A method for detecting invasion boundaries of osteosarcoma before chemotherapy in children based on MRI scans
By segmenting and dividing grayscale images of pediatric osteosarcoma from MRI scans into blocks and regions, and calculating enhancement coefficients for personalized enhancement, the problem of loss of detail information in existing technologies is solved, and clear display of the invasion boundary of osteosarcoma is achieved.
Patent Information
- Application Number
- CN202411337967.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-25
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-25
AI Technical Summary
Existing image enhancement methods lose some detail information when enhancing pediatric osteosarcoma areas on MRI scans, resulting in unclear invasion boundaries.
By uniformly dividing grayscale images of pediatric osteosarcoma from MRI scans into blocks, calculating similarity parameters and texture feature differences between sub-blocks, different regions are segmented, and enhancement coefficients are calculated based on texture richness and grayscale discriminability to perform personalized enhancement on the regions to be enhanced.
The image enhancement effect is improved, while retaining detailed information, making the invasion boundary of osteosarcoma clearer and easier for medical staff to observe.
Smart Images

Figure CN118864336B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of image enhancement, and in particular to a method for detecting invasion boundaries before chemotherapy of childhood osteosarcoma based on MRI scanning. Background Art
[0002] Osteosarcoma is a malignant bone tumor that is common in children and adolescents. It usually occurs at the ends of long bones, especially near the knee joint. It often has invasive growth characteristics and may invade adjacent soft tissues, blood vessels and nerves. It is usually necessary to use MRI (magnetic resonance imaging) technology to perform medical imaging scans on the patient's osteosarcoma site so that relevant medical staff can observe and analyze the invasion of osteosarcoma.
[0003] Since the contrast between the osteosarcoma area and the adjacent tissues and bones in the collected medical images is low, in the related technology, the image is usually enhanced as a whole using methods such as histogram equalization to improve the contrast of various areas in the image, so as to facilitate the observation of the boundary invasion of osteosarcoma on the surrounding tissues. However, due to the large differences in grayscale features and texture features between the osteosarcoma area and other areas, the existing enhancement methods will lose some detail information after image enhancement, reduce the image enhancement effect, and thus fail to clearly display the invasion boundary of osteosarcoma in the image. Summary of the invention
[0004] In order to solve the technical problem that the existing image enhancement method loses some detail information after image enhancement, reduces the image enhancement effect, and thus cannot clearly display the invasion boundary of osteosarcoma in the image, the purpose of the present invention is to provide a method for detecting the invasion boundary of childhood osteosarcoma before chemotherapy based on MRI scanning. The technical scheme adopted is as follows:
[0005] The present invention proposes a method for detecting invasion boundaries before chemotherapy for childhood osteosarcoma based on MRI scanning, the method comprising:
[0006] A grayscale image of an osteosarcoma at a target site of a pediatric patient scanned by MRI is obtained;
[0007] The grayscale image is uniformly divided into blocks to obtain a plurality of sub-blocks in the grayscale image, and a similarity parameter between any two sub-blocks is obtained according to a difference in grayscale values of pixels and a difference in gradient directions of pixels between any two sub-blocks; based on the similarity parameter between any two sub-blocks, the grayscale image is segmented to obtain a plurality of segmented regions;
[0008] According to the distribution of the gradient amplitude of the pixel points in each segmented area, the texture richness of each segmented area is obtained; based on the texture richness, a plurality of areas to be enhanced are screened out from all the segmented areas; any area to be enhanced is taken as a target area to be enhanced, and according to the difference in the distribution of the grayscale values of the pixel points between the target area to be enhanced and other areas to be enhanced except the target area to be enhanced, and the texture richness of the target area to be enhanced, an enhancement coefficient of the target area to be enhanced is obtained;
[0009] According to the distribution of the grayscale values of the pixels in each area to be enhanced in the grayscale image and the enhancement coefficient, the grayscale values of the pixels in the area to be enhanced are adjusted to obtain an enhanced image.
[0010] Further, obtaining the similarity parameter between any two sub-blocks includes:
[0011] Take any sub-block in the grayscale image as the target sub-block;
[0012] Analyze the overall level of the grayscale values of all pixels in the target sub-block to obtain the overall grayscale value of the target sub-block;
[0013] The sequence composed of the gradient directions of all pixels in the target sub-block is used as the texture feature sequence of the target sub-block;
[0014] According to the difference of the overall grayscale values between any two sub-blocks, obtaining the overall grayscale difference value between any two sub-blocks;
[0015] Obtaining a texture feature difference between any two sub-blocks according to a difference in gradient direction in the texture feature sequence between any two sub-blocks;
[0016] The overall grayscale difference value and the texture feature difference are integrated to obtain a comprehensive difference between any two sub-blocks, and the comprehensive difference is negatively correlated and normalized to obtain a similarity parameter between any two sub-blocks.
[0017] Further, obtaining the texture feature difference between any two sub-blocks according to the difference in the gradient direction in the texture feature sequence between any two sub-blocks includes:
[0018] Based on a dynamic time warping algorithm, the texture feature sequence between any two sub-blocks is processed to obtain the texture feature difference between any two sub-blocks.
[0019] Further, obtaining a plurality of segmented regions includes:
[0020] Based on the region growing algorithm and according to the similarity parameter between any two sub-blocks, region growing is performed on the grayscale image to obtain multiple segmented regions in the grayscale image, wherein the merging rule of region growing is that when the similarity parameter between the sub-blocks is less than a preset similarity threshold, merging is performed.
[0021] Furthermore, obtaining the texture richness of each segmented region includes:
[0022] Take any segmentation area as the target segmentation area, and take the extreme difference of the gradient amplitude of all pixels in the target segmentation area as the first richness parameter of the target segmentation area;
[0023] Analyze the overall level of the gradient amplitude of all pixels in the target segmentation area to obtain the overall gradient amplitude of the target segmentation area;
[0024] Obtaining a gradient amplitude deviation of each pixel point in the target segmentation area according to a difference between the gradient amplitude of each pixel point in the target segmentation area and the overall gradient amplitude;
[0025] The average value of the gradient amplitude deviation of all pixels in the target segmentation area is used as the second richness parameter of the target segmentation area;
[0026] The first richness parameter and the second richness parameter are integrated and normalized to obtain the texture richness of the target segmented area.
[0027] Furthermore, the step of selecting a plurality of areas to be enhanced from all segmented areas based on the texture richness includes:
[0028] The segmented area whose texture richness is greater than a preset richness threshold is used as the area to be enhanced.
[0029] Furthermore, obtaining the enhancement coefficient of the target area to be enhanced includes:
[0030] Analyze the overall level of the grayscale values of all pixels in each area to be enhanced to obtain the overall grayscale parameter of each area to be enhanced;
[0031] Obtaining a grayscale distinction between the target region to be enhanced and each other region to be enhanced according to a difference in the overall grayscale parameter between the target region to be enhanced and each other region to be enhanced except the target region to be enhanced;
[0032] An enhancement coefficient of the target region to be enhanced is obtained according to the grayscale distinction between the target region to be enhanced and each other region to be enhanced, and the texture richness of the target region to be enhanced.
[0033] Further, obtaining the enhancement coefficient of the target region to be enhanced according to the grayscale distinction between the target region to be enhanced and each other region to be enhanced, and the texture richness of the target region to be enhanced includes:
[0034] Performing negative correlation mapping on the average values of the grayscale distinctions between the target region to be enhanced and all other regions to be enhanced to obtain the enhancement degree of the target region to be enhanced;
[0035] Using the texture richness of the target area to be enhanced, weighting the enhancement degree to obtain an adjustment amount of the target area to be enhanced;
[0036] The sum of the enhancement degree and the adjustment amount is normalized to obtain an enhancement coefficient of the target area to be enhanced.
[0037] Further, obtaining the enhanced image includes:
[0038] For any area to be enhanced in the grayscale image, all pixels in the area to be enhanced are divided into two categories based on the grayscale value, and the category with the largest average grayscale value of the pixels is taken as the high grayscale category, and the category with the smallest average grayscale value of the pixels is taken as the low grayscale category;
[0039] Taking the product value of the grayscale value of the pixel point in the area to be enhanced and the enhancement coefficient of the area to be enhanced as the grayscale adjustment amount of the pixel point in the area to be enhanced;
[0040] If the pixel in the area to be enhanced belongs to the high grayscale category, the sum of the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel; if the pixel in the area to be enhanced belongs to the low grayscale category, the difference between the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel;
[0041] The grayscale values of the pixels in the area to be enhanced in the grayscale image are replaced with the adjusted grayscale values to obtain an enhanced image.
[0042] Furthermore, the method of dividing all pixels in the area to be enhanced into two categories based on grayscale values includes:
[0043] Based on the gray value, all the pixels in the enhanced area are clustered to obtain two clusters. The pixels in one cluster are regarded as one class, and the pixels in the other cluster are regarded as another class.
[0044] The present invention has the following beneficial effects:
[0045] The present invention takes into account that the existing image enhancement methods will lose some detail information after image enhancement, reduce the image enhancement effect, and thus fail to clearly display the invasion boundary of osteosarcoma in the image. Therefore, a grayscale image of the osteosarcoma at the test site of a pediatric patient who has undergone MRI scanning is first obtained. Considering that the grayscale image not only has background areas, but also has other areas such as tissues and bones other than the osteosarcoma area, and there are large differences in grayscale features and texture features between different areas, the grayscale image can be evenly divided into blocks to obtain multiple sub-blocks, and the similarity parameters are used to reflect the similarity of grayscale and texture features between the sub-blocks, thereby segmenting the grayscale image to accurately distinguish different areas in the grayscale image. Considering that the segmented area also includes a background area that does not contain any information, and there is no need to subsequently target the background. The target area is enhanced instead of the background area. For example, there are certain texture features in the osteosarcoma area and the tissue bone area. Therefore, the texture richness can be used to reflect the degree to which the segmented area contains texture features, and the area to be enhanced can be screened out. Considering that there are large differences in the grayscale and texture features in different areas to be enhanced, it is necessary to enhance different areas to be enhanced to different degrees, and it is necessary to ensure that the distinction between the enhanced areas to be enhanced is large, and the texture features in each area to be enhanced are more obvious. Therefore, the degree of enhancement of the target area to be enhanced can be reflected by the enhancement coefficient, and then the grayscale value of the pixel in the area to be enhanced is adjusted to obtain an enhanced image, thereby improving the final image enhancement effect, retaining the detail information in the image, and making the invasion boundary of osteosarcoma on the surrounding tissue and bone more clearly displayed. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0047] Figure 1 A flow chart of a method for detecting invasion boundaries before chemotherapy for osteosarcoma in children based on MRI scanning provided by one embodiment of the present invention;
[0048] Figure 2 A flow chart of a method for obtaining an enhancement coefficient of a target area to be enhanced provided by an embodiment of the present invention;
[0049] Figure 3 A schematic diagram of a grayscale image of an osteosarcoma at a test site of a child patient scanned by MRI provided by an embodiment of the present invention;
[0050] Figure 4A schematic diagram of an enhanced image provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0051] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a method for detecting invasion boundaries before chemotherapy for osteosarcoma in children based on MRI scanning proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0052] 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.
[0053] The following is a detailed description of a specific scheme of a method for detecting invasion boundaries before chemotherapy for osteosarcoma in children based on MRI scanning provided by the present invention in conjunction with the accompanying drawings.
[0054] See also Figure 1 , which shows a flow chart of a method for detecting invasion boundaries before chemotherapy for osteosarcoma in children based on MRI scanning according to an embodiment of the present invention, the method comprising:
[0055] Step S1: obtaining a grayscale image of an osteosarcoma at a test site of a child patient scanned by MRI.
[0056] The embodiment of the present invention first uses the MIR device to scan the part to be tested of the child patient with osteosarcoma symptoms, so as to obtain the original image of the part to be tested, wherein the part to be tested may be a part prone to osteosarcoma, such as the knee joint of the human body, and then grayscale processing is performed on the original image to obtain a single-channel grayscale image to improve the efficiency of subsequent image data processing. It should be noted that grayscale processing is a technical means well known to those skilled in the art and will not be described in detail here. Please refer to Figure 3 , which shows a schematic diagram of a grayscale image of an osteosarcoma at a test site of a child patient scanned by MRI according to an embodiment of the present invention.
[0057] At this point, the grayscale image of the osteosarcoma at the test site of the child patient has been collected.
[0058] Step S2: Evenly divide the grayscale image into blocks to obtain multiple sub-blocks in the grayscale image, and obtain the similarity parameters between any two sub-blocks based on the difference in grayscale values of pixels and the difference in gradient directions of pixels between any two sub-blocks; based on the similarity parameters between any two sub-blocks, segment the grayscale image to obtain multiple segmented areas.
[0059] Since existing image enhancement methods, such as grayscale histogram equalization, enhance the image as a whole, some detail information in the enhanced image is lost, and the enhancement effect is poor. In the grayscale image of osteosarcoma at the test site, there are not only background areas, but also other areas such as tissues and bones in addition to the osteosarcoma area. There are large differences in grayscale features and texture features between different areas. Therefore, an embodiment of the present invention first divides the grayscale image into uniform blocks to obtain multiple sub-blocks in the grayscale image. Subsequently, the similarity of the grayscale features and texture features displayed by each sub-block can be analyzed, and the sub-blocks with similar features can be merged to achieve segmentation of the grayscale image and obtain multiple different areas in the grayscale image.
[0060] In an embodiment of the present invention, the size of the sub-block is set to The specific size of the sub-block can also be set by the implementer according to the specific implementation scenario and is not limited here.
[0061] It should be noted that in the process of evenly dividing the grayscale image into blocks, due to boundary problems, it may be impossible to obtain complete sub-blocks at the image boundaries. At this time, the pixel points at the grayscale image boundary can be filled to ensure that complete sub-blocks are divided. Image boundary filling is a technical means well known to technical personnel in this field and will not be elaborated here.
[0062] From the above analysis, it can be seen that there are differences in grayscale and texture features between the background area, osteosarcoma area and the surrounding tissue and bone area, and the texture feature can be reflected by the gradient direction of the pixel point. Therefore, the embodiment of the present invention first uses the existing edge detection operator such as the Sobel operator or the Roberts operator to calculate the gradient direction of each pixel point in the grayscale image, wherein the gradient direction is usually represented by an angle, and then the difference in the grayscale value of the pixel point between any two sub-blocks and the difference in the gradient direction of the pixel point are analyzed. The similarity parameter obtained reflects the similarity of the grayscale and texture features between the sub-blocks. The larger the similarity parameter, the more similar the grayscale features and texture features between the two sub-blocks are. Subsequently, based on the similarity parameter, the sub-blocks can be merged to achieve segmentation of the grayscale image, thereby accurately distinguishing different areas in the grayscale image.
[0063] Preferably, in an embodiment of the present invention, a method for obtaining similarity parameters between any two sub-blocks specifically includes:
[0064] Firstly, an analysis is performed on any sub-block, and any sub-block in the grayscale image is taken as the target sub-block. The overall level of the grayscale values of all pixels in the target sub-block is analyzed to obtain the overall grayscale value of the target sub-block. The overall grayscale value reflects the overall grayscale distribution characteristics of the target sub-block. The sequence composed of the gradient directions of all pixels in the target sub-block is taken as the texture feature sequence of the target sub-block. The texture feature sequence contains the information of the texture distribution characteristics of the target sub-block.
[0065] In an embodiment of the present invention, the average or median of the grayscale values of all pixels in the target sub-block may be used as the overall grayscale value of the target sub-block to analyze the overall level of the grayscale values of all pixels in the target sub-block, which is not limited here.
[0066] In one embodiment of the present invention, the pixel points in the target sub-block may be arranged in an order from left to right and from top to bottom, and the gradient directions of the pixel points may be organized into a texture feature sequence. Other embodiments of the present invention may also be arranged in other orders, which are not limited here.
[0067] Then, according to the difference in the overall grayscale values between any two sub-blocks, the overall grayscale difference value between any two sub-blocks is obtained. The larger the overall grayscale difference value is, the greater the difference in the grayscale features between the sub-blocks is.
[0068] In one embodiment of the present invention, the absolute value of the difference between the overall grayscale values of any two sub-blocks may be used as the overall grayscale difference value between any two sub-blocks.
[0069] Furthermore, based on the difference in gradient direction in the texture feature sequence between any two sub-blocks, the texture feature difference between any two sub-blocks is obtained. The larger the texture feature difference, the greater the difference in texture features between the sub-blocks. Subsequently, the similarity parameters between any two sub-blocks can be accurately calculated by combining the texture feature difference and the overall grayscale difference value.
[0070] Preferably, in an embodiment of the present invention, a method for obtaining the texture feature difference between any two sub-blocks specifically includes:
[0071] Since the dynamic time warping algorithm can analyze the difference between two sequences, the texture feature sequence between any two sub-blocks can be processed based on the dynamic time warping algorithm to obtain the texture feature difference between any two sub-blocks. The dynamic time warping algorithm is a technical means well known to technical personnel in this field and will not be elaborated here.
[0072] In other embodiments of the present invention, the Euclidean distance or Manhattan distance of the texture feature sequences between any two sub-blocks may also be used as the texture feature difference between any two sub-blocks, which is not limited here.
[0073] Finally, the overall grayscale difference value and texture feature difference are integrated to obtain the comprehensive difference between any two sub-blocks. The larger the comprehensive difference, the greater the difference in grayscale features and texture features between the sub-blocks, which further indicates that the similarity of features between the sub-blocks is smaller. Therefore, the comprehensive difference can be negatively correlated and normalized to obtain the similarity parameter between any two sub-blocks.
[0074] In the embodiment of the present invention, the sum or product of the overall grayscale difference value and the texture feature difference can be used as the comprehensive difference to achieve the integration of the two, which is not limited here.
[0075] In one embodiment of the present invention, the normalization processing can be specifically, for example, maximum and minimum value normalization processing, and the normalization in subsequent steps can all adopt maximum and minimum value normalization processing. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of numerical values, which will not be described in detail.
[0076] As an example, in one embodiment of the present invention, the expression of the similarity parameter between any two sub-blocks may be specifically, for example, as follows:
[0077]
[0078] in, represents the similarity parameter between any two sub-blocks; Respectively represent the overall grayscale values of any two sub-blocks; Represents the overall grayscale difference value between any two sub-blocks; Represents the difference in texture features between any two sub-blocks; Represents the comprehensive difference between any two sub-blocks; represents the normalization function; Indicates the preset first adjustment parameter, which is used to prevent the denominator from being 0. The value range is In one embodiment of the present invention, Set to 0.01, The specific value of can also be set by the implementer according to the specific implementation scenario and is not limited here.
[0079] It should be noted that, in other embodiments of the present invention, negative correlation mapping may also be achieved through other basic mathematical operations, which are not limited here.
[0080] By obtaining the similarity parameters between any two sub-blocks, the grayscale image can be segmented based on the similarity parameters between the sub-blocks, and the sub-blocks with similar grayscale and texture can be merged into the same area, thereby obtaining multiple segmented areas and accurately distinguishing different areas in the grayscale image.
[0081] Preferably, in an embodiment of the present invention, a method for acquiring multiple segmented regions specifically includes:
[0082] Based on the region growing algorithm and according to the similarity parameters between any two sub-blocks, region growing is performed on the grayscale image to obtain multiple segmented regions in the grayscale image, wherein the merging rule of region growing is that when the similarity parameter between the sub-blocks is less than the preset similarity threshold, the sub-blocks are merged, wherein the region growing algorithm is a technical means well known to those skilled in the art and will not be elaborated here.
[0083] The preset similarity threshold is set to 0.7, and the specific value of the preset similarity threshold can also be set by the implementer according to the specific implementation scenario, and is not limited here.
[0084] In other embodiments of the present invention, all sub-blocks may be clustered based on similarity parameters between any two sub-blocks using existing clustering algorithms such as the K-means algorithm or the DBSCAN algorithm to obtain multiple segmented regions, which is not limited here.
[0085] Step S3: according to the distribution of the gradient amplitude of the pixel points in each segmented area, the texture richness of each segmented area is obtained; based on the texture richness, a plurality of areas to be enhanced are screened out from all the segmented areas; any area to be enhanced is taken as the target area to be enhanced, and according to the difference in the distribution of the grayscale values of the pixel points between the target area to be enhanced and other areas to be enhanced except the target area to be enhanced, and the texture richness of the target area to be enhanced, the enhancement coefficient of the target area to be enhanced is obtained.
[0086] Since the segmented area also includes a background area that does not contain any information, the subsequent steps do not need to enhance the background area. For non-background areas, such as osteosarcoma areas and tissue bone areas, which contain a certain degree of texture information, the subsequent steps need to enhance the osteosarcoma areas and tissue bone areas, and it is necessary to ensure that the texture information in the enhanced osteosarcoma areas and tissue bone areas is more prominent in order to improve the enhancement effect.
[0087] At the same time, the richness of texture information contained in different regions is different. For example, the background area basically does not contain texture information. For the osteosarcoma area and the tissue bone area, the richness of the texture information contained is also different. The richness of texture information contained in different regions can be reflected by the distribution of gradient amplitude values of pixels in the region. Therefore, the existing edge detection operators such as the Sobel operator or the Roberts operator are first used to calculate the gradient amplitude of each pixel in the grayscale image, and then the distribution of the gradient amplitude of the pixel in each segmented area is analyzed. The obtained texture richness reflects the richness of texture information contained in the segmented area. The greater the texture richness, the more texture information is contained in the segmented area. Subsequently, the area to be enhanced, namely the osteosarcoma area and the tissue bone area, can be screened out based on the texture richness. At the same time, different areas to be enhanced can be enhanced to different degrees based on the texture richness in the subsequent process to ensure that the texture information in the area to be enhanced is more prominent.
[0088] Preferably, in one embodiment of the present invention, a method for acquiring the texture richness of each segmented region specifically includes:
[0089] First, any segmentation area is taken as the target segmentation area, and the range of the gradient amplitude of all pixels in the target segmentation area is taken as the first richness parameter of the target segmentation area. The larger the first richness parameter is, the wider the distribution range of the gradient amplitude of the pixels in the target segmentation area is, which means that the target segmentation area contains richer texture information.
[0090] Then, the overall level of the gradient amplitude of all pixels in the target segmentation area is analyzed to obtain the overall gradient amplitude of the target segmentation area; the gradient amplitude deviation of each pixel in the target segmentation area is obtained according to the difference between the gradient amplitude of each pixel in the target segmentation area and the overall gradient amplitude; the average value of the gradient amplitude deviation of all pixels in the target segmentation area is used as the second richness parameter of the target segmentation area. The larger the second richness parameter is, the lower the consistency of the gradient amplitude of each pixel in the target segmentation area is, and thus the richer the texture information contained in the target segmentation area is.
[0091] In an embodiment of the present invention, the average value or median of the gradient amplitudes of all pixels in the target segmentation area may be used as the overall gradient amplitude of the target segmentation area to implement the overall level analysis of the gradient amplitudes of all pixels in the target segmentation area, which is not limited here.
[0092] In one embodiment of the present invention, the absolute value of the difference between the gradient amplitude of each pixel in the target segmentation area and the overall gradient amplitude can be used as the gradient amplitude deviation of each pixel in the target segmentation area to analyze the difference between the two.
[0093] Then, the first richness parameter and the second richness parameter are integrated and normalized to obtain the texture richness of the target segmented area.
[0094] In one embodiment of the present invention, the integration of the first richness parameter and the second richness parameter can be achieved by calculating the sum or product of the two, which is not limited here.
[0095] As an example, in one embodiment of the present invention, the expression of the texture richness of the target segmented region may be specifically, for example, as follows:
[0096]
[0097] in, Indicates the texture richness of the target segmentation area; Represents the maximum value of the gradient amplitude of all pixels in the target segmentation area; Represents the minimum value of the gradient amplitude of all pixels in the target segmentation area; Represents the range of the gradient amplitude of all pixels in the target segmentation area, that is, the first richness parameter of the target segmentation area; Indicates the first The gradient amplitude of each pixel; Represents the overall gradient amplitude of the target segmentation area; Indicates the number of pixels in the target segmentation area; Indicates the first The gradient amplitude deviation of each pixel; A second richness parameter representing the target segmentation region; Represents the normalization function.
[0098] The texture richness of each segmented area can be obtained by the same method as mentioned above. The greater the texture richness, the richer the texture information contained in the segmented area, and thus the greater the possibility that the segmented area is a non-background area. Therefore, based on the texture richness, multiple areas to be enhanced can be screened out from all segmented areas. The areas to be enhanced include osteosarcoma areas and tissue bone areas in non-background areas. The areas to be enhanced can be enhanced subsequently to improve the contrast between the osteosarcoma area and the surrounding tissue bone area, so that the invasion boundary of the osteosarcoma area on the surrounding tissue bone area can be more clearly displayed.
[0099] Preferably, in one embodiment of the present invention, a segmented area whose texture richness is greater than a preset richness threshold is used as an area to be enhanced, wherein the preset richness threshold is set to 0.25, and the specific value of the preset richness threshold can also be set by the implementer according to the specific implementation scenario, which is not limited here.
[0100] Since there are large differences in the grayscale and texture features between different areas to be enhanced, different areas to be enhanced need to be enhanced to different degrees to ensure that the distinction between the enhanced areas to be enhanced is greater and the texture features in each area to be enhanced are more obvious. Therefore, any area to be enhanced is first taken as the target area to be enhanced, and the difference in the distribution of pixel grayscale values between the target area to be enhanced and other areas to be enhanced except the target area to be enhanced, as well as the texture richness of the target area to be enhanced are analyzed. The degree of enhancement of the target area to be enhanced is reflected by the obtained enhancement coefficient. The larger the enhancement coefficient, the greater the subsequent degree of enhancement of the target area to be enhanced.
[0101] Preferably, in one embodiment of the present invention, the method for obtaining the enhancement coefficient of the target area to be enhanced specifically includes:
[0102] See also Figure 2 , which shows a flow chart of a method for obtaining an enhancement coefficient of a target area to be enhanced provided by an embodiment of the present invention.
[0103] Step S301: Analyze the overall level of the grayscale values of all pixels in each area to be enhanced to obtain the overall grayscale parameter of each area to be enhanced.
[0104] After enhancing each area to be enhanced in the grayscale image, it is necessary to ensure that the contrast between the areas to be enhanced is high. Therefore, the overall level of the grayscale values of all pixels in each area to be enhanced is first analyzed to obtain the overall grayscale parameters of each area to be enhanced. The overall grayscale parameters are used to reflect the grayscale distribution characteristics of each area to be enhanced.
[0105] In an embodiment of the present invention, the average or median of the grayscale values of all pixels in each area to be enhanced may be used as the overall grayscale parameter of each area to be enhanced to analyze the overall level of the grayscale values of all pixels in each area to be enhanced, which is not limited here.
[0106] Step S302: obtaining the grayscale distinction between the target region to be enhanced and each other region to be enhanced according to the difference in overall grayscale parameters between the target region to be enhanced and each other region to be enhanced except the target region to be enhanced.
[0107] The more similar the grayscale features between the target region to be enhanced and the other regions to be enhanced are, the lower the contrast or distinction between the target region to be enhanced and the other regions to be enhanced is, which means that a greater degree of enhancement is required for the target region to be enhanced. Therefore, the grayscale distinction between the target region to be enhanced and each other region to be enhanced can be obtained according to the difference in overall grayscale parameters between the target region to be enhanced and each other region to be enhanced except the target region to be enhanced. The smaller the grayscale distinction is, the lower the contrast between the target region to be enhanced and each other region to be enhanced is, and the higher the degree of enhancement of the target region to be enhanced is. Subsequently, the enhancement coefficient of the target region to be enhanced can be accurately calculated based on the grayscale distinction.
[0108] In one embodiment of the present invention, the absolute value of the difference in overall grayscale parameters between the target region to be enhanced and each other region to be enhanced except the target region to be enhanced can be used as the grayscale distinction between the target region to be enhanced and each other region to be enhanced to analyze the difference between the two.
[0109] Step S303: obtaining an enhancement coefficient of the target region to be enhanced according to the grayscale distinction between the target region to be enhanced and each other region to be enhanced, and the texture richness of the target region to be enhanced.
[0110] After enhancing the target area to be enhanced, while ensuring a high contrast between the target area to be enhanced and other areas to be enhanced, it is also necessary to ensure that the texture information in the target area to be enhanced is more prominent to improve the enhancement effect. Therefore, the enhancement coefficient of the target area to be enhanced can be obtained based on the grayscale distinction between the target area to be enhanced and each other area to be enhanced, combined with the texture richness of the target area to be enhanced.
[0111] Preferably, in one embodiment of the present invention, the method for obtaining the enhancement coefficient of the target area to be enhanced further includes:
[0112] A negative correlation mapping is performed on the average value of the grayscale distinction between the target area to be enhanced and all other areas to be enhanced to obtain the degree of enhancement of the target area to be enhanced. The greater the texture richness of the target area to be enhanced, the greater the degree of enhancement of the target area to be enhanced should be. Therefore, the texture richness of the target area to be enhanced can be used to weight the degree of enhancement to obtain the adjustment amount of the target area to be enhanced, and the sum of the degree of enhancement and the adjustment amount is normalized to obtain the enhancement coefficient of the target area to be enhanced.
[0113] As an example, in one embodiment of the present invention, the expression of the enhancement coefficient of the target area to be enhanced may be specifically, for example, as follows:
[0114]
[0115] in, Indicates the enhancement coefficient of the target area to be enhanced; Indicates the enhancement degree of the target area to be enhanced; Indicates the texture richness of the target area to be enhanced; Represents the overall grayscale parameter of the target area to be enhanced; Indicates the area other than the target area to be enhanced. The overall grayscale parameters of other areas to be enhanced; Indicates the number of other areas to be enhanced except the target area to be enhanced; Indicates the preset second adjustment parameter, which is used to prevent the denominator from being 0. The value range is In one embodiment of the present invention, Set to 0.01, The specific value of can also be set by the implementer according to the specific implementation scenario and is not limited here.
[0116] The enhancement coefficient of each area to be enhanced can be obtained by the same method as above.
[0117] Step S4: According to the distribution of the grayscale values of the pixels in each area to be enhanced in the grayscale image and the enhancement coefficient, the grayscale values of the pixels in the area to be enhanced are adjusted to obtain an enhanced image.
[0118] The enhancement coefficient reflects the degree of enhancement of each area to be enhanced. The larger the enhancement coefficient, the greater the degree of enhancement of each area to be enhanced. At the same time, in the specific enhancement process, it is necessary to increase or decrease the pixel grayscale value based on the distribution of the pixel grayscale value of each area to be enhanced. While enhancing the contrast between the areas to be enhanced, the texture features in each area to be enhanced are highlighted to improve the final enhancement effect. Therefore, according to the distribution of the pixel grayscale value of each area to be enhanced in the grayscale image and the enhancement coefficient, the pixel grayscale value in the area to be enhanced can be adjusted to obtain an enhanced image, improve the contrast between the osteosarcoma area and the surrounding tissue and bone areas, and retain the texture details in each area, so that the invasion boundary of the osteosarcoma in the enhanced image is displayed more clearly.
[0119] Preferably, in one embodiment of the present invention, the method for obtaining the enhancement coefficient of the target area to be enhanced further includes:
[0120] For any area to be enhanced in the grayscale image, all pixels in the area to be enhanced are divided into two categories based on the grayscale value, and the category with the largest average grayscale value of pixels is taken as the high grayscale category, and the category with the smallest average grayscale value of pixels is taken as the low grayscale category. Subsequently, the grayscale value of the pixel can be increased or decreased according to the category to which the pixel in the area to be enhanced belongs, so as to highlight the texture features in the area to be enhanced.
[0121] Preferably, in one embodiment of the present invention, the existing K-means clustering algorithm can be used, and all pixel points in the enhanced area can be clustered based on the grayscale value to obtain two clusters, and the pixel points in one of the clusters are regarded as one class, and the pixel points in the other cluster are regarded as another class.
[0122] In other embodiments of the present invention, the Otsu threshold segmentation algorithm can also be used to calculate the optimal segmentation threshold of the grayscale value of the pixel points in the area to be enhanced, and the pixel points in the area to be enhanced that are less than the optimal segmentation threshold are classified as one category, and the pixel points that are not less than the optimal segmentation threshold are classified as another category, which is not limited here.
[0123] The product of the grayscale value of the pixel in the area to be enhanced and the enhancement coefficient of the area to be enhanced is used as the grayscale adjustment amount of the pixel in the area to be enhanced. If the pixel in the area to be enhanced belongs to the high grayscale category, the sum of the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel; if the pixel in the area to be enhanced belongs to the low grayscale category, the difference between the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel. It should be noted that during the calculation process, if the adjusted grayscale value of the pixel exceeds 255, the adjusted grayscale value is set to 255, and then the grayscale value of the pixel in the area to be enhanced in the grayscale image is replaced by the adjusted grayscale value to obtain an enhanced image. Please refer to Figure 4 , which shows a schematic diagram of an enhanced image provided by an embodiment of the present invention.
[0124] As an example, in one embodiment of the present invention, the expression of the adjusted grayscale value of a pixel point may be specifically, for example, as follows:
[0125]
[0126] in, Indicates In the area to be enhanced Adjust the gray value of each pixel; Indicates In the area to be enhanced The gray value of each pixel; Indicates The enhancement coefficient of the area to be enhanced; Indicates In the area to be enhanced Grayscale adjustment amount for each pixel; Indicates In the area to be enhanced pixels; Indicates Low gray level categories of the area to be enhanced; Indicates The high grayscale category of the area to be enhanced.
[0127] The contrast between the osteosarcoma area and the surrounding tissue and bone area in the enhanced image is significantly enhanced, and the texture features in each area are more obvious, which makes it easier for medical staff to observe the invasion of osteosarcoma on surrounding tissues more clearly.
[0128] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0129] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
Claims
1. A method for detecting invasion boundaries before chemotherapy for osteosarcoma in children based on MRI scanning, characterized in that: The method comprises: A grayscale image of an osteosarcoma at a target site of a pediatric patient scanned by MRI is obtained; The grayscale image is uniformly divided into blocks to obtain a plurality of sub-blocks in the grayscale image, and a similarity parameter between any two sub-blocks is obtained according to a difference in grayscale values of pixels and a difference in gradient directions of pixels between any two sub-blocks; based on the similarity parameter between any two sub-blocks, the grayscale image is segmented to obtain a plurality of segmented regions; According to the distribution of the gradient amplitude of the pixel points in each segmented area, the texture richness of each segmented area is obtained; based on the texture richness, a plurality of areas to be enhanced are screened out from all the segmented areas; any area to be enhanced is taken as a target area to be enhanced, and according to the difference in the distribution of the grayscale values of the pixel points between the target area to be enhanced and other areas to be enhanced except the target area to be enhanced, and the texture richness of the target area to be enhanced, an enhancement coefficient of the target area to be enhanced is obtained; According to the distribution of the grayscale values of the pixels in each area to be enhanced in the grayscale image and the enhancement coefficient, the grayscale values of the pixels in the area to be enhanced are adjusted to obtain an enhanced image.
2. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The obtaining of similarity parameters between any two sub-blocks comprises: Take any sub-block in the grayscale image as the target sub-block; Analyze the overall level of the grayscale values of all pixels in the target sub-block to obtain the overall grayscale value of the target sub-block; The sequence composed of the gradient directions of all pixels in the target sub-block is used as the texture feature sequence of the target sub-block; According to the difference of the overall grayscale values between any two sub-blocks, obtaining the overall grayscale difference value between any two sub-blocks; Obtaining a texture feature difference between any two sub-blocks according to a difference in gradient direction in the texture feature sequence between any two sub-blocks; The overall grayscale difference value and the texture feature difference are integrated to obtain a comprehensive difference between any two sub-blocks, and the comprehensive difference is negatively correlated and normalized to obtain a similarity parameter between any two sub-blocks.
3. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 2, characterized in that: The obtaining the texture feature difference between any two sub-blocks according to the difference in the gradient direction in the texture feature sequence between any two sub-blocks comprises: Based on the dynamic time warping algorithm, the texture feature sequence between any two sub-blocks is processed to obtain the texture feature difference between any two sub-blocks.
4. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The obtaining of a plurality of segmented regions comprises: Based on the region growing algorithm and according to the similarity parameter between any two sub-blocks, region growing is performed on the grayscale image to obtain multiple segmented regions in the grayscale image, wherein the merging rule of region growing is that when the similarity parameter between the sub-blocks is less than a preset similarity threshold, merging is performed.
5. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The obtaining of the texture richness of each segmented region comprises: Take any segmentation area as the target segmentation area, and take the extreme difference of the gradient amplitude of all pixels in the target segmentation area as the first richness parameter of the target segmentation area; Analyze the overall level of the gradient amplitude of all pixels in the target segmentation area to obtain the overall gradient amplitude of the target segmentation area; Obtaining a gradient amplitude deviation of each pixel point in the target segmentation area according to a difference between the gradient amplitude of each pixel point in the target segmentation area and the overall gradient amplitude; The average value of the gradient amplitude deviation of all pixels in the target segmentation area is used as the second richness parameter of the target segmentation area; The first richness parameter and the second richness parameter are integrated and normalized to obtain the texture richness of the target segmented area.
6. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The step of selecting a plurality of areas to be enhanced from all segmented areas based on the texture richness includes: The segmented area whose texture richness is greater than a preset richness threshold is used as the area to be enhanced.
7. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The step of obtaining the enhancement coefficient of the target area to be enhanced comprises: Analyze the overall level of the grayscale values of all pixels in each area to be enhanced to obtain the overall grayscale parameter of each area to be enhanced; Obtaining a grayscale distinction between the target region to be enhanced and each other region to be enhanced according to a difference in the overall grayscale parameter between the target region to be enhanced and each other region to be enhanced except the target region to be enhanced; An enhancement coefficient of the target region to be enhanced is obtained according to the grayscale distinction between the target region to be enhanced and each other region to be enhanced, and the texture richness of the target region to be enhanced.
8. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 7, characterized in that: The step of obtaining an enhancement coefficient of the target region to be enhanced according to the grayscale distinction between the target region to be enhanced and each other region to be enhanced, and the texture richness of the target region to be enhanced comprises: Performing negative correlation mapping on the average values of the grayscale distinctions between the target region to be enhanced and all other regions to be enhanced to obtain the enhancement degree of the target region to be enhanced; Using the texture richness of the target area to be enhanced, weighting the enhancement degree to obtain an adjustment amount of the target area to be enhanced; The sum of the enhancement degree and the adjustment amount is normalized to obtain an enhancement coefficient of the target area to be enhanced.
9. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 1, characterized in that: The obtaining of the enhanced image comprises: For any area to be enhanced in the grayscale image, all pixels in the area to be enhanced are divided into two categories based on the grayscale value, and the category with the largest average grayscale value of the pixels is taken as the high grayscale category, and the category with the smallest average grayscale value of the pixels is taken as the low grayscale category; Taking the product value of the grayscale value of the pixel point in the area to be enhanced and the enhancement coefficient of the area to be enhanced as the grayscale adjustment amount of the pixel point in the area to be enhanced; If the pixel in the area to be enhanced belongs to the high grayscale category, the sum of the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel; if the pixel in the area to be enhanced belongs to the low grayscale category, the difference between the grayscale value of the pixel and the grayscale adjustment amount is used as the adjusted grayscale value of the pixel; The grayscale values of the pixels in the area to be enhanced in the grayscale image are replaced with the adjusted grayscale values to obtain an enhanced image.
10. The method for detecting invasion boundaries of children's osteosarcoma before chemotherapy based on MRI scanning according to claim 9, characterized in that: The method of dividing all pixels in the area to be enhanced into two categories based on the grayscale value includes: Based on the gray value, all the pixels in the enhanced area are clustered to obtain two clusters. The pixels in one cluster are regarded as one class, and the pixels in the other cluster are regarded as another class.
Citation Information
Patent Citations
Method for detecting tumor tissue boundary or tumor matrix cell distribution range
CN107661103A
Nuclear magnetic resonance tumor visual detection method based on image denoising
CN117764864A
Ultrasonic tumor examination system for obstetrics and gynecology department
CN118587239A
Methods and Systems for Determining Tumor Boundary Characteristics
US20140148679A1
Method for detecting tumor tissue boundaries or tumor stromal cell distribution range
WO2018019038A1
