An image navigation and positioning system for hepatobiliary stone surgery

Abstract

The present invention relates to the field of hepatobiliary stone image processing, and specifically to a hepatobiliary stone surgical image navigation and positioning system. The system includes an image acquisition module; a connected domain feature analysis module, which analyzes the grayscale continuity and grayscale uniformity in all connected domains in each hepatobiliary stone image; a classifier training module, which obtains the connected domain difference index of the image according to the grayscale uniformity difference and grayscale continuity difference between connected domains in each image, and then obtains the training coefficient; based on the training coefficient, the weight is corrected to obtain the final weight coefficient; and then the final classifier model is obtained; a hepatobiliary stone positioning system, which segments the image based on the model; and navigates and positions the stone based on the segmentation result. The present invention makes the classifier segment the stone area and the artifact area more accurate and efficient, and the positioning of the stone area is more accurate.

Description

Technical Field

[0001] The present invention relates to the field of hepatobiliary stone image processing, and in particular to an image navigation and positioning system for hepatobiliary stone surgery. Background Art

[0002] The hepatobiliary is the general term for the duct system associated with the liver and gallbladder, and plays a key role in the transport of bile. Hepatobiliary stones are a common disease of the hepatobiliary system. They are caused by changes in bile composition in the gallbladder or accumulation of solid matter caused by bile oversaturation, which may lead to symptoms such as biliary colic, cholecystitis, and even bile duct obstruction. Surgery is usually required for stone removal or resection for treatment. Therefore, accurate navigation and positioning of hepatobiliary stones has become one of the important links.

[0003] In the prior art, navigation and positioning of hepatobiliary stones requires segmentation of the MRI images of the patient's liver system in advance in order to accurately identify the location of the stones. A commonly used method in the image segmentation process of hepatobiliary stones is a classifier algorithm, which obtains a more accurate classifier by weighting and repeatedly training a known data sample set, and uses the classifier to classify the pixels in the hepatobiliary stone image, thereby achieving the segmentation of the stone area and the normal area. However, due to the presence of many interference factors when acquiring hepatobiliary stone images, various artifacts are easily generated in the image. Some artifacts have many similarities with the appearance of stones in the image, resulting in the classifier requiring a very large number of iterations to adapt to the pixel data of the hepatobiliary duct, which ultimately makes it difficult for the classifier to distinguish between the artifact area and the stone area, and thus cannot accurately locate the stone area. Summary of the invention

[0004] In order to solve the technical problem that various artifacts are easily generated in the images due to the presence of many interference factors when acquiring hepatobiliary stone images, some artifacts have many similarities with stones in the image, resulting in the classifier requiring a very large number of iterations to adapt to the pixel data of the hepatobiliary duct, which ultimately makes it difficult for the classifier to distinguish between artifact areas and stone areas, and thus cannot accurately locate the stone area, the purpose of the present invention is to provide an image navigation and positioning system for hepatobiliary stone surgery, and the technical scheme adopted is as follows: a hepatobiliary stone surgery image navigation and positioning system, the system comprising: an image acquisition module for acquiring all hepatobiliary stone images; a connected domain feature analysis module for acquiring all connected domains in each hepatobiliary stone image; optionally one connected domain is used as a reference connected domain, and according to the grayscale distribution characteristics of the pixels in the reference connected domain, a connected domain is obtained. The grayscale uniformity of the reference connected domain; the grayscale continuity of the reference connected domain is obtained according to the grayscale distribution of each edge pixel in the reference connected domain in the positive direction of the gradient and the negative direction of the gradient; a classifier training module is used to select any hepatobiliary stone image as the reference hepatobiliary image; according to the difference in grayscale uniformity and grayscale continuity between any two connected domains in the reference hepatobiliary image, a connected domain difference index in the reference hepatobiliary image is obtained; according to the difference in connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images, a training coefficient of the reference hepatobiliary image is obtained; according to the training coefficient of the reference hepatobiliary image, the original weight coefficient of the classifier iterative training process is corrected to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process; based on the final weight coefficient, a final classifier model is obtained;

[0005] The hepatobiliary stone localization system is used to segment each hepatobiliary stone image based on the final classifier model; and navigate and locate the stones according to the segmentation results.

[0006] Furthermore, the grayscale uniformity acquisition method includes: acquiring the grayscale uniformity according to a grayscale uniformity calculation formula, and the grayscale uniformity calculation formula is as follows: In the formula, Indicates the gray uniformity of the reference connected domain; Represents the number of pixels in the reference connected domain; Indicates the number of pixels in the preset neighborhood of each pixel in the reference connected domain; Indicates the first The gray value of each pixel; Indicates the first The preset neighborhood of pixels Grayscale values ​​of other pixels; Represents the grayscale mean of the reference connected domain; Represents the grayscale mean value in the hepatobiliary stone image where the reference connected domain is located; represents the absolute value function; Represents an exponential function with a natural constant as base.

[0007] Furthermore, the grayscale continuity degree acquisition method includes: acquiring the grayscale continuity degree according to a grayscale continuity degree calculation formula, and the grayscale continuity degree calculation formula is as follows: In the formula, Indicates the grayscale continuity of the reference connected domain; Represents the number of edge pixels of the reference connected domain; Represents the reference connected domain The number of pixels within a preset range in the positive direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the gray value of the pixel points within the preset range in the positive direction of the edge pixel gradient; Represents the reference connected domain The number of pixels within a preset range in the opposite direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the gray value of the pixel points within the preset range in the opposite direction of the edge pixel gradient; Represents the reference connected domain The edge pixel point in the positive direction of the gradient is within the preset range The gray value of each pixel; Represents the reference connected domain The edge pixel point gradient in the reverse direction of the preset range The gray value of each pixel; Represents an exponential function with a natural constant as base.

[0008] Furthermore, the method for obtaining the connected domain difference index includes: obtaining the connected domain difference index according to a connected domain difference index calculation formula, and the connected domain difference index calculation formula is as follows: In the formula, represents the connectivity domain difference index in the reference hepatobiliary image; Indicates the maximum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the maximum grayscale continuity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale continuity of the connected domain in the reference hepatobiliary image.

[0009] Furthermore, the method for obtaining the training coefficient of the reference hepatobiliary image includes: obtaining the training coefficient according to a training coefficient calculation formula, and the training coefficient calculation formula is as follows: In the formula, represents the training coefficient of the reference hepatobiliary image; represents the connectivity domain difference index in the reference hepatobiliary image; It represents the maximum value of the difference index of the connected domain in all hepatobiliary stone images; Indicates the number of hepatobiliary stones images; Indicates that in addition to the reference hepatobiliary image Connected domain difference index of hepatobiliary stone images.

[0010] Furthermore, the obtaining of the final weight coefficient of the reference hepatobiliary image in the classification iterative training process includes: using the training coefficient to adjust the original training weight in the iterative training process, and the calculation formula is as follows: In the formula, represents the final weight coefficient of the first iteration training of the reference hepatobiliary image; Indicates the number of hepatobiliary stones images; Indicates the reference hepatobiliary image in The final weight coefficient during the training iteration; Indicates that the classifier model is The final weight coefficient during the training iteration; Indicates The basic weight correction value during the iteration training; represents the training coefficient of the reference hepatobiliary image; Indicates The final weight coefficients during the training iteration, The final weight coefficient during the first iteration of training is The second-order difference between the final weight coefficients at the training iteration.

[0011] Furthermore, the preset range is set to 10 pixels.

[0012] A method for image navigation and positioning of hepatobiliary stone surgery, the method comprising: obtaining all hepatobiliary stone images; obtaining all connected domains in each hepatobiliary stone image; selecting a connected domain as a reference connected domain, and obtaining the grayscale uniformity of the reference connected domain according to the grayscale distribution characteristics of the pixels in the reference connected domain; obtaining the grayscale continuity of the reference connected domain according to the grayscale distribution of each edge pixel in the reference connected domain in the positive direction of the gradient and the reverse direction of the gradient; selecting a hepatobiliary stone image as a reference hepatobiliary image; and obtaining the grayscale uniformity between any two connected domains in the reference hepatobiliary image according to the grayscale uniformity of the pixels in the reference connected domain. The connected domain difference index in the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the training coefficient of the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the original weight coefficient of the classifier iterative training process is corrected according to the connected domain difference index of the reference hepatobiliary image to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process; the final classifier model is obtained based on the final weight coefficient; each hepatobiliary stone image is segmented based on the final classifier model; the stones are navigated and positioned according to the segmentation results.

[0013] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned image navigation and positioning system for hepatobiliary stone surgery are implemented.

[0014] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above-mentioned image navigation and positioning system for hepatobiliary stone surgery are implemented.

[0015] The present invention has the following beneficial effects: since the grayscale distribution in the stone area and the artifact area in the hepatobiliary stone image is very different, and the edge features are very different, the grayscale uniformity of the connected domain is obtained by analyzing the grayscale distribution features in the hepatobiliary stone image, and the grayscale continuity of the connected domain is obtained according to the grayscale distribution of pixels in the positive and negative directions of the gradient of the edge pixels of the connected domain; based on the grayscale uniformity and the grayscale continuity, the connected domain difference index in each hepatobiliary stone image is obtained, and different connected domains are distinguished; since the training weights in the iterative training process have not taken into account the grayscale features of the hepatobiliary stone image itself, and the lower the connected domain difference index, the easier it is to classify during iterative training. The higher the value, the higher the training weight of the hepatobiliary stone image should be, and the more special the connected domain difference index is, the fewer images with the same properties as the hepatobiliary stone image are. In order to enhance the classification ability of the classifier for the hepatobiliary stone image, the higher the training weight should be. Therefore, according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images, the training coefficient of the reference hepatobiliary image is obtained, and the original weight coefficient of the iterative training process of the classifier is corrected by the training coefficient to obtain the final weight coefficient; the final classifier model is obtained based on the final weight coefficient; each hepatobiliary stone image is segmented based on the final classifier model; and the stone is navigated and located according to the segmentation result. The present invention takes into account the characteristics of each hepatobiliary stone image itself when the classifier is classified, so that the classifier can segment the stone area and the artifact area more accurately, with higher efficiency, and more accurately locate the stone area. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] 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.

[0017] Figure 1 A block diagram of an image navigation and positioning system for hepatobiliary stone surgery provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0018] 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, structure, features and effects of an image navigation and positioning system for hepatobiliary stone surgery 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.

[0019] 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.

[0020] The specific scheme of the image navigation and positioning system for hepatobiliary stone surgery provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] See also Figure 1 , which shows an image navigation and positioning system for hepatobiliary stone surgery provided by an embodiment of the present invention, the system includes: an image acquisition module 101, a connected domain feature analysis module 102, a classifier training module 103, and a hepatobiliary stone positioning system 104.

[0022] The image acquisition module 101 is used to acquire all hepatobiliary stone images.

[0023] The embodiment of the present invention is applicable to the scenario of locating the stone area in the hepatobiliary duct. In order to locate the hepatobiliary duct stones, it is necessary to first obtain the hepatobiliary duct stone image and perform subsequent classifier training operations through the hepatobiliary duct stone image.

[0024] In one embodiment of the present invention, an MRI (magnetic resonance imaging) image dataset of human hepatobiliary stones is obtained through channels such as medical research institutions or hospitals, and the MRI images of human hepatobiliary stones are used as hepatobiliary stone images to serve as training data input for a classifier model.

[0025] In one embodiment of the present invention, the classifier model adopts the AdaBoost model. It should be noted that in other embodiments of the present invention, the Bagging (Bootstrap Aggregating) model may also be adopted, wherein the AdaBoost model and the Bagging model are both technical means well known to those skilled in the art, and are not limited or elaborated herein.

[0026] The connected domain feature analysis module 102 is used to obtain all connected domains in each hepatobiliary stone image; select any connected domain as a reference connected domain, and obtain the grayscale uniformity of the reference connected domain based on the grayscale distribution characteristics of the pixels in the reference connected domain; and obtain the grayscale continuity of the reference connected domain based on the grayscale distribution of each edge pixel in the reference connected domain in the positive and negative directions of the gradient.

[0027] In actual situations, the stone area and the artifact area in the hepatobiliary stone image will have similarities in pixel performance, resulting in the classifier model being unable to effectively distinguish them. Therefore, it is necessary to first obtain the difference index between the stone area and the artifact area of ​​the hepatobiliary stone image analysis, so all connected domains of the hepatobiliary stone image are obtained for subsequent analysis. In actual situations, the stone connected domain usually presents a relatively uniform high-density feature, while the artifact connected domain presents a large grayscale value fluctuation due to different types, causes and locations. Therefore, in an embodiment of the present invention, the grayscale uniformity of the reference connected domain is obtained based on the grayscale distribution characteristics of the pixels in the reference connected domain, and the grayscale difference between the connected domains is analyzed by the grayscale uniformity.

[0028] Preferably, in one embodiment of the present invention, the method for obtaining grayscale uniformity includes: obtaining grayscale uniformity according to a grayscale uniformity calculation formula, and the grayscale uniformity calculation formula is as follows: In the formula, Indicates the gray uniformity of the reference connected domain; Represents the number of pixels in the reference connected domain; Indicates the number of pixels in the preset neighborhood of each pixel in the reference connected domain; Indicates the first The gray value of each pixel; Indicates the first The preset neighborhood of pixels Grayscale values ​​of other pixels; Represents the grayscale mean of the reference connected domain; Represents the grayscale mean value in the hepatobiliary stone image where the reference connected domain is located; represents the absolute value function; Represents an exponential function with a natural constant as base.

[0029] In the grayscale uniformity calculation formula, the average grayscale difference between each pixel in the connected domain and each other pixel in its preset neighborhood is used as the reference. The smaller it is, the more uniform the grayscale distribution in the preset neighborhood of each pixel in the reference connected domain is, that is, the more uniform the grayscale distribution of the reference connected domain is; and the smaller the difference between the grayscale mean in the reference connected domain and the grayscale mean of the hepatobiliary stone image where the reference connected domain is located, the more uniform the grayscale change in the reference connected domain is. At this time, the more uniform the grayscale distribution of the reference connected domain is, and the more likely the reference connected domain is to be a stone area.

[0030] In one embodiment of the present invention, each pixel is taken as the center, and the surrounding 8 pixels are used as the preset neighborhood of each pixel. It should be noted that in other embodiments of the present invention, the preset neighborhood can be set voluntarily, and is not limited here.

[0031] In reality, stones can cause pixel mutations in local areas of hepatobiliary stone images, that is, the stone connected domain will have a relatively obvious boundary, while the artifact connected domain is mostly manifested as a connected domain where the grayscale value of the pixel points gradually decays, that is, the boundary of the artifact connected domain may be a smooth transition or a gradual transition. Therefore, by analyzing the grayscale changes of the edge pixels of the connected domain in the gradient direction to subsequently obtain the difference index between different connected domains, in an embodiment of the present invention, the grayscale continuity of the reference connected domain is obtained according to the grayscale distribution of each edge pixel point in the reference connected domain in the positive and negative directions of the gradient, and the edge difference between different connected domains is analyzed by the grayscale continuity.

[0032] Preferably, in one embodiment of the present invention, the method for obtaining the grayscale continuity comprises: obtaining the grayscale continuity according to a grayscale continuity calculation formula, and the grayscale continuity calculation formula is as follows: In the formula, Indicates the grayscale continuity of the reference connected domain; Represents the number of edge pixels of the reference connected domain; Represents the reference connected domain The number of pixels within a preset range in the positive direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the gray value of the pixel points within the preset range in the positive direction of the edge pixel gradient; Represents the reference connected domain The number of pixels within a preset range in the opposite direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the gray value of the pixel points within the preset range in the opposite direction of the edge pixel gradient; Represents the reference connected domain The edge pixel point in the positive direction of the gradient is within the preset range The gray value of each pixel; Represents the reference connected domain The edge pixel point gradient in the reverse direction of the preset range The gray value of each pixel; Represents an exponential function with a natural constant as base.

[0033] Preferably, in one embodiment of the present invention, the preset range is set to 10 pixels, that is, It should be noted that, in the embodiment of the present invention, the preset range can be set voluntarily and is not limited here.

[0034] In the grayscale continuity calculation formula, the average value of the first-order difference of the grayscale value of each edge pixel in the preset range of the gradient in the positive direction of the connected domain is calculated by referring to The average value of the first-order difference of the grayscale value of pixels within the preset range in the opposite direction of the gradient The sum of the values ​​between can be used to determine whether the grayscale change around the edge pixel is continuous. The smaller the sum, the greater the grayscale continuity of the edge pixel. If the difference between the average grayscale of the edge pixel in the positive direction of the gradient and the average grayscale of the pixel in the negative direction of the gradient is The smaller it is, the smoother the grayscale change of the edge pixel is, that is, the greater the grayscale continuity of the edge pixel is. At this time, the same analysis is performed on all edge pixels to obtain , and normalize it by negative correlation mapping to obtain the grayscale continuity of the reference connected domain.

[0035] The classifier training module 103 is used to select any hepatobiliary stone image as a reference hepatobiliary image; obtain a connected domain difference index in the reference hepatobiliary image based on the difference in grayscale uniformity and grayscale continuity between any two connected domains in the reference hepatobiliary image; obtain a training coefficient of the reference hepatobiliary image based on the difference in connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; correct the original weight coefficient of the classifier iterative training process based on the training coefficient of the reference hepatobiliary image to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process; and obtain a final classifier model based on the final weight coefficient.

[0036] According to the above process, the grayscale uniformity and grayscale continuity of different connected domains can be obtained. Since there are large differences in grayscale uniformity and grayscale continuity between the stone connected domain and the artifact connected domain, in an embodiment of the present invention, the connected domain difference index in the reference hepatobiliary image is obtained based on the grayscale uniformity difference and grayscale continuity difference between any two connected domains in the reference hepatobiliary image.

[0037] Preferably, in one embodiment of the present invention, the method for obtaining the connected domain difference index includes:

[0038] The connected domain difference index is obtained according to the connected domain difference index calculation formula. The connected domain difference index calculation formula is as follows: In the formula, represents the connectivity domain difference index in the reference hepatobiliary image; Indicates the maximum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the maximum grayscale continuity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale continuity of the connected domain in the reference hepatobiliary image.

[0039] In the calculation formula of the connected domain difference index, since there are large differences in grayscale uniformity and grayscale continuity between the stone connected domain and the artifact connected domain, the maximum grayscale uniformity difference between two different connected domains is calculated. The maximum difference between the grayscale continuity Among them, the greater the maximum value of the grayscale uniformity difference and the maximum value of the grayscale continuity difference, the greater the difference between the stone connected domain and the artifact connected domain in the hepatobiliary stone image, that is, the greater the connected domain difference index in the reference hepatobiliary image.

[0040] During the iterative training of the classifier model, the grayscale features of the hepatobiliary stone image itself are often ignored in the training weight adjustment of each weak classifier. In each hepatobiliary stone image, the lower the connected domain difference index, the higher the classification difficulty during iterative training, and the higher the training weight of the hepatobiliary stone image should be. The more special the connected domain difference index, the fewer images with the same properties as the hepatobiliary stone image. In order to enhance the classification ability of the classifier for the hepatobiliary stone image, the higher its training weight should be. Therefore, in the embodiment of the present invention, the training coefficient of the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the original weight coefficient of the iterative training process of the classifier is corrected according to the training coefficient of the reference hepatobiliary image to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process.

[0041] Preferably, in one embodiment of the present invention, the method for obtaining the training coefficient of the reference hepatobiliary image includes: obtaining the training coefficient according to a training coefficient calculation formula, and the training coefficient calculation formula is as follows: In the formula, represents the training coefficient of the reference hepatobiliary image; represents the connectivity domain difference index in the reference hepatobiliary image; It represents the maximum value of the difference index of the connected domain in all hepatobiliary stone images; Indicates the number of hepatobiliary stones images; Indicates that in addition to the reference hepatobiliary image Connected domain difference index of hepatobiliary stone images.

[0042] In the training coefficient calculation formula, the closer the connected domain difference index in the reference hepatobiliary image is to the maximum connected domain difference index in all hepatobiliary stone images, that is, The closer it is to 0, the greater the difference between the stone connected domain and the artifact connected domain in the reference hepatobiliary image, and the smaller the final training weight in the iterative training process; Analyze the mean difference of the connected domain difference index between the reference hepatobiliary image and each other hepatobiliary stone image , the larger the difference mean, the more unique the connected domain in the reference hepatobiliary image, and the larger the final training weight should be during the iterative training process; and The training coefficient of the reference hepatobiliary image is obtained by multiplication to adjust the original training weight of the reference hepatobiliary image.

[0043] The training coefficient of the reference hepatobiliary image is obtained according to the above process, and the original training weight in the iterative training process is corrected based on the training coefficient to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process.

[0044] Preferably, in one embodiment of the present invention, obtaining the final weight coefficient of the reference hepatobiliary image in the classification iterative training process includes: the following shows one embodiment, specifically as follows: using the training coefficient to adjust the original training weight in the iterative training process, the calculation formula is as follows: In the formula, represents the final weight coefficient of the first iteration training of the reference hepatobiliary image; Indicates the number of hepatobiliary stones images; Indicates the reference hepatobiliary image in The final weight coefficient during the training iteration; Indicates that the classifier model is The final weight coefficient during the training iteration; Indicates The basic weight correction value during the iteration training; represents the training coefficient of the reference hepatobiliary image; Indicates The final weight coefficients during the training iteration, The final weight coefficient during the first iteration of training is The second-order difference between the final weight coefficients at the training iteration.

[0045] In the final weight coefficient calculation formula, since the original training weight of each hepatobiliary stone image in the first training process is , does not need to be corrected, and the training coefficient of the reference hepatobiliary image is taken into account, so the final weight coefficient of the reference hepatobiliary image in the first training process is During the second and third iterative training processes, since the classifier model will perform its own weight correction, and the basic weight correction value of each iterative training process can be known through existing technology, the final weight coefficients of the second and third iterative training processes can be obtained through After the fourth iteration training process, the trend of the final weight coefficients in the first three iteration training processes changed more, indicating that the final weight coefficients of the first three reference hepatobiliary images deviated more from the true value, so by adjusting the second-order difference Make adjustments to obtain .

[0046] It should be noted that the number of iterative training processes selected for the second-order difference can be set by oneself and is not limited here.

[0047] At this point, the final weight coefficient of each hepatobiliary stone image in the iterative training process is obtained.

[0048] The prior art of the AdaBoost classifier can be used to obtain a final classifier model based on the final weight coefficient of each hepatobiliary stone image in the iterative training process.

[0049] The hepatobiliary stone positioning system 104 is used to segment each hepatobiliary stone image based on the final classifier model; and navigate and locate the stone according to the segmentation result.

[0050] In one embodiment of the present invention, the hepatobiliary stone image of each patient is input into the final classifier model obtained by the algorithm, the pixels in the image are classified, and the stone area and the normal area are segmented. The stones can then be located with the help of the segmented images, or the patient's hepatobiliary system can be three-dimensionally reconstructed using multiple segmented images to complete the navigation and positioning of the hepatobiliary stones.

[0051] At this point, the navigation and positioning of hepatobiliary stones are completed.

[0052] In summary, an image acquisition module is set up to acquire all hepatobiliary stone images; a connected domain feature analysis module is used to acquire all connected domains in each hepatobiliary stone image; any connected domain is selected as a reference connected domain, and the grayscale uniformity of the reference connected domain is obtained according to the grayscale distribution characteristics of the pixels in the reference connected domain; the grayscale continuity of the reference connected domain is obtained according to the grayscale distribution of each edge pixel in the reference connected domain in the positive direction and the negative direction of the gradient; a classifier training module is used to select any hepatobiliary stone image as a reference hepatobiliary image; according to the grayscale uniformity between any two connected domains in the reference hepatobiliary image, the grayscale continuity of the reference connected domain is obtained. The connected domain difference index in the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the training coefficient of the reference hepatobiliary image is obtained according to the training coefficient of the reference hepatobiliary image; the original weight coefficient of the classifier iterative training process is corrected according to the training coefficient of the reference hepatobiliary image to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process; the final classifier model is obtained based on the final weight coefficient; the hepatobiliary stone positioning system is used to segment each hepatobiliary stone image based on the final classifier model; the stones are navigated and positioned according to the segmentation results.

[0053] The second purpose of the embodiment of the present invention is to provide a method for image navigation and positioning of hepatobiliary stone surgery, the method comprising: obtaining all hepatobiliary stone images; obtaining all connected domains in each hepatobiliary stone image; selecting any connected domain as a reference connected domain, and obtaining the grayscale uniformity of the reference connected domain according to the grayscale distribution characteristics of the pixels in the reference connected domain; obtaining the grayscale continuity of the reference connected domain according to the grayscale distribution of each edge pixel in the reference connected domain in the positive direction and the negative direction of the gradient; selecting any hepatobiliary stone image as a reference hepatobiliary image; and obtaining the grayscale continuity of the reference connected domain according to the grayscale distribution of any two connected domains in the reference hepatobiliary image. The connected domain difference index in the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the training coefficient of the reference hepatobiliary image is obtained according to the difference in the connected domain difference index between the reference hepatobiliary image and other hepatobiliary stone images; the original weight coefficient of the classifier iterative training process is corrected according to the connected domain difference index of the reference hepatobiliary image to obtain the final weight coefficient of the reference hepatobiliary image in the classification iterative training process; the final classifier model is obtained based on the final weight coefficient; each hepatobiliary stone image is segmented based on the final classifier model; the stones are navigated and located according to the segmentation results.

[0054] The third purpose of an embodiment of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned hepatobiliary stone surgery image navigation and positioning system when executing the computer program.

[0055] The fourth objective of an embodiment of the present invention is to provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned image navigation and positioning system for hepatobiliary stone surgery.

[0056] 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.

[0057] 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. An image navigation and positioning system for hepatobiliary stone surgery, characterized in that: The system comprises: An image acquisition module, used to acquire all images of hepatobiliary stones; A connected domain feature analysis module is used to obtain all connected domains in each hepatobiliary stone image; any connected domain is selected as a reference connected domain, and the grayscale uniformity of the reference connected domain is obtained based on the grayscale distribution characteristics of the pixels in the reference connected domain; and the grayscale continuity of the reference connected domain is obtained based on the grayscale distribution of each edge pixel in the reference connected domain in the positive and negative gradient directions. A classifier training module is used to select a hepatobiliary stone image as a reference hepatobiliary image; and obtain a connected domain difference index in the reference hepatobiliary image based on the grayscale uniformity difference and grayscale continuity difference between any two connected domains in the reference hepatobiliary image; The method for obtaining the training coefficient of the reference hepatobiliary image includes: The training coefficient is obtained according to the training coefficient calculation formula, which is as follows: Where, represents the training coefficient of the reference hepatobiliary image; represents the connectivity domain difference index in the reference hepatobiliary image; It represents the maximum value of the difference index of the connected domain in all hepatobiliary stone images; Indicates the number of hepatobiliary stone images; Indicates that the reference hepatobiliary image Connected domain difference index of hepatobiliary stone images; The method of obtaining the final weight coefficient of the reference hepatobiliary image during the classification iterative training process includes: The training coefficient is used to adjust the original training weights during the iterative training process. The calculation formula is as follows: Where, represents the final weight coefficient of the first iteration training of the reference hepatobiliary image; Indicates the number of hepatobiliary stone images; Indicates the reference hepatobiliary image in The final weight coefficient during the training iteration; Indicates that the classifier model is The final weight coefficient during the training iteration; Indicates the The basic weight correction value during the iteration training; represents the training coefficient of the reference hepatobiliary image; Indicates the The final weight coefficient during the iterative training, The final weight coefficient during the first iteration of training is The second-order difference between the final weight coefficients during the iterative training; Obtaining a final classifier model based on the final weight coefficients; The hepatobiliary stone localization system is used to segment each hepatobiliary stone image based on the final classifier model; and navigate and locate the stones according to the segmentation results.

2. The image navigation and positioning system for hepatobiliary stone surgery according to claim 1, characterized in that: The method for obtaining the grayscale uniformity includes: The grayscale uniformity is obtained according to the grayscale uniformity calculation formula, which is as follows: Where, Indicates the grayscale uniformity of the reference connected domain; Indicates the number of pixels in the reference connected domain; Indicates the number of pixels in the preset neighborhood of each pixel in the reference connected domain; Indicates the first Gray value of each pixel; Indicates the first The preset neighborhood of the pixel Grayscale values ​​of other pixels; Indicates the first The grayscale mean value within the preset neighborhood of pixels; Represents the grayscale mean of the reference connected domain; represents the absolute value function; Represents an exponential function with a natural constant as its base.

3. The image navigation and positioning system for hepatobiliary stone surgery according to claim 1, characterized in that: The method for obtaining the grayscale continuity degree includes: The grayscale continuity degree is obtained according to the grayscale continuity degree calculation formula, and the grayscale continuity degree calculation formula is as follows: Where, Indicates the grayscale continuity of the reference connected domain; Indicates the number of edge pixels of the reference connected domain; Represents the reference connected domain The number of pixels within a preset range in the positive direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the grayscale value of the pixel points within the preset range in the positive direction of the gradient of the edge pixel points; Represents the reference connected domain The number of pixels within a preset range in the opposite direction of the gradient of an edge pixel; Represents the reference connected domain The first order difference of the grayscale value of the pixel points within the preset range in the opposite direction of the gradient of the edge pixel points; Represents the reference connected domain The edge pixel point in the positive direction of the gradient is within the preset range Gray value of each pixel; Represents the reference connected domain The edge pixel point gradient in the opposite direction of the preset range Gray value of each pixel; Represents an exponential function with a natural constant as its base.

4. The image navigation and positioning system for hepatobiliary stone surgery according to claim 1, characterized in that: The method for obtaining the connectivity domain difference index includes: The connectivity domain difference index is obtained according to the connectivity domain difference index calculation formula. The connectivity domain difference index calculation formula is as follows: Where, represents the connectivity domain difference index in the reference hepatobiliary image; Indicates the maximum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale uniformity of the connected domain in the reference hepatobiliary image; Indicates the maximum grayscale continuity of the connected domain in the reference hepatobiliary image; Indicates the minimum grayscale continuity of the connected domain in the reference hepatobiliary image.

5. The image navigation and positioning system for hepatobiliary stone surgery according to claim 3, characterized in that: The preset range is set to 10 pixels.

6. A method for image navigation and positioning in hepatobiliary stone surgery, characterized in that: The method includes: Obtain images of all hepatobiliary stones; Obtain all connected domains in each hepatobiliary stone image; select one connected domain as a reference connected domain, and obtain the grayscale uniformity of the reference connected domain based on the grayscale distribution characteristics of the pixels in the reference connected domain; obtain the grayscale continuity of the reference connected domain based on the grayscale distribution of each edge pixel in the positive and negative gradient directions; A hepatobiliary stone image is randomly selected as a reference hepatobiliary stone image; a connected domain difference index in the reference hepatobiliary stone image is obtained based on the difference in grayscale uniformity and grayscale continuity between any two connected domains in the reference hepatobiliary stone image; a training coefficient of the reference hepatobiliary stone image is obtained based on the difference in connected domain difference index between the reference hepatobiliary stone image and other hepatobiliary stone images; an original weight coefficient of the classifier iterative training process is corrected based on the connected domain difference index of the reference hepatobiliary stone image to obtain a final weight coefficient of the reference hepatobiliary stone image in the classification iterative training process; and a final classifier model is obtained based on the final weight coefficient. Segment each hepatobiliary stone image based on the final classifier model; navigate and locate the stones based on the segmentation results; The method for obtaining the training coefficient of the reference hepatobiliary image includes: The training coefficient is obtained according to the training coefficient calculation formula, which is as follows: Where, represents the training coefficient of the reference hepatobiliary image; represents the connectivity domain difference index in the reference hepatobiliary image; It represents the maximum value of the difference index of the connected domain in all hepatobiliary stone images; Indicates the number of hepatobiliary stone images; Indicates that the reference hepatobiliary image Connected domain difference index of hepatobiliary stone images; The method of obtaining the final weight coefficient of the reference hepatobiliary image during the classification iterative training process includes: The training coefficient is used to adjust the original training weights during the iterative training process. The calculation formula is as follows: Where, represents the final weight coefficient of the first iteration training of the reference hepatobiliary image; Indicates the number of hepatobiliary stone images; Indicates the reference hepatobiliary image in The final weight coefficient during the training iteration; Indicates that the classifier model is The final weight coefficient during the training iteration; Indicates the The basic weight correction value during the iteration training; represents the training coefficient of the reference hepatobiliary image; Indicates the The final weight coefficient during the iterative training, The final weight coefficient during the first iteration of training is The second-order difference between the final weight coefficients during the training iteration.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the image navigation and positioning system for hepatobiliary stone surgery as claimed in any one of claims 1 to 5 are implemented.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the image navigation and positioning system for hepatobiliary stone surgery as described in any one of claims 1 to 5 are implemented.

Images