An efficient detection method, system and device for craniopharyngioma lesions

Abstract

The present invention relates to the field of medical image processing technology, and specifically to an efficient detection method, system and device for craniopharyngioma lesions, including: obtaining a number of continuous images and performing threshold segmentation to obtain tissue regions and suspected tumor regions, obtaining edge feature values ​​of each image according to the shape characteristics of the tissue regions, obtaining the degree of compression of each image according to the position difference of tissue regions adjacent to the corresponding suspected tumor regions in adjacent images, thereby obtaining the probability of the same region of two adjacent groups of suspected diseased image groups and a tumor feature expression group and a craniopharyngioma image set, obtaining the probability of the craniopharyngioma region according to the change trend of the degree of compression of the tumor feature expression group in the craniopharyngioma image set, screening the images in the craniopharyngioma image set, and completing the detection of the craniopharyngioma lesion region. The present invention obtains images containing craniopharyngioma lesions by analyzing multiple MRI images, thereby obtaining images containing craniopharyngioma lesions.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to an efficient detection method, system and equipment for craniopharyngioma lesions. Background Art

[0002] Craniopharyngioma is a rare skull base tumor that usually occurs at the skull base. When seeking medical treatment, craniopharyngioma patients need to analyze their condition through MRI images generated by magnetic resonance imaging. In order to ensure that the patient's condition can be accurately analyzed, multiple images are usually generated during the detection process for comprehensive analysis.

[0003] The existing technology usually uses semantic segmentation or image recognition technology to analyze a single MRI image of a patient, and the analysis results obtained are of low accuracy. In addition, since a tumor is a volumetric disease, the patient's specific disease condition cannot be fully obtained by studying a single tumor cross-section in a single MRI image, thereby reducing the auxiliary effect on medical diagnosis and treatment. When multiple MRI images are analyzed in a collaborative manner, it is impossible to obtain images containing craniopharyngioma lesions in multiple MRI images, so that the MRI images cannot provide accurate and usable information. Summary of the invention

[0004] The present invention provides a highly efficient detection method, system and equipment for craniopharyngioma lesions to solve the existing problems.

[0005] The present invention provides an efficient detection method, system and device for craniopharyngioma lesions using the following technical solutions:

[0006] An embodiment of the present invention provides an efficient detection method for craniopharyngioma lesions, the method comprising the following steps:

[0007] Acquire several consecutive images;

[0008] Threshold segmentation is performed on each image to obtain the tissue area and suspected tumor area of ​​each image;

[0009] Obtaining edge feature values ​​of each image according to shape features of the tissue region of each image;

[0010] According to the position difference of the tissue regions adjacent to the corresponding two suspected tumor regions in the adjacent images, the squeezing degree of each image is obtained; and according to the squeezing degree of each image, a group of suspected diseased images is obtained;

[0011] The probability of the same area between two adjacent suspected disease image groups is obtained by the squeezing degree between the adjacent suspected disease image groups;

[0012] According to the probability of the same area between two adjacent groups of suspected diseased images, a tumor characteristic expression group and a craniopharyngioma image set were obtained;

[0013] According to the changing trend of the compression degree of the tumor characteristic expression group in the craniopharyngioma image collection, the craniopharyngioma regional probability of each tumor characteristic expression group was obtained;

[0014] According to the probability of craniopharyngioma area in each tumor characteristic expression group, the images in the craniopharyngioma image set are screened to obtain the craniopharyngioma analysis image group, and the detection of craniopharyngioma lesion area is completed.

[0015] Furthermore, the threshold segmentation of each image to obtain the tissue area and the suspected tumor area of ​​each image includes the following specific steps:

[0016] Each image was segmented by the maximum inter-class variance method to obtain several regions. Each region in the foreground area with a gray value greater than the threshold was recorded as a tissue region, and each region in the background area with a gray value less than the threshold was recorded as a suspected tumor region.

[0017] Furthermore, the step of obtaining the edge feature value of each image according to the shape feature of the tissue region of each image includes the following specific steps:

[0018] Perform edge detection on all tissue regions. For the current edge pixel point in the edge detection result, if the tangent direction between the current edge pixel point and the adjacent edge pixel point is greater than the preset angle threshold, record the current edge pixel point as the regional edge inflection point, traverse the edge of each tissue region, and count the number of regional edge inflection points of each tissue; establish a pixel coordinate system with the upper left corner pixel point of the image as the coordinate origin, and obtain the coordinates of each regional edge inflection point and the coordinates of the geometric center of each tissue region;

[0019] The calculation method of the edge eigenvalue of the tth image is:

[0020] ;

[0021] In the formula, represents the edge feature value of the t-th image, represents the number of tissue regions in the tth image, represents the number of inflection points of the region edge of the i-th tissue region in the t-th image, represents the mean gray value of all pixels in the i-th tissue area of ​​the t-th image, represents the angle between the vector of the jth inflection point of the ith tissue region of the tth image and the geometric center vector of the ith tissue region of the tth image, The vector representing the jth inflection point of the ith tissue region in the tth image, Represents the modulo operator for vectors.

[0022] Furthermore, the squeezing degree of each image is obtained according to the change of the tissue area adjacent to the suspected tumor area of ​​each image; and each suspected disease image group is obtained according to the squeezing degree of each image, including the following specific steps:

[0023] According to the order in which the images are generated, the geometric centers of all suspected tumor regions in all images are obtained, and for the geometric center of each suspected tumor region in each image, the geometric center of the suspected tumor region with the closest coordinates in the next image is obtained;

[0024] The suspected tumor region corresponding to the geometric center of the suspected tumor region with the closest coordinates in the next image is recorded as the control region of each suspected tumor region;

[0025] Each suspected tumor region and the control region constitute a suspected tumor control region group, and all suspected tumor control region groups in each image and the next image are obtained;

[0026] For a suspected tumor region in the first image in a suspected tumor control region group, the control region in the second image is recorded as the control tumor region of the suspected tumor region; in the first image, all tissue regions adjacent to the suspected tumor region are obtained, and in the image where the control tumor region of the suspected tumor region is located, each tissue region adjacent to the control tumor region is obtained to obtain the control region of each tissue region adjacent to the suspected tumor region;

[0027] The degree of squeezing of the tth image is calculated as:

[0028] ;

[0029] In the formula, Indicates the degree of squeezing of the tth image; represents the number of suspected tumor areas in the tth image, Indicates the difference between the tth image and the The number of tissue regions adjacent to the suspected tumor region, represents the edge feature value of the t-th image, represents the edge feature value of the t+1th image, Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The offset distance of each tissue region; Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The number of pixels in a tissue area; Indicates that in the t+1th image, the The first adjacent area to the suspected tumor The number of pixels in the control area of ​​each tissue area; It represents the absolute value function;

[0030] The squeezing degree of the tth image is normalized to interval, and obtain a normalized evaluation of the squeezing degree of each image; when the normalized evaluation of the squeezing degree of the t-th image is greater than a preset judgment threshold, the t-th image and the t+1-th image are recorded as a suspected disease image group.

[0031] Furthermore, the method of obtaining the probability of the same region between two adjacent groups of suspected diseased images through the relationship between the groups of suspected diseased images includes the following specific steps:

[0032] In the zth suspected disease image group, obtaining the mean of the offset distances of all suspected tumor control area groups in the suspected disease image group, and recording it as the suspected tumor offset distance of the zth suspected disease image group;

[0033] Obtain the suspected tumor offset distance of the z+1th suspected diseased image group;

[0034] The probability of the same region in the zth and z+1th suspected disease image groups is calculated as:

[0035] ;

[0036] In the formula, represents the probability of the same region between the zth and z+1th suspected disease image groups; represents the degree of compression of the first image in the z-th group of suspected diseased images, Indicates the degree of squeezing of the first image in the z+1th group of suspected diseased images; represents the suspected tumor offset distance of the zth suspected disease image group, represents the suspected tumor offset distance of the z+1th suspected diseased image group; represents an exponential function with a natural constant as base, represents the absolute value function, Represents a hyperparameter.

[0037] Furthermore, the method of obtaining the tumor characteristic expression group and the craniopharyngioma image set according to the probability of the same region of two adjacent groups of suspected diseased images includes the following specific steps:

[0038] The normalized evaluation of the probability of the same region of each suspected disease image group is obtained by normalization through the sigmoid function; when the normalized evaluation of the probability of the same region of the zth and z+1th suspected disease image groups is greater than the preset image group judgment threshold, the zth and z+1th suspected disease image combinations are respectively recorded as tumor characteristic expression groups;

[0039] Obtain all tumor feature expression groups after the zth suspected disease image group;

[0040] All tumor characteristic expression groups including the zth suspected disease image group constitute the zth craniopharyngioma image set.

[0041] Furthermore, the method of obtaining the craniopharyngioma region probability of each tumor characteristic expression group according to the difference in the degree of compression of the tumor characteristic expression group in the craniopharyngioma image set includes the following specific steps:

[0042] For the z-th craniopharyngioma image set, taking the tumor characteristic expression group with the highest probability in the same region in the craniopharyngioma image set as the segmentation point, the z-th craniopharyngioma image set is divided into the first sequence group and the second sequence group of the z-th craniopharyngioma image set, wherein the first sequence group includes from the tumor characteristic expression group with the highest probability in the same region to the first tumor characteristic expression group, and the second sequence includes from the next tumor characteristic expression group with the highest probability in the same region to the last tumor characteristic expression group;

[0043] For the first sequence group of the z-th craniopharyngioma image set Tumor characteristics group, The calculation method for the regional probability of craniopharyngioma in the tumor characteristic expression group is:

[0044] ;

[0045] represents the zth image in the craniopharyngioma image set The regional probability of craniopharyngioma in each tumor characteristic group, For the The tumor characteristics of the group show the degree of compression of the first image in the group. For the The group tumor characteristics show the degree of compression in the first image of the group.

[0046] Furthermore, the images in the craniopharyngioma image set are screened according to the craniopharyngioma region probability of each group of tumor characteristic expression groups to obtain a craniopharyngioma analysis image group and complete the detection of the craniopharyngioma lesion area, including the following specific steps:

[0047] Starting from the first group of tumor characteristic expression groups in the first sequence group of the zth craniopharyngioma image set, the craniopharyngioma region probability of each tumor characteristic expression group is calculated, and when the craniopharyngioma region probability is greater than or equal to 0, the next group of tumor characteristic expression groups in the first sequence group is calculated, until the next group of craniopharyngioma region probability is less than 0, the images of the tumor characteristic expression groups from the first group in the first sequence group to the current group are recorded as the craniopharyngioma analysis image group in the first sequence group of the zth craniopharyngioma image set;

[0048] Obtaining a craniopharyngioma analysis image group in the second sequence group of the z-th craniopharyngioma image set, a craniopharyngioma analysis image group in the first sequence group of the z-th craniopharyngioma image set, and a craniopharyngioma analysis image group in the second sequence group as a craniopharyngioma analysis image group representing craniopharyngioma regional characteristics;

[0049] The suspected tumor area of ​​each image in the craniopharyngioma analysis image group is marked, and the marked area is the craniopharyngioma lesion area.

[0050] The present invention also proposes an efficient detection system for craniopharyngioma lesions. When used, the steps of the aforementioned efficient detection method for craniopharyngioma lesions are performed. The system comprises:

[0051] An image acquisition module, used for acquiring multiple continuous images;

[0052] An image processing module, used to process different areas in multiple images;

[0053] The result screening module is used to screen the processed multiple images to obtain the craniopharyngioma lesion area.

[0054] The present invention also proposes an efficient detection device for craniopharyngioma lesions, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory is used to store the computer program, and the processor executes the computer program to implement the steps of the aforementioned efficient detection method for craniopharyngioma lesions.

[0055] The beneficial effects of the technical solution of the present invention are: obtaining a number of continuous images and performing threshold segmentation to obtain tissue areas and suspected tumor areas, obtaining edge feature values ​​of each image according to the shape characteristics of the tissue areas, which is helpful to find the corresponding areas in adjacent images; obtaining the degree of compression of each image according to the position difference of the tissue areas adjacent to the corresponding suspected tumor areas in the adjacent images, preliminarily screening out images containing craniopharyngioma areas, thereby obtaining a suspected diseased image group, and then obtaining the probability of the same area of ​​two adjacent suspected diseased image groups and a tumor feature expression group and a craniopharyngioma image set, showing the changes of the tumor area in different images; obtaining the probability of the craniopharyngioma area according to the change trend of the compression degree of the tumor feature expression group in the craniopharyngioma image set, screening the images in the craniopharyngioma image set, and accurately monitoring the craniopharyngioma lesion area. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be 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.

[0057] Figure 1 The present invention is a flowchart of the steps of an efficient detection method for craniopharyngioma lesions. DETAILED DESCRIPTION

[0058] 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 an efficient detection method, system and device for craniopharyngioma lesions proposed by the present invention, its specific implementation method, structure, features and effects, 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.

[0059] 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.

[0060] The specific scheme of an efficient detection method, system and device for craniopharyngioma lesions provided by the present invention is described in detail below with reference to the accompanying drawings.

[0061] First, see Figure 1 , which shows a flowchart of a method for efficiently detecting craniopharyngioma lesions provided by an embodiment of the present invention, the method comprising the following steps:

[0062] Step S001: Acquire several continuous images.

[0063] The purpose of this embodiment is to determine the craniopharyngioma lesion area by analyzing the MRI image. Since it is difficult to accurately reflect the patient's specific condition through only one MRI image of craniopharyngioma, multiple MRI images are required for joint analysis. Therefore, the patient's diseased part is scanned by an MRI device to obtain several continuous MRI images. It is particularly noted that the acquired MRI images are grayscale images.

[0064] Step S002: Obtain edge feature values ​​of each image according to the shape features of the tissue region of each image.

[0065] It should be noted that craniopharyngioma has a similar appearance in different images, so the shape of the craniopharyngioma area in each image is similar. However, since the edges of the craniopharyngioma shape are irregular, it is necessary to express the shape characteristics of the craniopharyngioma area through the edge features of the craniopharyngioma area in each image, so as to analyze the suspected craniopharyngioma areas in multiple adjacent images.

[0066] Specifically, for each acquired image, the maximum inter-class variance method is used to perform threshold segmentation on the grayscale image to obtain several regions, and the region with a grayscale value greater than the threshold is recorded as the foreground region, and the region with a grayscale value less than the threshold is recorded as the background region. Since the density of the tumor region is higher than that of the tissue, the tumor region is in the background region after threshold segmentation; each region in the foreground region is recorded as a tissue region, and each region in the background region is recorded as a suspected tumor region;

[0067] For each tissue region, edge detection is performed on the tissue region. For the current edge pixel in the edge detection result, among the edge pixels of each tissue region, if the tangent direction between the current edge pixel and the adjacent edge pixel is greater than the preset angle threshold, the current edge pixel is recorded as the region edge inflection point, and the edge of each tissue region is traversed to count the number of region edge inflection points of each tissue. A pixel coordinate system is established with the upper left corner pixel of the image as the coordinate origin to obtain the coordinates of each region edge inflection point and the coordinates of the geometric center of each tissue region.

[0068] The calculation method of the edge eigenvalue of the tth image is:

[0069] ;

[0070] In the formula, represents the edge feature value of the t-th image, represents the number of tissue regions in the tth image, represents the number of inflection points of the region edge of the i-th tissue region in the t-th image, represents the mean gray value of all pixels in the i-th tissue area of ​​the t-th image, represents the angle between the vector of the jth inflection point of the ith tissue region of the tth image and the geometric center vector of the ith tissue region of the tth image, The vector representing the jth inflection point of the ith tissue region in the tth image, Represents the modulo operator for vectors.

[0071] The angle between the vector of the jth inflection point of the ith tissue region of the tth image and the geometric center vector of the ith tissue region of the tth image is: Indicates the regional edge inflection point of the i-th tissue region The larger the product of its modulus length, the more prominent the regional edge inflection point of the i-th tissue region is, and the better it can express the edge shape characteristics of the craniopharyngioma region.

[0072] Step S003, obtaining the degree of squeezing of each image according to the changes in the tissue areas adjacent to the suspected tumor areas of each adjacent image; obtaining each suspected diseased image group according to the degree of squeezing of each image; obtaining the probability of the same areas of two adjacent suspected diseased image groups through the relationship between the suspected diseased image groups; obtaining the tumor characteristic expression group and the craniopharyngioma image set according to the probability of the same areas of two adjacent suspected diseased image groups.

[0073] It should be noted that since craniopharyngioma is formed by the division and growth of residual epithelial cells, it will squeeze the surrounding brain tissue as the tumor continues to grow. During the squeezing process, the position of the tumor remains basically unchanged, while the geometric center position of the surrounding tissue area will shift along the direction of squeezing. As the number of images increases, the greater the difference in the tumor area between the two images, the higher the pressure on the surrounding area. Therefore, based on the situation where the area position in adjacent images remains unchanged while the geometric center position of the surrounding area shifts, the offset of the surrounding area, that is, the degree of squeezing, is analyzed to obtain the difference in the degree of squeezing of the corresponding tissue areas in adjacent images.

[0074] Specifically, the geometric centers of all suspected tumor regions in all images are obtained in the order in which the images are generated, and for the geometric center of each suspected tumor region in each image, the geometric center of the suspected tumor region with the closest coordinates in the next image is obtained;

[0075] The suspected tumor region corresponding to the geometric center of the suspected tumor region with the closest coordinates in the next image is recorded as the control region of each suspected tumor region;

[0076] Each suspected tumor region and the control region constitute a suspected tumor control region group, and all suspected tumor control region groups in each image and the next image are obtained.

[0077] For a suspected tumor region in the first image of a suspected tumor control region group, the control region in the second image is recorded as the control tumor region of the suspected tumor region; in the first image, all tissue regions adjacent to the suspected tumor region are obtained, and in the image where the control tumor region of the suspected tumor region is located, each tissue region adjacent to the control tumor region is obtained to obtain the control region of each tissue region adjacent to the suspected tumor region.

[0078] The Euclidean distance between each tissue region adjacent to the suspected tumor region and the geometric center of the control region of each tissue region is recorded as the offset distance of each tissue region adjacent to the suspected tumor region.

[0079] Then the degree of squeezing of the tth image is calculated as:

[0080] ;

[0081] In the formula, Indicates the degree of squeezing of the tth image; represents the number of suspected tumor areas in the tth image, Indicates the difference between the tth image and the The number of tissue regions adjacent to the suspected tumor region, represents the edge feature value of the t-th image, represents the edge feature value of the t+1th image, Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The offset distance of each tissue region; Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The number of pixels in a tissue area; Indicates that in the t+1th image, the The first adjacent area to the suspected tumor The number of pixels in the control area of ​​each tissue area; It represents the absolute value function;

[0082] when The further away from 1, the less similar the edge feature values ​​are between the tth image and the t+1th image, indicating that the difference in tissue regions between the two images is greater. It is more likely that as the number of images increases, the tumor region will become larger, and the pressure on the surrounding area will be higher, so the compression degree will be greater. The larger the value, the more similar the tth image is to the first The first adjacent area to the suspected tumor The organizational area is affected by The greater the squeezing effect of a suspected tumor area, the greater the squeezing degree value; Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The greater the difference in the area of ​​each tissue region, the greater the difference in compression.

[0083] The squeezing degree of the tth image is normalized to interval, and obtain the normalized evaluation of the degree of compression of each image; preset a judgment threshold, when the normalized evaluation of the degree of compression of the t-th image is greater than the preset judgment threshold, it means that the characteristics of the tumor compressing the surrounding tissue area in the t-th image and the t+1-th image are stronger; the t-th image and the t+1-th image are recorded as a suspected diseased image group, and similarly, all suspected diseased image groups are obtained.

[0084] It should be further explained that, since the tumor will appear in multiple images after scanning due to its proliferation, the tumor area will squeeze the surrounding area during its growth, while its own position is relatively stable. Therefore, the geometric center position offset of the tumor area in multiple connected images is small and will not produce significant changes.

[0085] In the zth suspected disease image group, the mean of the offset distances of all suspected tumor control area groups in the suspected disease image group is obtained, which is recorded as the suspected tumor offset distance of the zth suspected disease image group; similarly, the suspected tumor offset distance of the z+1th suspected disease image group is obtained.

[0086] Then the probability of the same area in the zth and z+1th suspected disease image groups is calculated as:

[0087] ;

[0088] In the formula, represents the probability of the same region between the zth and z+1th suspected disease image groups; represents the degree of compression of the first image in the z-th group of suspected diseased images, Indicates the degree of squeezing of the first image in the z+1th group of suspected diseased images; represents the suspected tumor offset distance of the zth suspected disease image group, represents the suspected tumor offset distance of the z+1th suspected diseased image group; represents an exponential function with a natural constant as base, represents the absolute value function, Represents a hyperparameter, which is used to avoid the denominator being 0. In this embodiment, Let's take this as an example.

[0089] It represents the ratio of the squeezing degree of the suspected tumor area in the first image of the zth and z+1th suspected disease image groups. The farther away from 1 the value is, the greater the squeezing degree of the suspected tumor area on the adjacent area in the two adjacent suspected disease image groups is, that is, the size of the suspected tumor area changes with the change of the image. Therefore, the probability value of the same area in the zth and z+1th suspected disease image groups is greater; It is the difference in the position change of the suspected tumor area in the first image of the zth and z+1th suspected disease image groups. The smaller the value, the more likely it is that the center position of the suspected tumor area has not changed, and the more likely it is that they belong to the same suspected tumor area.

[0090] The probability of the same area between the zth and z+1th suspected disease image groups is normalized to interval, and obtain the normalized evaluation of the probability of the same area of ​​each suspected disease image group; preset the image group judgment threshold, when the normalized evaluation of the probability of the same area of ​​the zth and z+1th suspected disease image groups is greater than the preset image group judgment threshold, it means that the suspected tumor areas contained in the zth and z+1th suspected disease image groups are likely to be the same area, and the zth and z+1th suspected disease image combinations are respectively recorded as tumor feature expression groups, and similarly, all tumor feature expression groups after the zth suspected disease image group are obtained, including all tumor feature expression groups including the zth suspected disease image group, constitute the zth craniopharyngioma image set.

[0091] Step S004: Screen the images in the craniopharyngioma image set according to the craniopharyngioma region probability of each tumor characteristic expression group to obtain a craniopharyngioma analysis image group, and screen the craniopharyngioma image group to obtain the craniopharyngioma lesion region.

[0092] It should be noted that craniopharyngioma is mainly divided into two subtypes: amelogenocyte-type craniopharyngioma and papillary craniopharyngioma. Both subtypes present a shape that is large in the middle and small at both ends. Therefore, different positions of craniopharyngioma have different effects on the degree of compression of other adjacent areas. The diameter in the middle position of the craniopharyngioma will be larger than that at the two ends, and the degree of compression on the adjacent areas is the greatest, while the degree of compression on the adjacent areas at the two ends of the tumor is the smallest. Therefore, the deviation of the geometric center of the adjacent area in the middle position of the craniopharyngioma is larger than the deviation of the geometric center of the adjacent area at the two ends of the craniopharyngioma. Therefore, this embodiment obtains an image showing the characteristics of large in the middle and small at both ends in the craniopharyngioma image group by analyzing the image with the maximum expression of the craniopharyngioma, and obtains a craniopharyngioma analysis image group showing the regional characteristics of the craniopharyngioma.

[0093] Specifically, for the zth craniopharyngioma image set, the tumor characteristic expression group with the highest probability in the same area in the craniopharyngioma image set is taken as the dividing point, and the zth craniopharyngioma image set is divided into the first sequence group and the second sequence group of the zth craniopharyngioma image set, wherein the first sequence group includes from the tumor characteristic expression group with the highest probability in the same area to the first tumor characteristic expression group, and the second sequence is from the next tumor characteristic expression group of the tumor characteristic expression group with the highest probability in the same area to the last tumor characteristic expression group.

[0094] For the first sequence group of the z-th craniopharyngioma image set Tumor characteristics group, The calculation method for the regional probability of craniopharyngioma in the tumor characteristic expression group is:

[0095] ;

[0096] represents the zth image in the craniopharyngioma image set The regional probability of craniopharyngioma in each tumor characteristic group, For the The tumor characteristics of the group show the degree of compression of the first image in the group. For the The group tumor characteristics show the degree of compression in the first image of the group. It represents the change of the degree of compression of adjacent tumor feature expression groups in the first sequence group of the zth craniopharyngioma image set. If the craniopharyngioma has the characteristics of being large in the middle and small at both ends, the probability of the craniopharyngioma area is greater than or equal to 0. If not, the probability of the craniopharyngioma area is less than 0.

[0097] Further, starting from the first group of tumor characteristic expression groups in the first sequence group of the zth craniopharyngioma image set, the probability of the craniopharyngioma region is calculated. When the probability of the craniopharyngioma region is greater than or equal to 0, the next group of tumor characteristic expression groups in the first sequence group is calculated, until the probability of the craniopharyngioma region of the next group of the current group is less than 0, the images of the tumor characteristic expression groups from the first group in the first sequence group to the current group are recorded as the craniopharyngioma analysis image group in the first sequence group of the zth craniopharyngioma image set. Similarly, the craniopharyngioma analysis image group in the second sequence group of the zth craniopharyngioma image set, the craniopharyngioma analysis image group in the first sequence group of the zth craniopharyngioma image set, and the craniopharyngioma analysis image group in the second sequence group are obtained as the final analysis image showing the craniopharyngioma region characteristics; the suspected tumor area of ​​each image in the final analysis image is marked, and the marked area is the craniopharyngioma lesion area.

[0098] The marked area is image enhanced. This embodiment uses the histogram equalization algorithm as an example for enhancement. The histogram equalization algorithm is a well-known technology and will not be described in detail in this embodiment. The enhanced image is output to a high-resolution display device to display the craniopharyngioma lesion area.

[0099] In a second aspect, the present invention further proposes an efficient detection system for craniopharyngioma lesions, which executes the steps of the aforementioned efficient detection method for craniopharyngioma lesions when used, and the system comprises:

[0100] An image acquisition module, used for acquiring multiple continuous images;

[0101] An image processing module, used to process different areas in multiple images;

[0102] The result screening module is used to screen the processed multiple images to obtain the craniopharyngioma lesion area.

[0103] In a third aspect, the present invention also provides an efficient detection device for craniopharyngioma lesions, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory is used to store the computer program, and the processor executes the computer program to implement the steps of the aforementioned efficient detection method for craniopharyngioma lesions.

[0104] So far, the present invention is completed.

[0105] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An efficient detection method for craniopharyngioma lesions, characterized in that: The method comprises the following steps: Acquire several consecutive images; Threshold segmentation is performed on each image to obtain the tissue area and suspected tumor area of ​​each image; Obtaining edge feature values ​​of each image according to shape features of the tissue region of each image; According to the position difference of the tissue regions adjacent to the corresponding two suspected tumor regions in the adjacent images, the squeezing degree of each image is obtained; and according to the squeezing degree of each image, a group of suspected diseased images is obtained; The method of obtaining the squeezing degree of each image according to the position difference of the tissue regions adjacent to the corresponding two suspected tumor regions in the adjacent images and obtaining the suspected diseased image group according to the squeezing degree of each image includes the following specific steps: According to the order in which the images are generated, the geometric centers of all suspected tumor regions in all images are obtained, and for the geometric center of each suspected tumor region in each image, the geometric center of the suspected tumor region with the closest coordinates in the next image is obtained; The suspected tumor region corresponding to the geometric center of the suspected tumor region with the closest coordinates in the next image is recorded as the control region of each suspected tumor region; Each suspected tumor region and the control region constitute a suspected tumor control region group, and all suspected tumor control region groups in each image and the next image are obtained; For a suspected tumor region in the first image in a suspected tumor control region group, the control region in the second image is recorded as the control tumor region of the suspected tumor region; in the first image, all tissue regions adjacent to the suspected tumor region are obtained, and in the image where the control tumor region of the suspected tumor region is located, each tissue region adjacent to the control tumor region is obtained to obtain the control region of each tissue region adjacent to the suspected tumor region; The degree of squeezing of the tth image is calculated as: ； In the formula, Indicates the degree of squeezing of the tth image; represents the number of suspected tumor areas in the tth image, Indicates the difference between the tth image and the The number of tissue regions adjacent to the suspected tumor region, represents the edge feature value of the t-th image, represents the edge feature value of the t+1th image, Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The offset distance of each tissue region; Indicates the difference between the tth image and the The first adjacent area to the suspected tumor The number of pixels in a tissue area; Indicates that in the t+1th image, the The first adjacent area to the suspected tumor The number of pixels in the control area of ​​each tissue area; It represents the absolute value function; The squeezing degree of the tth image is normalized to interval, and obtain a normalized evaluation of the squeezing degree of each image; when the normalized evaluation of the squeezing degree of the t-th image is greater than a preset judgment threshold, the t-th image and the t+1-th image are recorded as a suspected disease image group; The probability of the same area between two adjacent suspected disease image groups is obtained by the squeezing degree between the adjacent suspected disease image groups; The method of obtaining the probability of the same region between two adjacent groups of suspected diseased images by the degree of squeezing between adjacent groups of suspected diseased images comprises the following specific steps: In the zth suspected disease image group, obtaining the mean of the offset distances of all suspected tumor control area groups in the suspected disease image group, and recording it as the suspected tumor offset distance of the zth suspected disease image group; Obtain the suspected tumor offset distance of the z+1th suspected diseased image group; The probability of the same region in the zth and z+1th suspected disease image groups is calculated as: ； In the formula, represents the probability of the same region between the zth and z+1th suspected disease image groups; represents the degree of compression of the first image in the z-th group of suspected diseased images, Indicates the degree of squeezing of the first image in the z+1th group of suspected diseased images; represents the suspected tumor offset distance of the zth suspected disease image group, represents the suspected tumor offset distance of the z+1th suspected diseased image group; represents an exponential function with a natural constant as base, represents the absolute value function, represents a hyperparameter; According to the probability of the same area between two adjacent groups of suspected diseased images, a tumor characteristic expression group and a craniopharyngioma image set were obtained; According to the changing trend of the compression degree of the tumor characteristic expression group in the craniopharyngioma image collection, the craniopharyngioma regional probability of each tumor characteristic expression group was obtained; The method of obtaining the craniopharyngioma region probability of each tumor characteristic expression group according to the change trend of the compression degree of the tumor characteristic expression group in the craniopharyngioma image set includes the following specific steps: For the z-th craniopharyngioma image set, taking the tumor characteristic expression group with the highest probability in the same region in the craniopharyngioma image set as the segmentation point, the z-th craniopharyngioma image set is divided into the first sequence group and the second sequence group of the z-th craniopharyngioma image set, wherein the first sequence group includes from the tumor characteristic expression group with the highest probability in the same region to the first tumor characteristic expression group, and the second sequence includes from the next tumor characteristic expression group with the highest probability in the same region to the last tumor characteristic expression group; For the first sequence group of the z-th craniopharyngioma image set Tumor characteristics group, The calculation method for the regional probability of craniopharyngioma in the tumor characteristic expression group is: ; represents the zth image in the craniopharyngioma image set The regional probability of craniopharyngioma in each tumor characteristic group, For the The tumor characteristics of the group show the degree of compression of the first image in the group. For the The group tumor features show the degree of compression in the first image of the group; According to the probability of craniopharyngioma area in each tumor characteristic expression group, the images in the craniopharyngioma image set are screened to obtain the craniopharyngioma analysis image group, and the detection of craniopharyngioma lesion area is completed.

2. An efficient detection method for craniopharyngioma lesions according to claim 1, characterized in that: The threshold segmentation of each image to obtain the tissue area and the suspected tumor area of ​​each image includes the following specific steps: Each image was segmented by the maximum inter-class variance method to obtain several regions. Each region in the foreground area with a gray value greater than the threshold was recorded as a tissue region, and each region in the background area with a gray value less than the threshold was recorded as a suspected tumor region.

3. The efficient detection method for craniopharyngioma lesions according to claim 1, characterized in that: The step of obtaining the edge feature value of each image according to the shape feature of the tissue region of each image includes the following specific steps: Perform edge detection on all tissue regions. For the current edge pixel in the edge detection result, if the tangent direction between the current edge pixel and the adjacent edge pixel is greater than the preset angle threshold, record the current edge pixel as the region edge inflection point, traverse the edge of each tissue region, and count the number of region edge inflection points of each tissue; A pixel coordinate system is established with the pixel point at the upper left corner of the image as the coordinate origin to obtain the coordinates of the edge inflection point of each region and the coordinates of the geometric center of each tissue region; The calculation method of the edge eigenvalue of the tth image is: ； In the formula, represents the edge feature value of the t-th image, represents the number of tissue regions in the tth image, represents the number of inflection points of the region edge of the i-th tissue region in the t-th image, represents the mean gray value of all pixels in the i-th tissue area of ​​the t-th image, represents the angle between the vector of the jth inflection point of the ith tissue region of the tth image and the geometric center vector of the ith tissue region of the tth image, The vector representing the jth inflection point of the ith tissue region in the tth image, Represents the modulo operator for vectors.

4. The efficient detection method for craniopharyngioma lesions according to claim 1, characterized in that: The method of obtaining a tumor characteristic expression group and a craniopharyngioma image set according to the probability of the same region between two adjacent groups of suspected diseased images includes the following specific steps: The normalized evaluation of the probability of the same region of each suspected disease image group is obtained by normalization through the sigmoid function; when the normalized evaluation of the probability of the same region of the zth and z+1th suspected disease image groups is greater than the preset image group judgment threshold, the zth and z+1th suspected disease image combinations are respectively recorded as tumor characteristic expression groups; Obtain all tumor feature expression groups after the zth suspected disease image group; All tumor characteristic expression groups including the zth suspected disease image group constitute the zth craniopharyngioma image set.

5. The efficient detection method for craniopharyngioma lesions according to claim 1, characterized in that: The images in the craniopharyngioma image set are screened according to the craniopharyngioma region probability of each group of tumor characteristic expression groups to obtain a craniopharyngioma analysis image group, and complete the detection of the craniopharyngioma lesion area, including the following specific steps: Starting from the first group of tumor characteristic expression groups in the first sequence group of the zth craniopharyngioma image set, the craniopharyngioma region probability of each tumor characteristic expression group is calculated, and when the craniopharyngioma region probability is greater than or equal to 0, the next group of tumor characteristic expression groups in the first sequence group is calculated, until the next group of craniopharyngioma region probability is less than 0, the images of the tumor characteristic expression groups from the first group in the first sequence group to the current group are recorded as the craniopharyngioma analysis image group in the first sequence group of the zth craniopharyngioma image set; Obtaining a craniopharyngioma analysis image group in the second sequence group of the z-th craniopharyngioma image set, a craniopharyngioma analysis image group in the first sequence group of the z-th craniopharyngioma image set, and a craniopharyngioma analysis image group in the second sequence group as a craniopharyngioma analysis image group representing craniopharyngioma regional characteristics; The suspected tumor area of ​​each image in the craniopharyngioma analysis image group is marked, and the marked area is the craniopharyngioma lesion area.

6. An efficient detection system for craniopharyngioma lesions, characterized in that: When used, the method for efficiently detecting craniopharyngioma lesions according to any one of claims 1 to 5 is performed, and the system comprises: An image acquisition module, used for acquiring multiple continuous images; An image processing module, used to process different areas in multiple images; The result screening module is used to screen the processed multiple images to obtain the craniopharyngioma lesion area.

7. An efficient detection device for craniopharyngioma lesions, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of an efficient detection method for craniopharyngioma lesions as described in any one of claims 1 to 5 are implemented.

Images