5. The method of claim 1, wherein the method further comprises: The credibility index of the highlight region is obtained according to the edge gradient intensity indicators, the gradient direction angles of the edge pixel points and the pixel grayscale values of the highlight region, comprising: The gradient direction cosine value and the gradient direction sine value of an edge pixel point of a highlight region are combined to form a two-dimensional unit vector; the average of the lengths of the two-dimensional unit vectors corresponding to the edge pixel points in the highlight region is calculated and normalized to obtain a gradient direction feature item; the difference between a first preset value and the normalized value of the entropy of the gray values of all the pixel points in the highlight region is calculated and recorded as a gray feature item; the edge gradient intensity index, the gradient direction feature item and the gray feature item of the highlight region are added to obtain the credibility index of the highlight region.

6. The method of claim 1, wherein the method further comprises: The credibility degree of the highlight region is obtained by using the attachment degree and the credibility index of the highlight region, and the method comprises the following steps: The attachment degree of a highlight region is taken as the base, and the credibility index of the highlight region is taken as the exponent to obtain the credibility degree of the highlight region.

7. The method of claim 1, wherein the method further comprises: determining a first angle between the first line and the second line; and determining a second angle between the first line and the third line. The calcification point region is obtained by screening the highlight regions according to the credibility degrees, and the method comprises the following steps: When the credibility degree of a highlight region is greater than a judgment threshold, the highlight region is the calcification point region.

8. The method of claim 1, wherein the method further comprises: The initial color Gaussian function parameters are corrected based on the credibility degree of a calcification point region, and the method comprises the following steps: The initial color Gaussian function parameters are multiplied by the difference between a first preset value and the credibility degree of a calcification point region to obtain the corrected color Gaussian function parameters of the calcification point region.

Images