High-definition medical image processing method

Through guided filtering algorithm, potential secretion index screening, edge detection and Fourier transform analysis, the problem of inaccurate enhancement of lesion areas in traditional methods is solved, and the precise enhancement and characteristic evaluation of lesion areas in high-definition medical imaging processing is achieved.

CN119090764AActive Publication Date: 2024-12-06CHENGDU SHENGAN MEDICAL TECH CO LTD
View PDF 16 Cites 0 Cited by

Patent Information

Application Number
CN202411089205.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-12-06
Estimated Expiration
2044-08-09

AI Technical Summary

Technical Problem

Traditional high-definition medical image processing methods lack targeting and cannot effectively analyze the features in the image, find out the lesion area and extract multiple features, resulting in inaccurate enhancement of the image of the lesion area.

Method used

The guided filtering algorithm is used to perform noise reduction processing, calculate the potential secretion index, and use the threshold segmentation method to screen the pixel points, combine the edge detection algorithm to connect the pixel points to form the lesion area image, and analyze the frequency components through Fourier transform to obtain The ambiguity index and enhancement coefficient achieve the enhancement of the lesion area.

Benefits of technology

The boundary determination accuracy and fuzzy characteristics evaluation of the lesion area are improved, effective enhancement of the lesion area images is achieved, and the accuracy and efficiency of medical image analysis are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090764A_ABST
    Figure CN119090764A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of image processing, in particular to a high-definition medical image processing method, which comprises the following steps of: performing noise reduction processing on an ileocecal part image through a guide filtering algorithm to obtain an ileocecal part image histogram, and calculating a potential secretion index of each pixel point in the ileocecal part image according to the ileocecal part histogram; the method comprises the following steps: setting a segmentation threshold to extract pixel points exceeding a potential secretion index, connecting the pixel points, highlighting a lesion area of an ileocecum, extracting edge features of an image of the lesion area by using an edge detection algorithm, obtaining the resolution of the edge pixel points, analyzing by using Fourier transform to obtain a fuzzy degree coefficient of the edge pixel points, and finally obtaining the fuzzy degree coefficient of the edge pixel points. And obtaining an enhancement coefficient of the lesion area image according to the resolution and the fuzzy degree coefficient of the edge pixel points, and realizing enhancement of the lesion area image of the ileocecum by adjusting the enhancement coefficient of the lesion area image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a high-definition medical image processing method. Background Art

[0002] Image processing is an important technology. With the continuous advancement of image processing technology, especially the development of deep learning algorithms, it can more accurately detect and identify tiny lesions and early lesions, improve the accuracy of disease diagnosis, and make detailed analysis of the patient's medical images to develop personalized treatment plans for each patient and improve the treatment effect. Image processing is used to analyze a large amount of medical image data, explore potential disease patterns and risk factors, and achieve early prediction and prevention of diseases. When processing medical images, traditional high-definition medical image processing methods are relatively simple and not targeted. They do not find the lesion area by analyzing the features in the image, and extract multivariate features of the lesion area. By analyzing the multivariate features of the lesion area image, the enhancement coefficient of the lesion area image is obtained, thereby achieving enhancement of the lesion area image. In order to solve this technical problem, we provide a high-definition medical image processing method. Summary of the invention

[0003] The object of the present invention is to provide a high-definition medical image processing method to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides a high-definition medical image processing method, comprising the following steps:

[0005] S1, using an image acquisition device to obtain a medical image of the ileocecal region, marking it as an ileocecal region image, and performing noise reduction processing on the ileocecal region image through a guided filtering algorithm, extracting pixel values ​​of the ileocecal region image after the noise reduction processing, and equalizing the pixel values ​​of the ileocecal region image to obtain an ileocecal region image histogram;

[0006] S2, calculating the potential secretion index of each pixel in the ileocecal image according to the ileocecal histogram, setting a segmentation threshold, using the threshold segmentation method to compare and classify the potential secretion index of each pixel in the ileocecal image with the segmentation threshold, and obtaining the pixels in the ileocecal image that exceed the segmentation threshold to form a set of suspected lesion pixels;

[0007] S3, extracting pixel points from the set of suspected lesion pixel points as pixel points of the lesion area, and connecting the pixel points of the lesion area using an edge detection algorithm to form an image of the lesion area, extracting edge features of the lesion area image, and obtaining the resolution of edge pixel points according to the edge features of the lesion area image;

[0008] S4, using Fourier transform to analyze the frequency components of the lesion area image, obtain a blur index of the lesion area image, and obtain a blur degree coefficient of the edge pixel point according to the blur index;

[0009] S5. deriving an enhancement coefficient of the lesion region image according to the resolution and blur coefficient of the edge pixel points, and enhancing the ileocecal lesion region image by adjusting the enhancement coefficient of the lesion region image.

[0010] As a further improvement of the technical solution, the step S1 performs noise reduction processing on the ileocecal image by using a guided filtering algorithm, specifically including:

[0011] A random image is selected from the ileocecal image as the guide image, denoted as A, and the ileocecal image to be denoised is used as the input image, denoted as B. The window radius of the guide filter is set to r, the regularization parameter is α, and for each pixel c in the ileocecal image to be denoised, a window D is established with it as the center. e , in window D e In the example, the output of the guided filter F c The calculation is done by the following formula, namely:

[0012] F c =a e *A c +b e

[0013] Among them, a e and b e is calculated by linear regression, A c represents the pixel value of the guidance image, F c Represents the ileocecal image after guided filtering denoising.

[0014] As a further improvement of this solution, the potential secretion index of the ileocecal region is calculated according to the ileocecal region histogram in S2, which specifically includes:

[0015] Suppose the ileocecal image F after noise reduction is c The intensity value range of (x, y) is [0, L-1], and its size is M*N, where L represents the total number of color depths in the ileocecal image. The intensity histogram H(k) is calculated using the following formula, namely:

[0016]

[0017] in, Represented as image F c The gray value at coordinate (x, y), Expressed as Kronecker function, 0≤k≤L-1, the Kronecker function is defined as:

[0018] The skewness and kurtosis of the intensity histogram H(k) are extracted, and a linear model is established based on the skewness and kurtosis of the intensity histogram H(k) to calculate the potential secretion index γ of each pixel. The expression of the potential secretion index γ of each pixel is:

[0019] γ=ω 1 *PD H(k) +ω 2 *FD H(k)

[0020] Among them, ω 1 and ω 2 The weight coefficient of the potential secretion index γ affecting each pixel point is obtained through experimental data. PD is expressed as the skewness of the intensity histogram H(k), and FD is expressed as the kurtosis of the intensity histogram H(k).

[0021] As a further improvement of the technical solution, the potential secretion index of each pixel in the ileocecal image is compared and classified with the segmentation threshold in S2, as follows:

[0022] The secretion index threshold is set, the potential secretion index of the lesion is higher than that of the surrounding area, and the potential secretion index of each pixel in the ileocecal image is compared with the secretion index threshold, the pixels in the ileocecal image whose potential secretion index is higher than the segmentation threshold are screened out, and the pixels in the ileocecal image that exceed the segmentation threshold are grouped into a set of suspected lesion pixels.

[0023] As a further improvement of the technical solution, the pixel points of the lesion area are connected in S3 to form an image of the lesion area, which specifically includes:

[0024] The edge detection algorithm is used to calculate the gradient amplitude and direction of each pixel in the lesion area, non-maximum suppression is performed in the gradient direction, and high and low thresholds are set. According to the high and low thresholds, the gradient amplitude is divided into strong edge and weak edge categories, and the strong edge pixels are tracked. The weak edge pixels adjacent to the strong edge pixels are divided into edge pixels and connected to form an image of the lesion area.

[0025] As a further improvement of the technical solution, the step of obtaining the resolution of edge pixels according to the edge features of the lesion area image in S3 specifically includes:

[0026] Extract the coordinates of all edge pixels in the lesion area image (x i ,y i), i is the index of the edge pixel, the actual distance represented by each pixel in the horizontal direction is set to Δx, the actual distance represented by each pixel in the vertical direction is set to Δy, and the coordinates of the edge pixel points are converted from pixel units to actual space units, that is:

[0027] X i =x i *Δx

[0028] Y i =y i *Δy

[0029] By calculating the distance between edge pixels and analyzing edge features, the edge pixel (X i , Y i ) and (X j , Y j ) is:

[0030]

[0031] The resolution of the edge pixels is obtained by calculating the minimum distance between the edge pixels. The calculation formula for the resolution of the edge pixels is:

[0032]

[0033] Among them, min(d ij ) represents the minimum distance between all edge pixels.

[0034] As a further improvement of the technical solution, the frequency components of the lesion area image are analyzed by Fourier transform in S4 to obtain the blur index of the lesion area image, which specifically includes:

[0035] The image of the lesion area is converted into a grayscale image, and the grayscale image is smoothed by Gaussian filtering. The grayscale image is defined as h(x', y'), and a two-dimensional fast Fourier transform is applied to the grayscale image, namely:

[0036] f(u,v)=f'{h(x',y')}

[0037] Among them, f(u, v) is the frequency domain representation, (u, v) is the frequency coordinate, and the spectrum amplitude S(u, v) is calculated according to f(u, v). The spectrum amplitude S(u, v) represents the change intensity of each frequency component (u, v) in the frequency domain, reflecting the energy contribution of these frequencies in the original image. The expression of the spectrum amplitude S(u, v) is:

[0038] S(u,v)=|f(u,v)|

[0039] According to the spectrum amplitude, the ambiguity index is defined by using the energy ratio of the high-frequency component to the low-frequency component. The expression of the ambiguity index is:

[0040]

[0041] Among them, W represents the index set of high-frequency components, G represents the index set of low-frequency components, and the index set of high-frequency components represents the information of details, edges and noise in the image. The blur coefficient of edge pixels is defined as: R=1-Q. If the value of R is close to 1, it indicates that the pixel is clear, otherwise, it means that the pixel is blurred.

[0042] As a further improvement of the technical solution, the enhancement coefficient of the lesion area image is obtained according to the resolution and blur coefficient of the edge pixel points in S5, which specifically includes:

[0043] The enhancement coefficient T is set to be inversely proportional to the blur coefficient R of the edge pixel point and proportional to the resolution of the edge pixel point. The expression of the enhancement coefficient T is:

[0044] T=t*(1-R)*δ

[0045] Among them, t is a proportional constant used to control the degree of enhancement. The pixel value of the lesion area image is adjusted using the enhancement coefficient T. The lesion area image is taken at position (x i ,y i )’s pixel value P(x i ,y i ), by calculating the pixel value and the background pixel value P'(x i ,y i The product of the difference between the two values ​​and the enhancement coefficient T is used to obtain the enhanced part, and the enhanced part is added back to the original pixel value P(x i ,y i ) to get the enhanced pixel value P ZQ (x i ,y i ), the enhanced pixel value P ZQ (x i ,y i ) is:

[0046] P ZQ (x i ,y i )=P(x i ,y i )+T*[(P(x i ,y i ))-P'(x i ,y i )]

[0047] The image details and brightness are enhanced by increasing the contrast with the background, thereby enhancing the image of the ileocecal lesion area.

[0048] A second object of the present invention is to provide a system for implementing a high-definition medical image processing method according to any one of claims 1 to 7, characterized in that it comprises:

[0049] The acquisition and processing unit is used to acquire the medical image of the ileocecal region, and pre-process the ileocecal region image by a guided filtering algorithm to obtain an ileocecal region image histogram;

[0050] The lesion analysis unit includes a calculation analysis module and a lesion generation module;

[0051] The calculation and analysis module is used to calculate the potential secretion of each pixel point according to the ileocecal image histogram, set the secretion index threshold, and use the threshold segmentation method to perform comparison and classification to form a set of lesion pixel points;

[0052] The lesion generation module connects the pixel points of the lesion area using an edge detection algorithm to form an image of the lesion area, and obtains the resolution of the edge pixel points according to the edge features of the lesion area image;

[0053] The fuzzy evaluation unit uses Fourier transform to analyze the frequency components of the lesion area image, obtains the fuzziness index of the lesion area image, and obtains the fuzziness coefficient of the edge pixel point according to the fuzziness index;

[0054] The enhancement determination unit obtains the enhancement coefficient of the lesion area image according to the resolution and blur coefficient of the edge pixel points, and enhances the ileocecal lesion area image by adjusting the enhancement coefficient of the lesion area image.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] In a high-definition medical image processing method, a guided filtering algorithm is used to perform noise reduction processing on an ileocecal image to obtain an ileocecal image histogram, which can better retain the edge information of the image and help to clearly present the anatomical structure of the ileocecal region after noise reduction. Then, a threshold segmentation method is used to compare and classify the potential secretion index of each pixel in the ileocecal image with the segmentation threshold, and the pixels in the ileocecal image that exceed the segmentation threshold are obtained, thereby reducing the amount of data for subsequent processing and improving processing efficiency. The resolution of edge pixels is obtained according to the edge features of the lesion area image, which can more accurately determine the boundary of the lesion area and help to accurately evaluate the size of the lesion. The frequency components of the lesion area image are analyzed by Fourier transform to obtain a blur index of the lesion area image, and the blur degree coefficient of the edge pixel is obtained according to the blur index, thereby comprehensively evaluating the overall blur characteristics of the lesion area, rather than just being limited to local pixels. The enhancement coefficient of the lesion area image is obtained according to the resolution and blur degree coefficient of the edge pixel, and the enhancement of the lesion area image is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is the overall workflow diagram of the present invention;

[0058] Figure 2 It is a schematic diagram of the overall structure of the present invention;

[0059] The meaning of each number in the figure is:

[0060] 1. Acquisition and processing unit; 2. Lesion analysis unit; 21. Calculation and analysis module; 22. Lesion generation module; 3. Fuzzy evaluation unit; 4. Enhanced determination unit. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] Example 1

[0063] See also Figure 1-Figure 2 As shown, this embodiment provides a high-definition medical image processing method, comprising the following steps:

[0064] S1. Acquire a medical image of the ileocecal region by using an image acquisition device, mark it as an ileocecal region image, perform noise reduction processing on the ileocecal region image by using a guided filtering algorithm, extract pixel values ​​of the ileocecal region image after the noise reduction processing, and equalize the pixel values ​​of the ileocecal region image to obtain an ileocecal region image histogram, improve the visibility and details of the ileocecal region image, enhance the contrast of the image, and make the hidden structure in the ileocecal region image more clearly visible;

[0065] In S1, the ileocecal image is subjected to noise reduction processing by using a guided filtering algorithm, specifically including:

[0066] A random image is selected from the ileocecal image as the guide image, denoted as A, and the ileocecal image to be denoised is used as the input image, denoted as B. The window radius of the guide filter is set to r, the regularization parameter is α, and for each pixel c in the ileocecal image to be denoised, a window D is established with it as the center. e , in window D e In the example, the output of the guided filter F c The calculation is done by the following formula, namely:

[0067] F c =a e *A c +b e

[0068] Among them, a e and b e is calculated by linear regression, A c represents the pixel value of the guidance image, F c Represents the ileocecal image after guided filtering denoising.

[0069]

[0070] b e =B'-a e *A'

[0071] Among them, J represents the number of pixels in window J centered on pixel c, which is used to average the calculation results in the window to obtain representative statistics, A' represents the mean of the guide image A in window J, reflecting the average brightness or intensity level of the guide image in the window, and B' represents the mean of the input image B in window J. It is expressed as the variance of the guide image A in the window J, which is used to measure the discrete degree or distribution width of the guide image pixel value in the window. e It is expressed as a coefficient calculated from the statistical information within the window, which is used to determine the linear relationship between the output image and the guidance image, b e is represented as another coefficient calculated from the statistics within the window, and is related to a eTogether they determine the final value of the output image. α is represented as a regularization parameter, which is used to prevent instability caused by a denominator of zero or a value that is too small. It also controls the filtering effect to avoid over-smoothing or over-fitting.

[0072] S2, calculating the potential secretion index of each pixel in the ileocecal image according to the ileocecal histogram, setting a segmentation threshold, using the threshold segmentation method to compare and classify the potential secretion index of each pixel in the ileocecal image with the segmentation threshold, and obtaining the pixels in the ileocecal image that exceed the segmentation threshold to form a set of suspected lesion pixels;

[0073] In S2, the potential secretion index of the ileocecal region is calculated based on the ileocecal region histogram, including:

[0074] Suppose the ileocecal image F after noise reduction is c The intensity value range of (x, y) is [0, L-1], and its size is M*N. L represents the total number of color depths in the ileocecal image. The intensity histogram H(k) is calculated using the following formula, namely:

[0075]

[0076] in, Represented as image F c The gray value at coordinate (x, y), It is expressed as Kronecker function, 0≤k≤L-1, and Kronecker function is defined as:

[0077] Because the ileocecal secretions usually have obvious differences in intensity from the surrounding tissues, the intensity of the potential secretions in the image usually has a certain contrast with the surrounding tissues. By analyzing the histogram, we can determine which intensity values ​​represent the potential secretions, extract the skewness and kurtosis of the intensity histogram H(k), and establish a linear model based on the skewness and kurtosis of the intensity histogram H(k) to calculate the potential secretion index γ of each pixel. The expression of the potential secretion index γ of each pixel is:

[0078] γ=ω 1 *PD H(k) +ω 2 *FD H(k)

[0079] Among them, ω 1 and ω 2The weight coefficient of the potential secretion index γ affecting each pixel point is obtained through experimental data. PD is expressed as the skewness of the intensity histogram H(k), and FD is expressed as the kurtosis of the intensity histogram H(k). Calculating the potential secretion index of each pixel point based on the intensity histogram can provide a solid foundation for subsequent medical image analysis, thereby improving the accuracy and reliability of diagnosis.

[0080] In S2, the potential secretion index of each pixel in the ileocecal image is compared with the segmentation threshold and classified as follows:

[0081] The ileocecal part of the human intestine produces secretions to protect the inner wall of the intestine from stimulation and damage. When a lesion occurs in the ileocecal part, the human body protects itself by producing a large amount of secretions. The large amount of secretions produced at the ileocecal lesion will spread in the intestine, resulting in a difference in the potential secretion index. The potential secretion index at the lesion is higher than that of the surrounding area. A secretion index threshold is set, and the potential secretion index at the lesion is higher than that of the surrounding area. The potential secretion index of each pixel in the ileocecal image is compared with the secretion index threshold, and the pixels in the ileocecal image whose potential secretion index is higher than the segmentation threshold are screened out. The pixels in the ileocecal image that exceed the segmentation threshold are grouped into a set of suspected lesion pixels, so as to efficiently and accurately identify and screen out potential ileocecal lesion areas, which not only improves the efficiency of medical image analysis.

[0082] S3, extracting pixel points from the set of suspected lesion pixel points as pixel points of the lesion area, and connecting the pixel points of the lesion area using an edge detection algorithm to form an image of the lesion area, extracting edge features of the lesion area image, and obtaining the resolution of edge pixel points according to the edge features of the lesion area image;

[0083] In S3, the pixel points of the lesion area are connected to form an image of the lesion area, which specifically includes:

[0084] The edge detection algorithm is used to calculate the gradient amplitude and direction of each pixel in the lesion area, non-maximum suppression is performed in the gradient direction, and high and low thresholds are set. According to the high and low thresholds, the gradient amplitude is divided into strong edge and weak edge categories, and the strong edge pixels are tracked. The weak edge pixels adjacent to the strong edge pixels are divided into edge pixels and connected to form an image of the lesion area.

[0085] In S3, the resolution of edge pixels is obtained according to the edge features of the lesion area image, specifically including:

[0086] The edge pixel points represent the boundary between the lesion area and the background in the image. By obtaining the coordinates of the edge pixel points, the specific boundary of the suspected lesion can be accurately located and defined. Therefore, the coordinates of all edge pixel points in the lesion area image (x i ,y i ), i is the index of the edge pixel. In medical image analysis, the measurement of real space units can provide more intuitive and understandable results. Doctors use real units to evaluate the size and location of lesions. Therefore, converting pixel units to real space units makes the interpretation of the results clearer. Therefore, the actual distance represented by each pixel in the horizontal direction is set to Δx, and the actual distance represented by each pixel in the vertical direction is Δy. The coordinates of the edge pixel points are converted from pixel units to real space units, that is:

[0087] X i =x i *Δx

[0088] Y i =y i *Δy

[0089] By calculating the distance between edge pixels and analyzing edge features, the edge pixel (X i , Y i ) and (X j , Y j ) is:

[0090]

[0091] By calculating the minimum distance between edge pixels, the resolution of edge pixels is obtained to help determine the clarity of details in the image. The calculation formula for the resolution of edge pixels is:

[0092]

[0093] Among them, min(d ij ) represents the minimum distance between all edge pixels.

[0094] S4. Analyze the frequency components of the lesion area image using Fourier transform to obtain a blur index of the lesion area image, and obtain a blur coefficient of the edge pixel point based on the blur index.

[0095] In S4, the frequency components of the image of the lesion area are analyzed by Fourier transform to obtain the blur index of the image of the lesion area, which specifically includes:

[0096] Different frequency components in the image correspond to different features. For example, high-frequency components usually represent edges and details, while low-frequency components represent general shapes and backgrounds. Through Fourier transform, we can better understand the frequency characteristics of the image, and the blur of the lesion area can be quantified by analyzing its frequency components. Blurred synthetic images usually lose high-frequency information. Therefore, the blur index can be extracted through frequency domain analysis to assist in understanding the nature of the lesion. Therefore, the image of the lesion area is converted into a grayscale image, and the grayscale image is smoothed by Gaussian filtering. The grayscale image is defined as h(x', y'). A two-dimensional fast Fourier transform is applied to the grayscale image, that is:

[0097] f(u,v)=f'{h(x',y')}

[0098] Among them, f(u, v) is the frequency domain representation, (u, v) is the frequency coordinate, and the spectrum amplitude S(u, v) is calculated according to f(u, v). The spectrum amplitude S9u, v) represents the change intensity of each frequency component (u, v) in the frequency domain, reflecting the energy contribution of these frequencies in the original image. The expression of the spectrum amplitude S(u, v) is:

[0099] S(u,v)=|f(u,v)|

[0100] According to the spectrum amplitude, the ambiguity index is defined by the energy ratio of the high-frequency component to the low-frequency component. The expression of the ambiguity index is:

[0101]

[0102] Among them, W represents the index set of high-frequency components, G represents the index set of low-frequency components, the index set of high-frequency components represents the information of details, edges and noise in the image; the index set of low-frequency components represents the change information of the image's illumination and color, and the blur index Q reflects the relative energy relationship between low-frequency and high-frequency components. When the Q value is small, it means that the high-frequency component is relatively strong and the image is clear.

[0103] The blur coefficient of the edge pixel is defined as: R = 1-Q. If the value of R is close to 1, it indicates that the pixel is clear, otherwise, it indicates that the pixel is blurred.

[0104] S5. deriving an enhancement coefficient of the lesion region image according to the resolution and blur coefficient of the edge pixel points, and enhancing the ileocecal lesion region image by adjusting the enhancement coefficient of the lesion region image.

[0105] In S5, the enhancement coefficient of the lesion area image is obtained according to the resolution and blur coefficient of the edge pixel points, which specifically includes:

[0106] The enhancement coefficient T is set to be inversely proportional to the blur coefficient R of the edge pixel points and proportional to the resolution of the edge pixel points. The expression of the enhancement coefficient T is:

[0107] T=t*(1-R)*δ

[0108] Among them, t is a proportional constant used to control the degree of enhancement. The pixel value of the lesion area image is adjusted using the enhancement coefficient T. The lesion area image is taken at position (x i ,y i )’s pixel value P(x i ,y i ), by calculating the pixel value and the background pixel value P'(x i ,y i ) is multiplied by the product of the difference between the two values ​​and the enhancement coefficient T to obtain the enhanced part. The background pixel value is obtained by the field average method, which is a prior art and will not be described here. The enhanced part is added back to the original pixel value P(x i ,y i ) to get the enhanced pixel value P ZQ (x i ,y i ), the enhanced pixel value P ZQ (x i ,y i ) is:

[0109] P ZQ (x i ,y i )=P(x i ,y i )+T*[(P(x i ,y i ))-P'(x i ,y i )]

[0110] By increasing the contrast with the background, the details and brightness of the image are enhanced, thereby highlighting the lesion area and enhancing the image of the ileocecal lesion area.

[0111] The present invention uses a guided filtering algorithm to reduce the noise of an ileocecal image while effectively retaining the edge and details of the image, and extracts pixel values ​​of the ileocecal image that has completed the noise reduction process, obtains an ileocecal image histogram by equalizing the pixel values ​​of the ileocecal image, enhances the contrast, and calculates the potential secretion index of each pixel in the ileocecal image according to the ileocecal image histogram, compares and classifies the potential secretion index of each pixel according to a threshold segmentation method to form a set of suspected lesion pixel points, connects the pixel points of the lesion area to form an image of the lesion area, and uses an edge detection method to extract the edge features of the lesion area image to obtain the resolution of the edge pixel points, then uses Fourier transform to analyze the blur index of the lesion area image, calculates the blur degree coefficient of the edge pixel points according to the blur index, obtains the enhancement coefficient of the lesion area image by the resolution of the edge pixel points and the blur degree coefficient, and enhances the lesion area image of the ileocecal region by adjusting the enhancement coefficient of the lesion area image.

[0112] Example 2

[0113] A second object of the present invention is to provide a system for implementing any one of the above-mentioned high-definition medical image processing methods, comprising:

[0114] The acquisition and processing unit 1 is used to acquire a medical image of the ileocecal region, and preprocess the ileocecal region image by a guided filtering algorithm to obtain an ileocecal region image histogram;

[0115] The lesion analysis unit 2 includes a calculation and analysis module 21 and a lesion generation module 22;

[0116] The calculation and analysis module 21 is used to calculate the potential secretion of each pixel point according to the ileocecal image histogram, set the secretion index threshold, and use the threshold segmentation method to perform comparison and classification to form a set of pathological pixel points;

[0117] The lesion generation module 22 connects the pixels of the lesion area using an edge detection algorithm to form an image of the lesion area, and obtains the resolution of the edge pixels according to the edge features of the lesion area image;

[0118] The fuzzy evaluation unit 3 uses Fourier transform to analyze the frequency components of the lesion area image, obtains the fuzziness index of the lesion area image, and obtains the fuzziness coefficient of the edge pixel point according to the fuzziness index;

[0119] The enhancement determination unit 4 obtains the enhancement coefficient of the lesion area image according to the resolution and blur coefficient of the edge pixel points, and enhances the ileocecal lesion area image by adjusting the enhancement coefficient of the lesion area image.

[0120] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A high-definition medical image processing method, characterized in that: The following steps are involved: S1, using an image acquisition device to obtain a medical image of the ileocecal region, marking it as an ileocecal region image, and performing noise reduction processing on the ileocecal region image through a guided filtering algorithm, extracting pixel values ​​of the ileocecal region image after the noise reduction processing, and equalizing the pixel values ​​of the ileocecal region image to obtain an ileocecal region image histogram; S2, calculating the potential secretion index of each pixel in the ileocecal image according to the ileocecal histogram, setting a segmentation threshold, and using the threshold segmentation method to compare and classify the potential secretion index of each pixel in the ileocecal image with the segmentation threshold, and obtaining the pixels in the ileocecal image that exceed the segmentation threshold to form a set of lesion pixels; S3, extracting pixel points from the set of suspected lesion pixel points as pixel points of the lesion area, and connecting the pixel points of the lesion area using an edge detection algorithm to form an image of the lesion area, extracting edge features of the lesion area image, and obtaining the resolution of edge pixel points according to the edge features of the lesion area image; S4, using Fourier transform to analyze the frequency components of the lesion area image, obtain a blur index of the lesion area image, and obtain a blur degree coefficient of the edge pixel point according to the blur index; S5. deriving an enhancement coefficient of the lesion region image according to the resolution and blur coefficient of the edge pixel points, and enhancing the ileocecal lesion region image by adjusting the enhancement coefficient of the lesion region image.

2. The high-definition medical image processing method according to claim 1, characterized in that: In S1, the ileocecal image is subjected to noise reduction processing by using a guided filtering algorithm, which specifically includes: A random image is selected from the ileocecal image as the guide image, denoted as A, and the ileocecal image to be denoised is used as the input image, denoted as B. The window radius of the guide filter is set to r, the regularization parameter is α, and for each pixel c in the ileocecal image to be denoised, a window D is established with it as the center. e , in window D e In the example, the output of the guided filter F c The calculation is done by the following formula, namely: F c =a e *A c +b e Among them, a e and b e is calculated by linear regression, A c represents the pixel value of the guidance image, F c Represents the ileocecal image after guided filtering denoising.

3. The high-definition medical image processing method according to claim 2, characterized in that: The step S2 calculates the potential secretion index of the ileocecal region according to the ileocecal region histogram, specifically comprising: Suppose the ileocecal image F after noise reduction is c The intensity value range of (x, y) is [0, L-1], and its size is M*N, where L represents the total number of color depths in the ileocecal image. The intensity histogram H(k) is calculated using the following formula, namely: in, Represented as image F c The gray value at coordinate (x, y), Expressed as Kronecker function, 0≤k≤L-1, the Kronecker function is defined as: The skewness and kurtosis of the intensity histogram H(k) are extracted, and a linear model is established based on the skewness and kurtosis of the intensity histogram H(k) to calculate the potential secretion index γ of each pixel. The expression of the potential secretion index γ of each pixel is: γ=ω1*PD H(k) +ω2*FD H(k) Among them, ω1 and ω2 are the weight coefficients of the potential secretion index γ affecting each pixel obtained through experimental data, PD is expressed as the skewness of the intensity histogram H(k), and FD is expressed as the kurtosis of the intensity histogram H(k).

4. The high-definition medical image processing method according to claim 3, characterized in that: In S2, the potential secretion index of each pixel in the ileocecal image is compared with the segmentation threshold and classified as follows: The secretion index threshold is set, the potential secretion index of the lesion is higher than that of the surrounding area, and the potential secretion index of each pixel in the ileocecal image is compared with the secretion index threshold, the pixels in the ileocecal image whose potential secretion index is higher than the segmentation threshold are screened out, and the pixels in the ileocecal image that exceed the segmentation threshold are grouped into a set of suspected lesion pixels.

5. The high-definition medical image processing method according to claim 1, characterized in that: The pixel points of the lesion area are connected in S3 to form an image of the lesion area, which specifically includes: The edge detection algorithm is used to calculate the gradient amplitude and direction of each pixel in the lesion area, non-maximum suppression is performed in the gradient direction, and high and low thresholds are set. According to the high and low thresholds, the gradient amplitude is divided into strong edge and weak edge categories, and the strong edge pixels are tracked. The weak edge pixels adjacent to the strong edge pixels are divided into edge pixels and connected to form an image of the lesion area.

6. The high-definition medical image processing method according to claim 1, characterized in that: The step S3 of obtaining the resolution of edge pixels according to the edge features of the lesion area image specifically includes: Extract the coordinates of all edge pixels in the lesion area image (x i ,y i ), i is the index of the edge pixel, the actual distance represented by each pixel in the horizontal direction is set to Δx, the actual distance represented by each pixel in the vertical direction is set to Δy, and the coordinates of the edge pixel points are converted from pixel units to actual space units, that is: X i =x i *Δx Y i =y i *Δy By calculating the distance between edge pixels and analyzing edge features, the edge pixel (X i , Y i ) and (X j , Y j ) is: The resolution of the edge pixels is obtained by calculating the minimum distance between the edge pixels. The calculation formula for the resolution of the edge pixels is: Among them, min(d ij ) represents the minimum distance between all edge pixels.

7. The high-definition medical image processing method according to claim 1, characterized in that: In S4, the frequency components of the lesion area image are analyzed by Fourier transform to obtain the blur index of the lesion area image, which specifically includes: The image of the lesion area is converted into a grayscale image, and the grayscale image is smoothed by Gaussian filtering. The grayscale image is defined as h(x', y'), and a two-dimensional fast Fourier transform is applied to the grayscale image, namely: f(u,v)=f{h(x',y')} Among them, f(u, v) is the frequency domain representation, (u, v) is the frequency coordinate, and the spectrum amplitude S(u, v) is calculated according to f(u, v). The spectrum amplitude S(u, v) represents the change intensity of each frequency component (u, v) in the frequency domain, reflecting the energy contribution of these frequencies in the original image. The expression of the spectrum amplitude S(u, v) is: S(u,v)=|f(u,v)| According to the spectrum amplitude, the ambiguity index is defined by using the energy ratio of the high-frequency component to the low-frequency component. The expression of the ambiguity index is: Among them, W represents the index set of high-frequency components, G represents the index set of low-frequency components, and the index set of high-frequency components represents the information of details, edges and noise in the image. The blur coefficient of edge pixels is defined as: R=1-Q. If the value of R is close to 1, it indicates that the pixel is clear, otherwise, it means that the pixel is blurred.

8. The high-definition medical image processing method according to claim 7, characterized in that: The enhancement coefficient of the lesion area image is obtained according to the resolution and blur coefficient of the edge pixel points in S5, which specifically includes: The enhancement coefficient T is set to be inversely proportional to the blur coefficient R of the edge pixel point and proportional to the resolution of the edge pixel point. The expression of the enhancement coefficient T is: T=t*(1-R)*δ Among them, t is a proportional constant used to control the degree of enhancement. The pixel value of the lesion area image is adjusted using the enhancement coefficient T. The lesion area image is taken at position (x i ,y i )’s pixel value P(x i ,y i ), by calculating the pixel value and the background pixel value P'(x i ,y i The product of the difference between the two values ​​and the enhancement coefficient T is used to obtain the enhanced part, and the enhanced part is added back to the original pixel value P(x i ,y i ), and the enhanced pixel value P is obtained ZQ (x i ,y i ), the enhanced pixel value P ZQ (x i ,y i ) is: P ZQ (x i ,y i )=P(x i ,y i )+T*[(P(x i ,y i ))-P’(x i ,y i )] The image details and brightness are enhanced by increasing the contrast with the background, thereby enhancing the image of the ileocecal lesion area.

9. A system for implementing a high-definition medical image processing method according to any one of claims 1 to 8, characterized in that: include: The acquisition and processing unit (1) is used to acquire a medical image of the ileocecal region and pre-process the ileocecal region image by using a guided filtering algorithm to obtain an ileocecal region image histogram; The lesion analysis unit (2) comprises a calculation and analysis module (21) and a lesion generation module (22); The calculation and analysis module (21) is used to calculate the potential secretion of each pixel point according to the ileocecal image histogram, set the secretion index threshold, and use the threshold segmentation method to perform comparison and classification to form a set of pathological pixel points; The lesion generation module (22) connects the pixel points of the lesion area using an edge detection algorithm to form an image of the lesion area, and obtains the resolution of the edge pixel points according to the edge features of the lesion area image; The fuzzy evaluation unit (3) uses Fourier transform to analyze the frequency components of the lesion area image, obtains a fuzziness index of the lesion area image, and obtains a fuzziness coefficient of the edge pixel point based on the fuzziness index; The enhancement determination unit (4) obtains an enhancement coefficient of the lesion region image according to the resolution and blur coefficient of the edge pixel points, and enhances the ileocecal lesion region image by adjusting the enhancement coefficient of the lesion region image.

Citation Information

Patent Citations

  • Confidence map-based method for distinguishing and detecting virtual object of augmented reality scene

    CN102509104A

  • Improved DME edema area neural network segmentation model construction method

    CN111340829A

  • Gastric signet cell carcinoma image intelligent classification system based on Unet transfer learning

    CN112861994A

  • Nuclear magnetic resonance image enhancement method based on image processing

    CN115100304A

  • Medical image enhancement method for intestinal endoscope

    CN116758069A