Computer-aided plastic surgery navigation method and system

Abstract

The present invention relates to the field of image processing technology, and more specifically to a computer-assisted plastic surgery navigation method and system, comprising: acquiring a bone image, obtaining suspected pixels based on a grayscale histogram of the bone image, constructing an edge vector based on the bone edges of the bone image to obtain all center pixels, obtaining the bone thickness and edge angle of each center pixel, calculating the bone thickness abnormality factor parameter of each suspected pixel, combining the grayscale value and the distance of each suspected pixel from the nearest bone edge to obtain all abnormal pixels, obtaining all target pixels based on the grayscale value difference between each abnormal pixel and the surrounding pixels, clustering the target pixels to obtain a number of target clusters, and obtaining suspected bone protuberance abnormal regions after segmentation. The present invention aims to solve the problem of incorrect segmentation caused by placing seed points in low-density bones, thereby improving the accuracy of bone protuberance abnormal region segmentation.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a computer-assisted plastic surgery navigation method and system. Background Art

[0002] Because most plastic surgeries involve the craniofacial bones and central nervous system, the skull is diverse and its anatomical structure is complex. The risks of craniofacial surgery are high, the treatment is difficult, and satisfactory surgical results are difficult to achieve. The rapid development and application of computer hardware and artificial intelligence technology have enabled the construction of a three-dimensional model of the plastic surgery site by using high-performance computers and software analysis algorithms to align preoperative diagnostic images with the surgical field of view based on the real structure of the human body.

[0003] When machine vision detection technology is used to accurately locate abnormal bone protrusions, the watershed algorithm is often used to segment the abnormal bone protrusions in bone protrusion CT images. The selection of the watershed seed point has a great influence on the segmentation effect of the abnormal bone surface area. The segmentation effect is best when the watershed seed point is selected inside the abnormal bone surface area. Since the abnormal bone surface area often appears as an area with lower grayscale values ​​compared to the normal area in the CT image, the watershed algorithm often selects the pixel with the smallest grayscale value in the CT image as the seed point of the watershed algorithm. However, due to the different bone density at different positions in the bone area, the grayscale value of the low-density bone will also be lower than that of the normal-density bone area. Directly selecting the pixel with the smallest grayscale value as the seed point may incorrectly place the seed point in the low-density bone, thereby affecting the accuracy of the watershed algorithm for the segmentation of the abnormal bone protrusion area, and further affecting the accuracy of the subsequent orthopedic surgery-assisted navigation technology. Summary of the Invention

[0004] The present invention provides a computer-assisted plastic surgery assisted navigation method and system to solve the problem that the existing method of selecting the pixel point with the smallest grayscale value as the seed point may mistakenly place the seed point in the low-density bone, thereby affecting the accuracy of the watershed algorithm in segmenting the abnormal bone protrusion area, and further affecting the accuracy of subsequent orthopedic surgery assisted navigation technology.

[0005] The computer-assisted plastic surgery navigation method and system of the present invention adopt the following technical solutions:

[0006] One embodiment of the present invention provides a computer-aided plastic surgery navigation method, the method comprising the following steps:

[0007] Acquire bone CT images to obtain bone images;

[0008] Obtain suspected pixel points based on the grayscale histogram of the bone image;

[0009] Constructing an edge vector based on the bone edge of the bone image to obtain all central pixel points of the bone image; constructing a thickness line based on each central pixel point of the bone image to obtain the first intersection point and the second intersection point of each central pixel point; obtaining the bone thickness and edge angle of each central pixel point based on the first intersection point and the second intersection point of each central pixel point; obtaining the bone thickness abnormality factor parameter of each suspected pixel point based on the bone thickness and edge angle of the central pixel point on the same thickness line as the suspected pixel point and the grayscale value of the suspected pixel point;

[0010] All abnormal pixels are obtained based on the bone roughness abnormality factor parameter and grayscale value of each suspected pixel, as well as the distance between each suspected pixel and the nearest bone edge;

[0011] All target pixels are obtained based on the grayscale value difference between each abnormal pixel and the surrounding pixels;

[0012] The target pixel points are clustered to obtain several target clusters; the target clusters are segmented to obtain suspected bone protrusion abnormal areas; the suspected bone protrusion abnormal areas are used to develop a surgical assisted navigation method.

[0013] Furthermore, the method of obtaining suspected pixels based on the grayscale histogram of the bone image includes the following specific steps:

[0014] The number of pixels at each gray level in the grayscale histogram of the bone image is fitted using the GMM Gaussian mixture model to obtain the grayscale distribution fitting curve of the bone image. The AMPD peak detection algorithm is used to extract the grayscale distribution fitting curve of the bone image to obtain the peaks and troughs of the bone image.

[0015] The gray level corresponding to the trough of the bone image is used as the segmentation threshold of the bone image, and all pixels in the bone image with a gray value less than the segmentation threshold are recorded as suspected pixels.

[0016] Furthermore, the step of constructing edge vectors according to the bone edges of the bone image to obtain all central pixel points of the bone image includes the following specific steps:

[0017] The Canny edge detection algorithm is used to obtain the edge curve of the bone image. All the edge curves of the bone image are input into the Hough line detection algorithm to obtain several detection lines. The two longest detection lines are selected as the edge pixel lines of the bone image. Each edge pixel line of the bone image corresponds to a bone edge. Two edge vectors are constructed based on the two edge pixel lines of the bone image. The specific construction method of the edge vector is as follows: the direction of the edge pixel line from the left side of the maxillary face to the right side of the maxillary face is recorded as the vector direction of the edge vector, and the length of the edge pixel line is recorded as the modulus value of the edge vector.

[0018] The two edge vectors are added together to obtain a superposition vector of the bone image, and all the pixel points that the superposition vector of the bone image passes through in the bone image are recorded as the central pixel points of the bone image.

[0019] Furthermore, the step of constructing a thickness line according to each central pixel of the bone image to obtain a first intersection point and a second intersection point of each central pixel includes the following specific steps:

[0020] A straight line in the normal direction of the superimposed vector of the bone image is recorded as a thickness line. After passing through each central pixel point, the thickness line intersects with the two bone edges of the bone image at two points, which are recorded as the first intersection point and the second intersection point of each central pixel point.

[0021] Furthermore, the step of obtaining the bone thickness and edge angle of each central pixel point according to the first intersection point and the second intersection point of each central pixel point includes the following specific steps:

[0022] The distance between the first intersection point and the second intersection point of each central pixel point is recorded as the bone thickness of the central pixel point; the angle between the gradient vectors of the first intersection point and the second intersection point of each central pixel point is recorded as the edge angle of each central pixel point.

[0023] Furthermore, the bone thickness abnormality factor parameter of each suspected pixel is obtained according to the bone thickness and edge angle of the central pixel on the same thickness line as the suspected pixel, and the grayscale value of the suspected pixel. The corresponding specific formula is as follows:

[0024]

[0025] Where, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Indicates the edge angle of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the bone thickness of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; represents a logarithmic function with a natural constant as the base; is the linear normalization function, is the absolute value function.

[0026] Furthermore, the specific steps of obtaining all abnormal pixels according to the bone thickness abnormality factor parameter and grayscale value of each suspected pixel and the distance of each suspected pixel from the nearest bone edge are as follows:

[0027] Obtaining an abnormality degree value of each suspected pixel in the bone image according to a bone coarseness abnormality factor parameter of each suspected pixel in the bone image and a grayscale value of each suspected pixel;

[0028] The calculation method of the abnormality value of the kth suspected pixel in the bone image is:

[0029]

[0030] Where, Indicates the abnormality value of the kth suspected pixel in the bone image, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the distance between the kth suspected pixel in the bone image and the nearest bone edge; is the linear normalization function, is the absolute value function, is an exponential function with a natural constant as its base;

[0031] A suspected threshold is preset, and when the abnormality value of the k-th suspected pixel point is greater than the preset suspected threshold, the k-th suspected pixel point is recorded as an abnormal pixel point.

[0032] Furthermore, the method of obtaining all target pixels based on the grayscale value difference between each abnormal pixel and the surrounding pixels includes the following specific steps:

[0033] A neighborhood range is constructed with each abnormal pixel as the center, and the target degree value of each abnormal pixel in the bone image is obtained according to the abnormal degree value of each abnormal pixel in the bone image and the difference between the grayscale value of each abnormal pixel and the surrounding pixels in the neighborhood range;

[0034] The target degree value of the nth abnormal pixel is calculated as follows:

[0035]

[0036] Where, Indicates the target degree value of the nth abnormal pixel; Indicates the gray value of the nth abnormal pixel; Represents the grayscale value of the mth pixel in the neighborhood of the nth abnormal pixel; Represents the distance between the nth abnormal pixel and the mth pixel in the neighborhood; Indicates the maximum side length of the neighborhood range; is the abnormality degree value of the nth abnormal pixel in the bone image; is the linear normalization function, is the absolute value function;

[0037] An abnormal threshold is preset. When the target degree value of the nth abnormal pixel point is greater than the abnormal threshold, the nth abnormal pixel point is recorded as the target pixel point.

[0038] Furthermore, clustering the target pixels to obtain a plurality of target clusters; segmenting the target clusters to obtain suspected bone protuberance abnormal areas includes the following specific steps:

[0039] The target pixel grayscale value and pixel position are used as input, and the DBSCAN density clustering algorithm is used on the target pixel to obtain several target clusters. The mean target degree value of all target pixels in each target cluster is calculated, and the target cluster with the highest mean target degree value is selected as the high-probability bone protuberance abnormality area.

[0040] The watershed algorithm is used to select suspected bone protrusion abnormal areas. The pixel with the lowest grayscale value in the high-probability bone protrusion abnormal area is selected as the watershed seed point. The bone image is segmented by watershed based on the selected watershed seed point, and the segmentation result area is recorded as the suspected bone protrusion abnormal area.

[0041] The present invention also proposes a computer-assisted plastic surgery assisted navigation system, which includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned computer-assisted plastic surgery assisted navigation method.

[0042] The beneficial effects of the technical solution of the present invention are: collecting bone images, obtaining suspected pixels according to the grayscale histogram of the bone image, constructing edge vectors according to the bone edges of the bone image to obtain all center pixels, constructing thickness lines to obtain the first intersection and the second intersection of each center pixel, obtaining the bone thickness and edge angle of each center pixel, conducting targeted analysis of the abnormal area based on the characteristics of the abnormal bone protuberance area, calculating the bone thickness abnormality factor parameters of each suspected pixel, combining the grayscale value and the distance of each suspected pixel from the nearest bone edge to obtain all abnormal pixels, improving the accuracy of watershed segmentation seed point selection by considering the proportional relationship between bone density and grayscale value, obtaining all target pixels according to the grayscale value difference between each abnormal pixel and the surrounding pixels, further improving the accuracy of watershed seed point selection by analyzing the positional relationship of each suspected pixel within the bone image, clustering the target pixels to obtain a number of target clustering clusters, segmenting the suspected abnormal bone protuberance area, and improving the accuracy of subsequent orthopedic surgery assisted navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 any creative work.

[0044] Figure 1 This is a flowchart of the steps of the computer-assisted plastic surgery navigation method of the present invention. DETAILED DESCRIPTION

[0045] To further illustrate the technical means and effectiveness of the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the computer-assisted plastic surgery navigation method and system proposed by the present invention, including its specific implementation, structure, features, and effectiveness. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0046] 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.

[0047] The specific scheme of the computer-assisted plastic surgery navigation method and system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0048] See also Figure 1 , which shows a flowchart of a computer-aided plastic surgery navigation method provided by one embodiment of the present invention, the method comprising the following steps:

[0049] Step S001: Acquire a bone CT image to obtain a bone image.

[0050] The purpose of this embodiment is to extract the abnormal bone protrusion area in the bone CT image, so it is necessary to first use a CT device to acquire the bone CT image.

[0051] Furthermore, since the bone protrusion is located inside the bone, and the collected bone CT image is interfered by muscles and a blank black background, it is necessary to input the obtained bone CT image into a trained neural network to obtain the bone area in the bone CT image; the neural network used in this embodiment is ResNET, and the data set for training the neural network is obtained as follows: a large number of bone CT images are collected, and the bone area is artificially marked with a bounding box in each bone CT image, and the marking result is recorded as the label of each bone CT image; all bone CT images are marked to obtain a training data set, and the neural network is trained using the training data set. The loss function used in the training process is a mean square error loss function; the specific training process is a well-known content of the neural network, and the specific training process will not be repeated in this embodiment.

[0052] Each bone region is subjected to denoising using a Gaussian filter algorithm, and the denoising result of each bone region is a bone image. It should be noted that the Gaussian filter algorithm is a well-known technology and will not be described in detail in this embodiment.

[0053] At this point, the bone image is obtained.

[0054] Step S002: Obtain suspected pixel points based on the grayscale histogram of the bone image.

[0055] It should be noted that the bone image includes abnormal bone protrusion areas with low grayscale values, normal bone areas with low bone density and low grayscale values, and normal bone areas with high bone density and high grayscale values. The grayscale values ​​of the normal bone areas with high bone density are larger than those of the abnormal bone protrusion areas and the normal bone areas with low bone density. The grayscale histogram of the bone image shows a more obvious bimodal nature, so the normal bone areas with high bone density can be screened out based on the characteristics of the grayscale histogram.

[0056] Specifically, this embodiment uses an arbitrary skeletal image as an example for analysis; the analysis method for other skeletal images is the same. The grayscale histogram of the skeletal image is obtained, and the number of pixels at each grayscale level in the grayscale histogram of the skeletal image is fitted using the GMM Gaussian mixture model to obtain a grayscale distribution fitting curve for the skeletal image. The AMPD peak detection algorithm is then used to extract the grayscale distribution fitting curve of the skeletal image to obtain the peaks and troughs of the skeletal image. It should be noted that the GMM Gaussian mixture model and AMPD peak detection algorithm described in this embodiment are both well-known technologies and will not be further elaborated in this embodiment.

[0057] Furthermore, since the grayscale histogram of the bone image is bimodal, the pixels in the smaller peak range correspond to abnormal bone protrusion areas with lower grayscale values ​​or normal areas with low bone density. Therefore, this embodiment uses the grayscale level corresponding to the trough of the bone image as the segmentation threshold of the bone image, and records all pixels in the bone image with grayscale values ​​less than the segmentation threshold as suspected pixels.

[0058] At this point, all the suspected pixels in the bone image are obtained.

[0059] Step S003: Obtain the bone roughness abnormality factor parameter of each suspected pixel point according to the grayscale value and gradient direction of each suspected pixel point in the bone image.

[0060] It should be noted that in order to further determine whether the suspected pixel position is in an abnormal bone protrusion area or a normal area with low bone density, it is necessary to analyze the pixel grayscale and bone characteristics. The bone density at different bone thickness and bone positions is different, and there is a regular proportional relationship between bone density and grayscale value, while the grayscale value of the pixel in the abnormal bone protrusion area will destroy this regular proportional relationship; the smaller the direction angle of the gradient vector of the edge pixel of the bone image, the smaller the angle formed by the straight line segment where the edge pixel of the bone image is located, and the smaller the bone thickness is in the middle part of the bone. Therefore, this embodiment obtains the bone thickness abnormality factor parameter of the suspected pixel through the relationship between the suspected pixel and the bone thickness of the position where the suspected pixel is located, and the grayscale change relationship of the bone position.

[0061] Specifically, the Canny edge detection algorithm is used to obtain the edge curve of the skeleton image, all the edge curves of the skeleton image are input into the Hough line detection algorithm to obtain several detection straight lines, and the two longest detection straight lines are selected as the edge pixel lines of the skeleton image, and the edge pixel lines of each skeleton image are corresponding to a skeleton edge. According to the two edge pixel lines of the skeleton image, two edge vectors are constructed, and the specific construction mode of the edge vector is: the direction of the edge pixel line from the left side of the jaw face to the right side of the jaw face is recorded as the vector direction of the edge vector, and the length of the edge pixel line is recorded as the modulus value of the edge vector. It should be noted that the Hough line detection described in the present embodiment is an existing known technology, and the present embodiment will not be described in detail.

[0062] Furthermore, the two edge vectors are added to obtain the superposition vector of the bone image, and all the pixel points through which the superposition vector of the bone image passes in the bone image are recorded as the central pixel points of the bone image; a straight line drawn in the normal direction of the superposition vector of the bone image is recorded as a thickness line, and the thickness line intersects with the two bone edges of the bone image at two points after passing through each central pixel point, which are respectively recorded as the first intersection point and the second intersection point of each central pixel point, and the distance between the first intersection point and the second intersection point of each central pixel point is recorded as the bone thickness at the central pixel point; the angle between the gradient vectors of the first intersection point and the second intersection point of each central pixel point is recorded as the edge angle of each central pixel point.

[0063] Furthermore, this embodiment obtains the bone thickness abnormality factor parameter of each suspected pixel point based on the bone thickness and edge angle of the central pixel point on the same thickness line as the suspected pixel point, and the grayscale value of the suspected pixel point.

[0064] Specifically, the calculation method of the bone roughness abnormality factor parameter of the kth suspected pixel point in the bone image is:

[0065]

[0066] Where, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Indicates the edge angle of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the bone thickness of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; represents a logarithmic function with a natural constant as the base; is a linear normalization function that normalizes the data value to the range [0,1]. is the absolute value function.

[0067] Take the arc tangent of the angle, Represents the product of the distance and the arc tangent value. The larger the product, the larger the edge angle of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image, and the thicker the bone at that location. It is the ratio of the bone thickness parameter corresponding to the k-th suspected pixel in the bone image to the grayscale value of the k-th suspected pixel. The closer the ratio is to 1, the more proportional the grayscale value of the k-th suspected pixel is to the bone thickness, and the smaller the probability that the k-th suspected pixel is an abnormal bone protrusion area.

[0068] Similarly, the bone thickness abnormality factor parameter of each suspected pixel point is obtained according to the relationship between each suspected pixel point in the bone image and the bone thickness at the position where the suspected pixel point is located and the grayscale change of the bone position.

[0069] At this point, the bone thickness abnormality factor parameters of each suspected pixel are obtained.

[0070] Step S004: Obtain the abnormality degree value of each suspected pixel in the bone image according to the bone roughness abnormality factor parameter of each suspected pixel in the bone image and the grayscale value of each suspected pixel; obtain the abnormal pixel in the bone image according to the abnormality degree value of each suspected pixel in the bone image.

[0071] It should be noted that because there is more tissue inside the bone than at the edge of the bone, the bone density at the edge of the bone is higher than inside the bone. When there is no bone protrusion inside the bone image, the closer the suspected pixel is to the bone edge, the higher the bone density and the larger the grayscale value. However, due to the presence of bone protrusions, the suspected pixels in the abnormal bone protrusion area disrupt this grayscale regularity. For example, the grayscale value of the suspected pixels in the abnormal bone protrusion area is darker near the bone edge. Therefore, this embodiment calculates the abnormality value of the suspected pixel based on the grayscale value of the suspected pixel and the distance from the suspected pixel to the bone edge, combined with the bone thickness abnormality factor of the suspected pixel.

[0072] Specifically, the abnormality degree value of the kth suspected pixel in the bone image is calculated as follows:

[0073]

[0074] Where, Indicates the abnormality value of the kth suspected pixel in the bone image, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the distance between the kth suspected pixel in the bone image and the nearest bone edge; is a linear normalization function that normalizes the data value to the range [0,1]. is the absolute value function, is an exponential function with a natural constant as its base.

[0075] The closer it is to 1, the less the k-th suspected pixel point satisfies the grayscale change rule that the closer it is to the bone edge, the larger the grayscale value, and the more likely it is a pixel point in the abnormal bone protrusion area; As a confidence parameter, the larger the value is, the more credible the abnormality of the obtained pixel point is.

[0076] Furthermore, a suspected threshold is preset In this embodiment, the suspected threshold is In other implementations, other values ​​may be set, which are not limited in this embodiment; when the abnormality value of the kth suspected pixel point satisfies When the kth suspected pixel point is recorded as an abnormal pixel point; when the abnormality value of the kth suspected pixel point meets When , the kth suspected pixel is not processed.

[0077] Similarly, based on the grayscale value of each suspected pixel and the distance from the suspected pixel to the bone edge, combined with the bone roughness abnormality factor of each suspected pixel, the abnormality degree value of the suspected pixel is obtained, and the abnormality degree of each suspected pixel is judged using the suspected threshold to obtain all abnormal pixels in the bone image.

[0078] At this point, the abnormality level value of each suspected pixel point is obtained, and all abnormal pixels in the bone image are obtained according to the abnormality level value of each suspected pixel point in the bone image.

[0079] Step S005: Obtain a target degree value for each abnormal pixel in the bone image based on the difference between the abnormal degree value of each abnormal pixel in the bone image and the grayscale values ​​of the surrounding pixels; and obtain a target pixel of the bone image based on the target degree value of each abnormal pixel.

[0080] It should be noted that the grayscale of non-bone protuberance abnormalities in skeletal images is more uniform than that of abnormal bone protuberances. However, because bone protuberances contain connective tissue and blood vessels, the grayscale inside bone protuberances in CT images is more chaotic and uneven than that of normal bone areas. Therefore, this embodiment determines a target degree value based on the grayscale uniformity characteristics of the abnormal pixel within its different neighborhoods, and then obtains the target pixel based on the target degree value.

[0081] Specifically, the target degree value of the nth abnormal pixel is calculated as follows:

[0082]

[0083] Where, Indicates the target degree value of the nth abnormal pixel; Indicates the gray value of the nth abnormal pixel; Represents the grayscale value of the mth pixel in the neighborhood of the nth abnormal pixel; Represents the distance between the nth abnormal pixel and the mth pixel in the neighborhood; Indicates the maximum side length of the neighborhood range. In this embodiment, the preset selection range is described as a pixel area with a radius of 5 centered on the abnormal pixel point. In other implementations, it can be set to other values ​​and is not limited in this embodiment; is the abnormality value of the nth abnormal pixel in the bone image, is a linear normalization function that normalizes the data value to Within the interval, is the absolute value function.

[0084] Will and Make the difference and take the absolute value. The larger the absolute value, the greater the difference between the pixels in the current neighborhood, and the worse the uniformity of the current neighborhood. Will The inverse is that the closer the distance between the currently traversed neighborhood pixel and the center point is, the more reliable the result obtained by taking the absolute value of the difference between the two pixel grayscales is. To represent the grayscale uniformity of the neighborhood, the grayscale uniformity features within different neighborhoods are superimposed; As a confidence parameter, the larger the value is, the more credible the target degree value of the abnormal pixel point is.

[0085] Furthermore, the abnormal threshold is preset In this embodiment, the suspected threshold is In other embodiments, other values ​​may be set, which are not limited in this embodiment; when the target degree value of the nth abnormal pixel point meets When the target degree value of the nth abnormal pixel satisfies When , the nth abnormal pixel is not processed.

[0086] Similarly, according to the difference between the grayscale value of each abnormal pixel and the grayscale value of the surrounding pixels, combined with the abnormal degree value of each abnormal pixel, the target degree value of each abnormal pixel is judged using the suspected threshold to obtain all target pixels in the bone area.

[0087] At this point, the target degree value of each abnormal pixel point is obtained, and all target pixels in the bone image are obtained according to the target degree value of each abnormal pixel point.

[0088] Step S006: cluster the target pixel points in the bone image to obtain several target clusters; obtain high-probability bone protrusion abnormality areas based on the target degree values ​​of all target pixel points in the target clusters; use the watershed algorithm on the high-probability bone protrusion abnormality areas to obtain suspected bone protrusion abnormality areas; the suspected bone protrusion abnormality areas are used to formulate appropriate surgical assisted navigation methods.

[0089] Specifically, the target pixel grayscale value and pixel position are used as input, and the DBSCAN density clustering algorithm is applied to the target pixel to obtain several target clusters. The mean target degree values ​​of all target pixels in each target cluster are calculated, and the target cluster with the highest mean target degree value is selected as the high-probability bone protuberance abnormality area. It should be noted that DBSCAN density clustering is a well-known technique and will not be described in detail in this embodiment.

[0090] Furthermore, a watershed algorithm is used to select suspected bone protuberance regions. Within the high-probability bone protuberance region, the pixel with the lowest grayscale value is selected as a watershed seed point. The bone image is then segmented using the selected watershed seed point. The resulting segmentation region is designated as the suspected bone protuberance region, which is then used to develop an appropriate surgical assisted navigation method. It should be noted that the watershed algorithm is a well-known technique and will not be further elaborated in this embodiment.

[0091] In a second aspect, the present invention also provides a computer-assisted plastic surgery assisted navigation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned computer-assisted plastic surgery assisted navigation method.

[0092] It should be noted that the The model only shows negative correlation and the output of the constraint model is in In the interval, As the input of this model, it can be replaced by other models with the same purpose in specific implementation. This embodiment is just based on The model is used as an example for description without making any specific limitations.

[0093] So far, the present invention is completed.

[0094] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A computer-aided plastic surgery navigation method, characterized in that: The method comprises the following steps: Acquire bone CT images to obtain bone images; Obtain suspected pixel points based on the grayscale histogram of the bone image; Constructing an edge vector based on the bone edge of the bone image to obtain all central pixel points of the bone image; constructing a thickness line based on each central pixel point of the bone image to obtain the first intersection point and the second intersection point of each central pixel point; obtaining the bone thickness and edge angle of each central pixel point based on the first intersection point and the second intersection point of each central pixel point; obtaining the bone thickness abnormality factor parameter of each suspected pixel point based on the bone thickness and edge angle of the central pixel point on the same thickness line as the suspected pixel point and the grayscale value of the suspected pixel point; All abnormal pixels are obtained based on the bone roughness abnormality factor parameter and grayscale value of each suspected pixel, as well as the distance between each suspected pixel and the nearest bone edge; All target pixels are obtained based on the grayscale value difference between each abnormal pixel and the surrounding pixels; Clustering the target pixels to obtain several target clusters; segmenting the target clusters to obtain suspected bone protrusion abnormal areas; the suspected bone protrusion abnormal areas are used to develop a surgical assisted navigation method; The specific steps of obtaining the suspected pixel points according to the grayscale histogram of the bone image are as follows: The number of pixels at each gray level in the grayscale histogram of the bone image is fitted using the GMM Gaussian mixture model to obtain the grayscale distribution fitting curve of the bone image. The AMPD peak detection algorithm is used to extract the grayscale distribution fitting curve of the bone image to obtain the peaks and troughs of the bone image. The gray level corresponding to the trough of the bone image is used as the segmentation threshold of the bone image, and all pixels in the bone image with a gray value less than the segmentation threshold are recorded as suspected pixels; The step of constructing edge vectors according to the bone edges of the bone image to obtain all central pixel points of the bone image includes the following specific steps: The Canny edge detection algorithm is used to obtain the edge curve of the bone image. All the edge curves of the bone image are input into the Hough line detection algorithm to obtain several detection lines. The two longest detection lines are selected as the edge pixel lines of the bone image. Each edge pixel line of the bone image corresponds to a bone edge. Two edge vectors are constructed based on the two edge pixel lines of the bone image. The specific construction method of the edge vector is as follows: the direction of the edge pixel line from the left side of the maxillary face to the right side of the maxillary face is recorded as the vector direction of the edge vector, and the length of the edge pixel line is recorded as the modulus value of the edge vector. The two edge vectors are added to obtain a superposition vector of the bone image, and all the pixel points that the superposition vector of the bone image passes through in the bone image are recorded as the central pixel points of the bone image; The step of constructing a thickness line according to each central pixel of the bone image to obtain a first intersection point and a second intersection point of each central pixel includes the following specific steps: A straight line is drawn in the normal direction of the superimposed vector of the bone image as a thickness line. The thickness line passes through each central pixel and intersects with two bone edges of the bone image at two points, which are respectively recorded as the first intersection point and the second intersection point of each central pixel. The specific steps of obtaining the bone thickness and edge angle of each central pixel point according to the first intersection point and the second intersection point of each central pixel point are as follows: The distance between the first intersection point and the second intersection point of each central pixel point is recorded as the bone thickness of the central pixel point; the angle between the gradient vectors of the first intersection point and the second intersection point of each central pixel point is recorded as the edge angle of each central pixel point.

2. The computer-aided plastic surgery navigation method according to claim 1, characterized in that: The bone thickness abnormality factor parameter of each suspected pixel is obtained according to the bone thickness and edge angle of the central pixel on the same thickness line as the suspected pixel, and the gray value of the suspected pixel. The corresponding specific formula is as follows: Where, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Indicates the edge angle of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the bone thickness of the central pixel point on the same thickness line as the k-th suspected pixel point in the bone image; represents a logarithmic function with a natural constant as the base; is the linear normalization function, is the absolute value function.

3. The computer-aided plastic surgery navigation method according to claim 1, characterized in that: The specific steps of obtaining all abnormal pixels according to the bone thickness abnormality factor parameter and grayscale value of each suspected pixel and the distance between each suspected pixel and the nearest bone edge are as follows: Obtaining an abnormality degree value of each suspected pixel in the bone image according to a bone coarseness abnormality factor parameter of each suspected pixel in the bone image and a grayscale value of each suspected pixel; The calculation method of the abnormality value of the kth suspected pixel in the bone image is: Where, Indicates the abnormality value of the kth suspected pixel in the bone image, Represents the bone roughness abnormality factor parameter of the kth suspected pixel in the bone image; Represents the gray value of the kth suspected pixel in the bone image; Indicates the distance between the kth suspected pixel in the bone image and the nearest bone edge; is the linear normalization function, is the absolute value function, is an exponential function with a natural constant as its base; A suspected threshold is preset, and when the abnormality value of the k-th suspected pixel point is greater than the preset suspected threshold, the k-th suspected pixel point is recorded as an abnormal pixel point.

4. The computer-aided plastic surgery navigation method according to claim 3, characterized in that: The specific steps of obtaining all target pixels based on the grayscale value difference between each abnormal pixel and the surrounding pixels are as follows: A neighborhood range is constructed with each abnormal pixel as the center, and the target degree value of each abnormal pixel in the bone image is obtained according to the abnormal degree value of each abnormal pixel in the bone image and the difference between the grayscale value of each abnormal pixel and the surrounding pixels in the neighborhood range; The target degree value of the nth abnormal pixel point is calculated as follows: Where, Indicates the target degree value of the nth abnormal pixel; Indicates the gray value of the nth abnormal pixel; Represents the grayscale value of the mth pixel in the neighborhood of the nth abnormal pixel; Represents the distance between the nth abnormal pixel and the mth pixel in the neighborhood; Indicates the maximum side length of the neighborhood range; is the abnormality degree value of the nth abnormal pixel in the bone image; is the linear normalization function, is the absolute value function; An abnormal threshold is preset. When the target degree value of the nth abnormal pixel point is greater than the abnormal threshold, the nth abnormal pixel point is recorded as the target pixel point.

5. The computer-aided plastic surgery navigation method according to claim 1, characterized in that: The target pixel points are clustered to obtain a plurality of target clusters; and the target clusters are segmented to obtain suspected bone protuberance abnormal areas, including the following specific steps: The target pixel grayscale value and pixel position are used as input, and the DBSCAN density clustering algorithm is used on the target pixel to obtain several target clusters. The mean target degree value of all target pixels in each target cluster is calculated, and the target cluster with the highest mean target degree value is selected as the high-probability bone protuberance abnormality area. The watershed algorithm is used to select suspected bone protrusion abnormal areas. The pixel with the lowest grayscale value in the high-probability bone protrusion abnormal area is selected as the watershed seed point. The bone image is segmented by watershed based on the selected watershed seed point, and the segmentation result area is recorded as the suspected bone protrusion abnormal area.

6. A computer-aided plastic surgery navigation system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the computer-assisted plastic surgery navigation method according to any one of claims 1 to 5 are implemented.

Images