A method for identifying deformed regions of nasal sinus images under nasal endoscopy

Abstract

The present invention relates to the field of abnormal sinus image recognition, and particularly to a method for marking deformed regions of sinus images under a nasal endoscope. The present invention segments sinus images to obtain gray surface regions, and obtains edge curves in the gray surface regions; obtains inflection points in the curves, fits inflection point curves, obtains curve difference coefficients through the shape differences between the two, and further obtains cumulative curve difference coefficients for each image; uses the same to perform clustering on all images to obtain sinus image clustering clusters and sort them; obtains specific coefficients for each cluster according to the differences in cumulative curve difference coefficients between different clusters; and further screens the clusters to obtain abnormal sinus image clusters; marks the deformed regions according to the abnormal clusters. The present invention can accurately mark all sinus deformed regions, so that relevant medical personnel can accurately judge abnormal conditions in sinus images.

Description

Technical Field

[0001] The present invention relates to the field of abnormal sinus image recognition, and particularly to a method for marking deformed regions of sinus images under a nasal endoscope. Background Art

[0002] Nasal endoscopy is a common clinical diagnostic method mainly used to examine relevant lesions in the nasal cavity and sinuses. However, traditional manual diagnostic methods have problems such as strong subjectivity, low efficiency, and easy omission of lesion regions in the clinical stage. With the development of related computer technologies, image processing technologies, especially the automatic marking method for deformed regions based on computer vision, can greatly improve the accuracy and efficiency of diagnosis. The present invention proposes an innovative image processing method that uses computer vision technology to mark deformed regions in nasal endoscopy images to assist doctors in performing relevant clinical diagnoses efficiently and accurately.

[0003] Nasal endoscopy is a common clinical diagnostic method mainly used to examine relevant lesions in the nasal cavity and sinuses. However, traditional manual diagnostic methods have problems such as strong subjectivity, low efficiency, and easy omission of lesion regions in the clinical stage, and cannot accurately mark deformed regions in sinus images, ultimately resulting in the inability of relevant medical staff to accurately judge abnormal conditions in sinus images. Summary of the Invention

[0004] In order to solve the technical problems that the traditional diagnostic method for deformed regions of the sinuses has strong subjectivity, low efficiency, and easy omission of lesion regions in the clinical stage, and cannot accurately mark deformed regions in sinus images, ultimately resulting in the inability of relevant medical staff to accurately judge abnormal conditions in sinus images, the purpose of the present invention is to provide a method for marking deformed regions of sinus images under a nasal endoscope, and the specific technical solution adopted is as follows:

[0005] A method for marking deformed regions of sinus images under a nasal endoscope, the method comprising:

[0006] Obtain a preset number of sinus images and the acquisition time nodes of each sinus image;

[0007] Segment each sinus image according to the clustering algorithm to obtain the gray area in each sinus image; perform edge detection on the gray area to obtain all the edge curves of the gray area; obtain all the inflection points of each edge curve; obtain the corresponding inflection point curve of each edge curve according to the distribution characteristics of all the inflection points of each edge curve; obtain the curve difference coefficient between each edge curve and the inflection point curve according to the shape difference between each edge curve and the inflection point curve; accumulate all the curve difference coefficients in the gray area of each sinus image to obtain the accumulated curve difference coefficient of each sinus image; use the accumulated curve difference coefficient to cluster a preset number of sinus images to obtain all the sinus image clustering clusters; sort the sinus image clustering clusters according to the acquisition time nodes of each sinus image in each sinus image clustering cluster;

[0008] Optionally select one sinus image clustering cluster as the reference clustering cluster; obtain the specific coefficient of the reference clustering cluster according to the difference in the accumulated curve difference coefficient between the sinus images in the reference clustering cluster and the sinus images in other sinus image clustering clusters; screen all the sinus image clustering clusters according to the specific coefficient to obtain the abnormal sinus image cluster; identify the deformed area of the sinus image according to the abnormal sinus image cluster.

[0009] Further, the method for obtaining the gray area includes:

[0010] Cluster each sinus image according to the pixel gray value to obtain three pixel clustering clusters;

[0011] Sort the three pixel clustering clusters in descending order according to the average pixel gray value, and take the second pixel clustering cluster as the gray clustering cluster;

[0012] Take the area where the pixels corresponding to the gray clustering cluster are located as the gray area in the sinus image.

[0013] Further, the method for obtaining the corresponding inflection point curve of each edge curve includes:

[0014] Use Newton interpolation method to perform curve fitting on the inflection points in each edge curve to obtain the corresponding inflection point curve of each edge curve, and the inflection point curve corresponds one-to-one with the starting and ending points of the edge curve.

[0015] Further, the method for obtaining the curve difference coefficient includes:

[0016] Construct a Cartesian coordinate system for the edge curve and the inflection point curve, and obtain the edge curve ordinate and inflection point curve ordinate of each abscissa;

[0017] Calculate the difference between the ordinate of the corresponding edge curve and the ordinate of the corresponding inflection point curve for each abscissa as the first difference; sum up the first differences of all abscissas to obtain the curve difference coefficient between each edge curve and the inflection point curve.

[0018] Further, according to the acquisition time nodes of each sinus image in each sinus image clustering cluster, sort the sinus image clustering clusters, including:

[0019] Count the start time node and end time node of the maximum number of adjacent time node sinus images in each sinus image clustering cluster, and take the time between the end time node and the start time node as the time stamp of each sinus image clustering cluster;

[0020] Sort each sinus image clustering cluster according to the time stamp to obtain the sequential arrangement of all sinus image clustering clusters.

[0021] Further, the method for obtaining the specificity coefficient includes:

[0022] Obtain the specificity coefficient according to the specificity coefficient calculation formula, and the specificity coefficient calculation formula is as follows:

[0023]

[0024] In the formula, represents the specificity coefficient of the reference clustering cluster; represents the number of sinus images in the reference clustering cluster; represents the cumulative curve difference coefficient of the th sinus image in the reference clustering cluster; represents the number of sinus image clustering clusters; represents the average value of the cumulative curve difference coefficients of the sinus images in other sinus image clustering clusters except the reference clustering cluster; represents the average value of the cumulative curve difference coefficients of the sinus images in the previous sinus image clustering cluster of the reference clustering cluster; represents the average value of the cumulative curve difference coefficients of the sinus images in the next sinus image clustering cluster of the reference clustering cluster; represents the maximum value of the cumulative curve difference coefficients of all sinus images in the reference clustering cluster; represents the Iverson bracket, where if the condition inside the bracket holds, the value inside the bracket is 1, and if the condition inside the bracket does not hold, the value inside the bracket is 0.

[0025] Further, the method for obtaining the abnormal sinus image cluster includes:

[0026] Take the sinus image clustering cluster with a specificity coefficient greater than a preset first threshold as the abnormal sinus image cluster.

[0027] A malformation area identification system for nasal endoscopic sinus images. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for identifying the malformation area of nasal endoscopic sinus images are implemented.

[0028] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned method for identifying the malformation area of nasal endoscopic sinus images are implemented.

[0029] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for identifying the malformation area of nasal endoscopic sinus images are implemented.

[0030] The present invention has the following beneficial effects:

[0031] The present invention first obtains sinus images and the acquisition time nodes of each sinus image, facilitating the analysis of the process of collecting images by a nasal endoscope. When the light on the head of the nasal endoscope irradiates the sinus region, there will be obvious overexposure near the light and underexposure far from the light. There will be extreme cases of overexposure and underexposure in the areas corresponding to the bright and dark surfaces. Therefore, the present invention selects the gray surface area. Since the edge curvature of the image area corresponding to the deformed area changes greatly, edge detection is performed on the gray surface area to obtain all the edge curves of the gray surface area. Since the inflection point is the position where the direction changes significantly in the sinus image, all the inflection points of each edge curve are obtained. Curve fitting is performed on the inflection points on each edge curve to obtain the inflection point curve. By comparing the shapes between the inflection point curve and the edge curve, it can be analyzed whether there is an obvious curvature change in the edge. Since the greater the difference between the inflection point curve and the actual edge curve, it indicates that there is a large curvature change in the edge pixel points on the edge curve. Therefore, the curve difference coefficient between each edge curve and the inflection point curve is analyzed. Since there are multiple edge curves and corresponding inflection point curves in the gray surface area of the sinus image, multiple curve difference coefficients can be obtained. Therefore, all the curve difference coefficients in the gray surface area of the sinus image are accumulated to obtain the accumulated curve difference coefficient of each sinus image. The accumulated curve difference coefficient is used to cluster a preset number of sinus images to obtain all the sinus image clustering clusters. Among the multiple sinus image clustering clusters, the greater the accumulated curve difference coefficient in the sinus images within the cluster, the more abnormal the sinus image clustering cluster is, and the more likely the sinus images within the cluster are obtained by shooting through the deformed area. Therefore, the specific coefficient of the sinus image clustering cluster is analyzed. The specific coefficient is used to screen out the abnormal clusters of sinus images. The deformed areas of the sinus images are marked according to the abnormal clusters of sinus images. The present invention can accurately mark all the deformed areas of the sinuses, enabling relevant medical personnel to accurately judge the abnormal conditions in the sinus images. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0033] Figure 1 It is a flowchart of a method for marking deformed areas of sinus images under a nasal endoscope provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, with reference to the accompanying drawings and preferred embodiments, a method for marking deformed regions of sinus images under a nasal endoscope, including its specific implementation manner, structure, features, and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0036] The following specifically describes the specific solution of a method for marking deformed regions of sinus images provided by the present invention with reference to the accompanying drawings.

[0037] Please refer to Figure 1 , which shows a method for marking deformed regions of sinus images provided by an embodiment of the present invention. The method includes:

[0038] Step S1: Obtain a preset number of sinus images and the acquisition time node of each sinus image.

[0039] The embodiment of the present invention is mainly applied to the scenario of marking deformed regions in sinus images. Therefore, sinus images are first obtained. In actual situations, during the process of collecting images using a nasal endoscope, a slender endoscope instrument is inserted through the patient's nasal cavity, and a high-resolution camera is used to display the local image of the sinus region in real time. The endoscope usually has an adjustable light source to provide a clear field of view through the lens. During the exploration process, the image information is transmitted to the display screen in real time. In one embodiment of the present invention, one sinus image is collected every 200 ms, and the preset number is set to 20. It should be noted that in other embodiments of the present invention, both the acquisition time node and the preset number can be set by oneself and are not limited herein.

[0040] In one embodiment of the present invention, the collected sinus images are subjected to gray-scale processing and denoising operations to obtain the final sinus images participating in the following steps. It should be noted that both the gray-scale processing and the denoising operations are technical means well-known to those skilled in the art and will not be elaborated herein.

[0041] Step S2: Segment each sinus image according to the clustering algorithm to obtain the gray area in each sinus image; perform edge detection on the gray area to obtain all the edge curves of the gray area; obtain all the inflection points of each edge curve; obtain the corresponding inflection point curve of each edge curve according to the distribution characteristics of all the inflection points of each edge curve; obtain the curve difference coefficient between each edge curve and its corresponding inflection point curve according to the shape difference between each edge curve and its corresponding inflection point curve; accumulate all the curve difference coefficients in the gray area of each sinus image to obtain the accumulated curve difference coefficient of each sinus image; use the accumulated curve difference coefficient to cluster a preset number of sinus images to obtain all the sinus image clustering clusters; sort the sinus image clustering clusters according to the acquisition time node of each sinus image in each sinus image clustering cluster.

[0042] In the actual situation, as the nasal endoscope moves continuously, the sinus area will change from bright to dark, and the originally darker area will become brighter. Moreover, when the nasal endoscope explores the sinus area, the surrounding space is very narrow. Therefore, when irradiating the sinus area with the light on the head of the nasal endoscope, there will be obvious overexposure near the light and underexposure far from the light. Therefore, the areas corresponding to the bright surface, gray surface, and dark surface are segmented from the sinus image. Among them, there will be extreme situations of overexposure and underexposure in the areas corresponding to the bright surface and dark surface. Therefore, in the embodiments of the present invention, the gray area is selected to perform the operations of the following steps. Since the gray value difference between the gray area and the bright area and the dark area is large, in the embodiments of the present invention, each sinus image is segmented according to the clustering algorithm to obtain the gray area in each sinus image.

[0043] In one embodiment of the present invention, the K-means clustering algorithm is used to cluster each sinus image. It should be noted that in other embodiments of the present invention, other clustering algorithms may also be used, such as the hierarchical clustering algorithm. Among them, the clustering algorithm is a well-known technical means for those skilled in the art and will not be limited and described herein.

[0044] Preferably, in one embodiment of the present invention, the method for obtaining the gray area includes:

[0045] Cluster each sinus image according to the gray value of the pixel points to obtain three pixel point clustering clusters; sort the three pixel point clustering clusters from large to small according to the average gray value of the pixel points, and take the second pixel point clustering cluster as the gray surface clustering cluster; take the area where the pixel points corresponding to the gray surface clustering cluster are located as the gray area in the sinus image.

[0046] Due to the abnormality in the sinus region, which is manifested as a large change in the curvature of the edge of the image region corresponding to the deformed region in the sinus image. Combining with the moderate gray-scale change in the gray surface region mentioned in the above process, and there is a convex change in this region, which can make the shadow of the convex being irradiated appear in the image and will not be covered by overexposed or underexposed regions. Therefore, in the embodiment of the present invention, edge detection is performed on the gray surface region to obtain all the edge curves of the gray surface region.

[0047] Among the edge curves, the inflection point is the position where the direction changes significantly in the sinus image, often appears on the edge of the sinus image, and is often used as a key feature in the sinus image and participates in the segmentation process within the gray surface region of the sinus image. Therefore, in the embodiment of the present invention, all the inflection points of each edge curve are obtained. It should be noted that the method for obtaining the inflection point is a well-known technical means for those skilled in the art and will not be elaborated here.

[0048] Curve fitting is performed on the inflection points on each edge curve to obtain the inflection point curve. By comparing the shapes between the inflection point curve and the edge curve, it can be analyzed whether there is an obvious curvature change in the edge. Therefore, in the embodiment of the present invention, according to the distribution characteristics of all the inflection points of each edge curve, the corresponding inflection point curve of each edge curve is obtained.

[0049] Preferably, in an embodiment of the present invention, the method for obtaining the corresponding inflection point curve of each edge curve includes:

[0050] Using Newton interpolation method to perform curve fitting on the inflection points in each edge curve to obtain the corresponding inflection point curve of each edge curve, and the inflection point curve corresponds one-to-one with the starting and ending points of the edge curve. It should be noted that in other embodiments of the present invention, other curve fitting methods such as the least squares method can also be used to obtain the inflection point curve, and the above curve fitting methods are all well-known technical means for those skilled in the art and will not be limited and elaborated here.

[0051] Since the greater the difference between the inflection point curve and the actual edge curve, it indicates that the edge pixels on the edge curve have a large curvature change or irregularity. And the curvature change can identify the contour and shape of the sinus region. The greater the curvature change, it indicates that the contour and shape of the sinus region have changed more violently, and it is more likely that there is a deformed region in the sinus region. Therefore, in the embodiment of the present invention, according to the shape difference between each edge curve and the inflection point curve, the curve difference coefficient between each edge curve and the inflection point curve is obtained.

[0052] Preferably, in an embodiment of the present invention, the method for obtaining the curve difference coefficient includes:

[0053] A Cartesian coordinate system is constructed for the edge curve and the inflection point curve. Since the above process has mentioned that the starting endpoints and the ending endpoints of the edge curve and the inflection point curve correspond one to one, the horizontal coordinates of all pixel points on the edge curve and the inflection point curve correspond one to one, and the corresponding vertical coordinates of the edge curve and the inflection point curve are obtained for each horizontal coordinate.

[0054] The difference between the ordinate of the corresponding edge curve of each horizontal coordinate and the ordinate of the corresponding inflection point curve is calculated as the first difference; the first differences of all horizontal coordinates are accumulated and summed to obtain the curve difference coefficient between each edge curve and the inflection point curve, wherein the larger the curve difference coefficient, the greater the difference between the edge curve and the corresponding inflection point curve, and the more drastic the change of the edge curve.

[0055] Since there are multiple edge curves and corresponding inflection point curves in the gray surface area of the sinus image, multiple curve difference coefficients can be obtained, so all curve difference coefficients in the gray surface area of the sinus image are accumulated to obtain the accumulated curve difference coefficient of each sinus image. The accumulated curve difference coefficient is used to judge whether the edge of each sinus image changes drastically, and participates in subsequent operation steps.

[0056] If there is a deformed area in the sinus image, the cumulative curve difference coefficient is large, and if there is no deformed area in the sinus image, the cumulative curve difference coefficient is small, and there is a large difference between the two. Therefore, in an embodiment of the present invention, a preset number of sinus images are clustered using the cumulative curve difference coefficient to obtain all sinus image clusters, wherein the degree of change of the edges within the same sinus image clusters is similar. It should be noted that in one embodiment of the present invention, a K-means clustering algorithm is used to cluster a preset number of sinus images to obtain all sinus image clusters. The K-means clustering algorithm is a technical means well known to those skilled in the art and is not limited or elaborated herein.

[0057] Since the sinus image clusters are clustered according to the cumulative curve difference coefficient, the sinus images in the same cluster may not be adjacent at the acquisition time node, so in the embodiment of the present invention, the sinus image clusters are sorted.

[0058] Preferably, in one embodiment of the present invention, the sinus image clusters are sorted according to the acquisition time node of each sinus image in each sinus image cluster, including:

[0059] Since the time - node differences for adjacent sinus images being acquired are very small, and the difference coefficients of the accumulation curves of multiple adjacent sinus images are similar, most of the sinus images in the same cluster are adjacent in terms of acquisition time - nodes. Therefore, for each cluster of sinus images, the start time - node and end time - node of the maximum number of adjacent - time - node sinus images are statistically determined, and the time between the end time - node and the start time - node is used as the time - stamp for each cluster of sinus images.

[0060] Sort each cluster of sinus images according to the time - stamp to obtain the sequential arrangement of all clusters of sinus images.

[0061] Step S3: Arbitrarily select a cluster of sinus images as the reference cluster; obtain the specificity coefficient of the reference cluster based on the difference in the accumulation - curve difference coefficients between the sinus images in the reference cluster and those in other clusters of sinus images; screen all clusters of sinus images according to the specificity coefficient to obtain the abnormal cluster of sinus images; identify the deformed area of the sinus images based on the abnormal cluster of sinus images.

[0062] Among multiple clusters of sinus images, the larger the difference coefficient of the accumulation curves within a cluster, the more abnormal the cluster of sinus images, and the more likely the sinus images within the cluster are obtained by shooting through the deformed area. Therefore, in the embodiments of the present invention, the specificity coefficient of the reference cluster is obtained based on the difference in the accumulation - curve difference coefficients between the sinus images in the reference cluster and those in other clusters of sinus images.

[0063] Preferably, in one embodiment of the present invention, the method for obtaining the specificity coefficient includes:

[0064] Obtain the specificity coefficient according to the specificity - coefficient calculation formula, and the specificity - coefficient calculation formula is as follows:

[0065]

[0066] In the formula, represents the specificity coefficient of the reference cluster; represents the number of sinus images in the reference cluster; represents the accumulation - curve difference coefficient of the m - th sinus image in the reference cluster; represents the number of clusters of sinus images; represents the average value of the accumulation - curve difference coefficients of the sinus images in the previous cluster of sinus images of the reference cluster; represents the average value of the accumulation - curve difference coefficients of the sinus images in the next cluster of sinus images of the reference cluster; Represents the maximum value of the cumulative curve difference coefficient for all sinus images in the reference clustering cluster; Represents the Iverson bracket, where if the condition inside the bracket holds, the value inside the bracket is 1, and if the condition inside the bracket does not hold, the value inside the bracket is 0.

[0067] In the specific coefficient calculation formula, the mean value of the cumulative curve difference coefficients of the sinus images in the reference clustering cluster Compared with the mean value of the cumulative curve difference coefficients in other sinus image clustering clusters Is large, indicating that the sinus images in the reference clustering cluster are more abnormal. At this time, further analysis is performed on the sinus images in the reference clustering cluster; Compared with The larger it is, the more abnormal the sinus images in the reference clustering cluster are. At this time, the specific coefficient of the reference clustering cluster is larger; the difference between the mean value of the cumulative curve difference coefficients of the two adjacent sinus image clustering clusters before and after the reference clustering cluster compared with the mean value of the cumulative curve difference coefficients of the reference clustering cluster The larger it is, the greater the change of the reference clustering cluster compared with the adjacent sinus image clustering clusters; the difference in the mean value of the cumulative curve difference coefficients of the two adjacent sinus image clustering clusters before and after the reference clustering cluster The smaller it is, the more obvious the abnormality of the sinus images in the reference clustering cluster. At this time, the specific coefficient of the reference clustering cluster is larger. At the same time, within the reference clustering cluster, if the difference in the cumulative curve difference coefficients between sinus images is smaller, it indicates that the sinus images within the reference clustering cluster are generally more inclined to be abnormal. At this time, the specific coefficient of the reference clustering cluster is larger.

[0068] Preferably, in an embodiment of the present invention, the specific coefficient is normalized, and the sinus image clustering cluster with the normalized specific coefficient greater than the preset first threshold is used as the sinus image abnormal cluster. In an embodiment of the present invention, the preset first threshold is set to 0.95. It should be noted that in other embodiments of the present invention, the preset first threshold is set by itself and is not limited herein.

[0069] After screening out the sinus image abnormal cluster, all the sinus images in the sinus image abnormal cluster are marked. Among them, all the sinus images in the sinus image abnormal cluster can be used to assist in positioning the target of the possible malformed area. Relevant personnel can further observe and evaluate the suspected malformed area based on these images to judge its clinical significance and further diagnosis and treatment needs. It should be emphasized that this process is only an auxiliary tool for doctors, and automated screening and image marking cannot replace the judgment of professional doctors. Doctors should comprehensively evaluate and diagnose the malformed area in combination with clinical experience and other diagnostic means.

[0070] Thus, the marking of the malformed area of the sinus image is completed.

[0071] In summary, obtain a preset number of sinus images and the acquisition time nodes of each sinus image; segment each sinus image according to the clustering algorithm to obtain the gray area in each sinus image; perform edge detection on the gray area to obtain all the edge curves of the gray area; obtain all the inflection points of each edge curve; obtain the corresponding inflection point curve of each edge curve according to the distribution characteristics of all the inflection points of each edge curve; obtain the curve difference coefficient between each edge curve and the inflection point curve according to the shape difference between each edge curve and the inflection point curve; accumulate all the curve difference coefficients in the gray area of each sinus image to obtain the accumulated curve difference coefficient of each sinus image; use the accumulated curve difference coefficient to cluster the preset number of sinus images to obtain all the sinus image clusters; sort the sinus image clusters according to the acquisition time nodes of each sinus image in each sinus image cluster; arbitrarily select a sinus image cluster as the reference cluster; obtain the specific coefficient of the reference cluster according to the difference in the accumulated curve difference coefficients between the sinus images in the reference cluster and the sinus images in other sinus image clusters; screen all the sinus image clusters according to the specific coefficient to obtain the abnormal sinus image cluster; and identify the deformed area of the sinus image according to the abnormal sinus image cluster.

[0072] An embodiment of the present invention provides a system for identifying a deformed area of a sinus image under a nasal endoscope, the system includes a memory, a processor, and a computer program, where the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs in the processor, it can implement the method described in steps S1-S3.

[0073] It should be noted that: the above sequence of embodiments of the present invention is only for description and does not represent the advantages or disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for identifying deformed regions in nasal endoscopy sinus images, characterized in that, The method includes: Obtaining a preset number of sinus images and the acquisition time nodes of each sinus image; Segmenting each sinus image according to a clustering algorithm to obtain the gray surface area in each sinus image; The method for obtaining the gray surface area includes: Clustering each of the sinus images according to the gray values of pixel points to obtain three pixel point clustering clusters; Sorting the three pixel point clustering clusters in descending order according to the average gray value of pixel points, and taking the second pixel point clustering cluster as the gray surface clustering cluster; Taking the area where the pixel points corresponding to the gray surface clustering cluster are located as the gray surface area in the sinus image; Performing edge detection on the gray surface area to obtain all edge curves of the gray surface area; obtaining all inflection points of each of the edge curves; obtaining the corresponding inflection point curve of each of the edge curves according to the distribution characteristics of all the inflection points of each of the edge curves; obtaining the curve difference coefficient between each of the edge curves and the corresponding inflection point curve according to the shape difference between each of the edge curves and the inflection point curve; accumulating all the curve difference coefficients in the gray surface area of each sinus image to obtain the cumulative curve difference coefficient of each sinus image; clustering the preset number of sinus images by using the cumulative curve difference coefficient to obtain all sinus image clustering clusters; sorting the sinus image clustering clusters according to the acquisition time nodes of each sinus image in each sinus image clustering cluster; Optionally selecting one sinus image clustering cluster as a reference clustering cluster; obtaining the specificity coefficient of the reference clustering cluster according to the difference in the cumulative curve difference coefficient between the sinus images in the reference clustering cluster and the sinus images in other sinus image clustering clusters; The formula for calculating the specificity coefficient is as follows: In the formula, represents the specific coefficient of the reference clustering cluster; represents the number of sinus images in the reference clustering cluster; represents the cumulative curve difference coefficient of the th sinus image in the reference clustering cluster; represents the number of sinus image clustering clusters; represents the mean value of the cumulative curve difference coefficients of the sinus images in the sinus image clustering clusters other than the reference clustering cluster; represents the mean value of the cumulative curve difference coefficients of the sinus images in the previous sinus image clustering cluster of the reference clustering cluster; represents the mean value of the cumulative curve difference coefficients of the sinus images in the next sinus image clustering cluster of the reference clustering cluster; represents the maximum value of the cumulative curve difference coefficients of all sinus images in the reference clustering cluster; [] represents the Iverson bracket, where if the condition inside the bracket holds, the value inside the bracket is 1, and if the condition inside the bracket does not hold, the value inside the bracket is 0; Screening all sinus image clustering clusters according to the specificity coefficient to obtain a sinus image abnormal cluster; identifying the deformed area of the sinus image according to the sinus image abnormal cluster.

2. The method for identifying a deformed area of a nasal endoscopic sinus image according to claim 1, characterized in that, The method for obtaining the corresponding inflection point curve of each of the edge curves includes: Performing curve fitting on the inflection points in each edge curve by using Newton interpolation method to obtain the corresponding inflection point curve of each of the edge curves, and the inflection point curve corresponds one-to-one with the starting endpoint and the ending endpoint of the edge curve.

3. A method for identifying a deformed area of a nasal sinus image under a nasal endoscope according to claim 1, characterized in that, The method for obtaining the curve difference coefficient includes: Constructing a Cartesian coordinate system for the edge curve and the inflection point curve, and obtaining the edge curve ordinate and the inflection point curve ordinate of each abscissa; Calculating the difference between the corresponding edge curve ordinate and the corresponding inflection point curve ordinate of each abscissa as the first difference; accumulating and summing the first differences of all abscissas to obtain the curve difference coefficient between each of the edge curves and the inflection point curve.

4. A method for identifying a deformed area of a nasal sinus image under a nasal endoscope according to claim 1, characterized in that, Sorting the sinus image clustering clusters according to the acquisition time nodes of each sinus image in each sinus image clustering cluster includes: Counting the starting time node and the ending time node of the maximum number of adjacent time node sinus images in each sinus image clustering cluster, and taking the time between the ending time node and the starting time node as the time stamp of each sinus image clustering cluster; Sorting each sinus image clustering cluster according to the time stamp to obtain the sequential arrangement of all sinus image clustering clusters.

5. A method for identifying a deformed area of a nasal sinus image under a nasal endoscope, characterized in that, The method for obtaining the abnormal cluster of sinus images includes: taking the cluster of sinus images with a specific coefficient greater than a preset first threshold as the abnormal cluster of sinus images.

6. A malformed area identification system for nasal endoscopic sinus images, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a method for identifying a deformed area of a sinus image under a nasal endoscope according to any one of claims 1 to 5.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of a method for identifying a deformed area of a sinus image under a nasal endoscope according to any one of claims 1 to 5.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a method for identifying a deformed area of a sinus image under a nasal endoscope according to any one of claims 1 to 5.

Images