CN111815613B - A method for identifying liver cirrhosis disease stages based on the analysis of envelope line morphological features - Google Patents

Abstract

The present invention provides a liver cirrhosis disease stage recognition method based on the analysis of envelope line morphological characteristics, which comprises: obtaining the predicted membrane and the real membrane of the liver capsule according to the ultrasonic image; obtaining the variance VoS of the segmental slope from the predicted membrane, The coefficient of variation CV of the slope difference of adjacent segments, the number of fluctuations NoF; and the number of line segments NoL is obtained from the real film; in the preliminary prediction, VoS, NoF and CV are input as input features to identify "normal-early" and " mid-late stage”; if the preliminary identification result is “normal-early”, input NoL and CV as features into the mild recognition model to judge normal and mild liver cirrhosis; if the preliminary identification result is “mid-term- Late stage”, NoL and VoS are input as features into the mid-late stage recognition model to identify moderate cirrhosis and severe cirrhosis. The present invention combines the characteristics of the predicted membrane and the real membrane of the liver capsule for analysis, and can fully display the morphological characteristics of the liver capsule.

Description

A method for identifying liver cirrhosis disease stages based on the analysis of envelope line morphological features

technical field

The invention relates to the field of medical imaging, in particular to a method for identifying liver cirrhosis disease stages based on the analysis of envelope line morphological characteristics.

Background technique

Cirrhosis is a clinically common chronic progressive liver disease, specifically defined as the histological development of fibrous bands around regenerative nodules caused by chronic liver injury, and with the further development of cirrhosis, it can lead to portal hypertension and end-stage Liver disease has a high mortality rate, and it is very necessary for early detection and treatment of liver cirrhosis. At present, the clinical diagnosis of liver cirrhosis mainly depends on the subjective manual judgment of doctors. Due to the influence of objective factors such as experience, there may be great differences in individual diagnosis. Therefore, the computer-aided diagnosis system for liver cirrhosis based on quantitative analysis is very important. Necessary.

The traditional method of auxiliary diagnosis of liver cirrhosis based on liver capsule images is mainly based on a single real membrane or a single predicted membrane, and then realizes the classification and identification of its features based on machine learning algorithms. Such methods may lead to the loss of some features, thus As a result, the accuracy of feature classification and recognition is low, and the disease stage cannot be accurately judged, which has an adverse effect on clinical auxiliary diagnosis. Therefore, it cannot meet the needs of the current clinical auxiliary diagnosis.

Contents of the invention

In view of the defects in the prior art, the purpose of the present invention is to provide a method for identifying liver cirrhosis disease stages based on the analysis of capsule line morphological characteristics, which can obtain the predicted membrane and real membrane of the liver capsule from the ultrasound image, and obtain the recognition method therefrom. Features, realizing the staging recognition of liver cirrhosis.

The technical scheme provided by the invention is:

A method for identifying liver cirrhosis disease stages based on the analysis of envelope line morphological characteristics, comprising:

(S1) Obtain the predicted membrane and real membrane of the liver capsule according to the ultrasonic image;

(S2) Obtain the variance VoS of the subsection slope of the liver capsule, the coefficient of variation CV of the slope difference of adjacent segments, the number of times NoF of fluctuation changes from the predicted film; and obtain the number NoL of the line segments of the liver capsule from the real film;

(S3) The variance VoS of the segmental slope, the number of fluctuations NoF and the coefficient of variation CV of the slope difference of adjacent segments are input into the preliminary identification model as input features, and the two judgment types output by the preliminary identification model are "normal-early stage" and "mid-late";

(S4) carry out secondary classification according to the result of primary recognition; In secondary classification,

If the result of preliminary recognition is "normal-early stage", the number of line segments NoL and the coefficient of variation CV of the slope difference of adjacent segments are input into the light recognition model as features, and the two recognition types output by the light recognition model are "normal". and "mild cirrhosis";

If the result of the preliminary recognition is "middle-late stage", the number of line segments NoL and the variance VoS of the segmental slope are input into the mid-late stage recognition model as features, and the two recognition types output by the moderate and mild recognition model are "moderate liver cirrhosis ” and “severe cirrhosis”.

A further improvement of the present invention is that in the process of obtaining the number NoL of line segments, the number of breakpoints in the real membrane is counted, and the number NoL of line segments of the liver envelope is determined according to the number of breakpoints.

A further improvement of the present invention is that when obtaining the variance VoS of the segmental slope of the liver capsule, the predicted membrane of the liver capsule is divided into several segments according to a predetermined interval, the slope of each segment is calculated separately, and the The variance of the slope to get the variance VoS of the segmented slope.

以及标准差STD_Kd；将标准差STD_Kd处以均值/>

即可得到相邻段斜率差的变异系数CV。The further improvement of the present invention is that in the process of obtaining the coefficient of variation CV of the slope difference of adjacent segments, the slope difference between each adjacent segment of the predicted film is calculated, and the mean value of each slope difference is obtained

And the standard deviation STD _Kd ; put the standard deviation STD _Kd at the mean value />

The coefficient of variation CV of the slope difference of adjacent segments can be obtained.

A further improvement of the present invention is that in the process of obtaining the number of fluctuations NoF, the absolute value of the slope difference between each adjacent segment is calculated, and the number of times the absolute value of the slope difference is greater than the fluctuation threshold is counted as the fluctuation change The number of times NoF.

A further improvement of the present invention is that, in the process of separating the predicted membrane of the liver envelope, each segment has the same number of pixels; when calculating the slope of the segment, the slope between the two endpoints of the segment is used as the segment. Segment slope, or use the slope of the line fitted by each pixel in the segment as the segment slope.

A further improvement of the present invention is that the ultrasonic image is a superficial section image of the liver, which includes the liver capsule at the upper edge of the liver.

A further improvement of the present invention lies in that the preliminary recognition model is a support vector machine model, which is trained by ten times of five-fold cross-validation.

A further improvement of the present invention is that the mild recognition model and the mid-late recognition model are both K-means clustering models, which are trained by ten times of 50-fold cross-validation.

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

1) Combined with the analysis of the characteristics of the predicted membrane and the real membrane of the liver capsule, the morphological characteristics of the liver capsule can be fully displayed.

2) Using the two-stage classification model of support vector machine and K-means clustering, the classification accuracy is improved, which provides a reliable guarantee for clinical auxiliary diagnosis.

Description of drawings

Other characteristics, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

Fig. 1 is the flow chart of the identification method of liver cirrhosis disease stage based on envelope line morphological feature analysis;

Fig. 2 is the schematic diagram of the recognition model adopted in the present invention;

3 is a flow chart of a method for extracting liver capsule from an ultrasound image of liver cirrhosis based on digital image processing technology;

Fig. 4 is a schematic diagram of the principle of sliding window detection;

Figure 5 is a schematic diagram of the principle of the liver capsule traversal search algorithm for images without ascites;

Fig. 6 is a schematic diagram of the principles of the liver capsule traversal search algorithm for images with ascites;

Figure 7 is a schematic diagram of the real membrane extraction process.

Detailed ways

The present invention will be described in detail below in conjunction with specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

As shown in Figure 1, the embodiment of the present invention includes a method for identifying liver cirrhosis disease stages based on the analysis of envelope line morphological characteristics, the method includes the following steps:

(S1) Obtain the predicted and real membranes of the liver capsule from the ultrasound image. The ultrasound image used in this embodiment is a superficial slice image of the liver, and the ultrasound image is the liver capsule at the upper edge of the liver. As shown in Figure 5, the lower half of the liver superficial section image is the liver parenchyma, the upper half is other organs, and the liver capsule is located in the middle of the image. The prediction film contains the trend information of the liver capsule, and the prediction film contains the fluctuation of the liver capsule, which has rich details; the combination of the prediction film and the real film can provide rich features for the automatic judgment of downstream liver diseases .

(S2) Obtain the variance VoS of the segmental slope of the liver capsule, the coefficient of variation CV of the slope difference between adjacent segments, and the number of fluctuations NoF from the predicted film; and obtain the number NoL of the line segments of the liver capsule from the real film. The above features are used as input features for subsequent recognition tasks.

Specifically, the predicted membrane of the liver capsule is a preliminary identification result of the liver capsule obtained from the ultrasound image, and the number NoL of line segments of the liver capsule is calculated from the predicted membrane, and is used to represent the continuity of the liver capsule. In the process of obtaining the number of line segments (Number of Line, NoL), the number of breakpoints in the real membrane is counted, and the number of line segments NoL of the liver envelope is determined according to the number of breakpoints. Its expression is:

NoL=k

L={l _i |i=1,...,k} (1)

L{l _i |i1,...,k} represents a collection of line segments constituting the entire liver capsule. In this embodiment, the "line segment" in "the number of line segments NoL" is not a straight line between two points in geometry, and the "line segment" in this embodiment refers to the points in the real membrane of the liver capsule. Segments, which can be straight or curved.

The variance of the segmental slope of the liver capsule (Variance of slope, VoS) and the coefficient of variation CV of the slope difference between adjacent segments were used to represent the smoothness of the liver capsule. Smoothness mainly describes the fluctuation of the liver capsule, which can effectively describe the overall trend and fluctuation of the liver capsule. Due to the discontinuity of the liver capsule at different stages, in order to avoid the influence of the discontinuous area on the feature analysis, this embodiment adopts the same spacing (same number of pixels) in each segment to divide the segments, and then divides each segment by The segmental slope in the segment is analyzed comprehensively, so as to obtain the overall liver capsule characteristics.

When obtaining the variance VoS of the segmental slope of the liver capsule, the predicted membrane of the liver capsule is divided into several segments according to a predetermined interval, and the slope of each segment is calculated separately, and the variance of the slope of each segment is calculated to obtain the score The variance VoS of the segment slope.

Specifically, assuming that there is a breakpoint in the image of the liver capsule, the predicted membrane is divided into m segments due to the breakpoint, where m=(1,2,...,i), the segment of the predicted membrane is divided into n segments at the same interval P Segments, where n=(1,2,...,j), then there are N _L segments in the whole liver capsule image.

where n _i represents the number of segments in the i-th segment of the predicted membrane. Therefore, it can be known that the slope of each segment in each segment is shown in formula 3, and the final slope variance of the entire liver capsule is shown in formula 4.

表示整条肝包膜中所有分段斜率的平均值。本实施例中，计算分段的斜率时，采用分段的两个端点之间的斜率作为分段斜率，在具体实施过程中，也可采用分段中各像素拟合后的直线的斜率作为分段斜率。In the formula, K _ij represents the slope of the j-th segment in the i-th segment, (x _Eij , y _Eij ) and (x _Fij , y _Fij ) represent the coordinates of the first and last pixels of each segment, respectively,

Indicates the average of the slopes of all segments in the entire liver capsule. In this embodiment, when calculating the slope of the segment, the slope between the two endpoints of the segment is used as the segment slope. In the specific implementation process, the slope of the straight line fitted by each pixel in the segment can also be used as the Segment slope.

The coefficient of variation is a statistic of the degree of variation of each observed value in the measurement index. Therefore, on the basis of obtaining the slope of each segment, this embodiment analyzes the degree of variation of the difference between the slopes of adjacent segments, so as to better realize the liver Description of envelope volatility.

以及标准差STD_Kd；将标准差STD_Kd处以均值/>

即可得到相邻段斜率差的变异系数CV。相邻段斜率差的变异系数CV如公式6所示。In the process of calculating the coefficient of variation CV of the slope difference of adjacent segments, the slope difference between each adjacent segment of the predicted film is calculated, and the average value of each slope difference is obtained

And the standard deviation STD _Kd ; put the standard deviation STD _Kd at the mean value />

The coefficient of variation CV of the slope difference of adjacent segments can be obtained. The coefficient of variation CV of the slope difference between adjacent segments is shown in Equation 6.

表示斜率差的均值。In the formula _{, Kd} represents the slope difference of adjacent small line segments, STD _Kd represents the standard deviation of the slope difference,

Indicates the mean of the slope differences.

In the process of obtaining the number of fluctuations NoF, the absolute value of the slope difference between adjacent segments is calculated, and the number of times the absolute value of the slope difference is greater than the fluctuation threshold is counted as the number of fluctuations NoF. In this embodiment, the fluctuation threshold is 0.3. When the absolute value |K _d | of K _d between adjacent segments is greater than 0.3, it is regarded as a fluctuation. The total number of fluctuations in the predicted membrane is the number of fluctuations (NoF).

(S3) The variance VoS of the segmental slope, the number of fluctuations NoF and the coefficient of variation CV of the slope difference of adjacent segments are input into the preliminary identification model as input features, and the two judgment types output by the preliminary identification model are "normal-early stage" and "mid-late".

As shown in Figure 2, in the specific implementation process, the three characteristics of the variance VoS of the segmental slope, the number of fluctuations NoF, and the coefficient of variation CV of the slope difference of adjacent segments form a three-dimensional vector, which is input into the preliminary identification model. The identification model is a support vector machine model, which is trained by ten times of five-fold cross-validation. The model can obtain two types of identification results: "normal-early" and "middle-late". The identification accuracy of the model can be obtained by averaging the identification results of these two categories.

(S4) Carry out secondary classification according to the result of the primary recognition. As shown in Figure 2, in the secondary classification, it is necessary to consider determining the corresponding model based on the results of the preliminary identification, specifically:

If the result of preliminary recognition is "normal-early stage", the number of line segments NoL and the coefficient of variation CV of the slope difference of adjacent segments are input into the light recognition model as features, and the two recognition types output by the light recognition model are "normal". and "mild cirrhosis". The mild recognition model can analyze the continuity and smoothness of the liver capsule to obtain the final recognition result.

If the result of the preliminary recognition is "middle-late stage", the number of line segments NoL and the variance VoS of the segmental slope are input into the mid-late stage recognition model as features, and the two recognition types output by the moderate and mild recognition model are "moderate liver cirrhosis ” and “severe cirrhosis”.

The above-mentioned mild recognition model and the mid-late recognition model are both K-means clustering models, which are trained by ten times of five-fold cross-validation. The average value of the recognition results of the above models can be taken as the recognition accuracy.

The step (S1) of this embodiment is used to process the image of the superficial section of the liver, and the image of the superficial section of the liver is an ultrasonic image acquired at a specific position of the patient. The superficial section image of the liver is shown in Figure 5, which includes the liver capsule at the upper edge of the liver, the lower half of the image is the liver parenchyma, and the upper half is other organs.

As shown in Figure 3, in the step (S1) of this embodiment, extracting the predicted membrane and the real membrane of the liver envelope includes the following steps:

(S11) Traversing the upper part of the ultrasound image by using a sliding window algorithm to identify whether there is a hepatic ascites area in the ultrasound image. This step specifically includes:

(S111) Adopt a L×L square window and slide from left to right and from top to bottom in the upper half of the ultrasonic image with a step size L, thereby traversing the upper half of the ultrasonic image; during the traversal process, each The average grayscale of each window, and compare the average grayscale with the window threshold, if the average grayscale is less than the window threshold, it is judged that there is hepatic ascites feature in this window; the sliding window detection principle is shown in Figure 4.

Specifically, if the size of the ultrasound image is M×N, the size of the sliding window traversal area (the upper half of the ultrasound image) is M×N/2, so it can be seen that when the window traverses a row or a column according to the step size L, the following can be obtained respectively: The number of windows shown in Equation 7 can be used to obtain the total number of windows shown in Equation 8 when traversing the entire upper half of the ultrasound image.

_Sr = M/L;

_Sc = (N/2)/L (7)

S _a =S _r ×S _c (8)

where S _r , S _c , S _a represent the number of sliding windows in one row, one column and the entire upper half of the ultrasound image, respectively. When judging the presence of hepatic ascites features in each window, the window threshold used was 60.

(S112) Count the number S _f of windows containing hepatic ascites features, and divide it by the number of windows S _a in the upper part of the ultrasound image to obtain the characteristic coefficient of hepatic ascites; when the characteristic coefficient of hepatic ascites is greater than the fixed threshold P, judge the ultrasound image including the hepatic ascites area. The rules for judging whether to include the hepatic ascites area are shown in Table 1.

Table 1 Criteria for identifying ascites areas in ultrasound images

(S12) Process the ultrasonic image by using a Gaussian blur algorithm to reduce image noise and detail level.

In this step, the Gaussian blur filter is used to process the ultrasonic image, and its definition is shown in formula (9).

where r is the blur radius and σ is the standard deviation of the normal distribution. In two-dimensional space, the convolution matrix composed of pixels whose distribution is not zero is transformed with the original image, and the value of each pixel is the weighted average of the surrounding adjacent pixel values. The value of the original pixel has the largest Gaussian distribution value, so it has the largest weight. As the adjacent pixels get farther and farther away from the original pixel, their weights are getting smaller and smaller, so that on the basis of retaining the edge effect, the ultrasonic Image denoising.

(S13) Using the method of multi-scale detail enhancement and fuzzy set transformation on the Gaussian-blurred ultrasonic image to perform local image enhancement first and then global image enhancement. This step specifically includes:

(S131) Using three Gaussian kernel functions of different scales to perform convolution operation with the Gaussian blurred ultrasonic image I to obtain three Gaussian blurred images B ₁ , B ₂ , B ₃ , the expressions of which are:

B ₁ =G ₁ *I, B ₂ =G ₂ *I, B ₃ =G ₃ -I (10)

Among them, G ₁ , G ₂ , and G ₃ are Gaussian kernels with standard deviations of 1, 2, and 4 respectively;

(S132) Subtract the ultrasonic image I from the three Gaussian blurred images B ₁ , B ₂ , B ₃ respectively to obtain three detail images D ₁ , D ₂ , D ₃ , the expression of which is:

D ₁ =IB ₁ , D ₂ =B ₁ -B ₂ , D ₃ =B ₂ -B ₃ (11)

(S133) Setting different weights to integrate each detail image into a final detail image D', which is expressed as:

D′＝(1-ω ₁ ×sgn(D ₁ ))×D ₁ +ω ₂ ×D ₂ +ω ₃ ×D ₃

(12)

Among them, ω ₁ , ω ₂ , ω ₃ are weights respectively; usually, ω ₁ , ω ₂ , ω ₃ are set to 0.5, 0.5, 0.25 respectively.

(S133) Perform overall enhancement on the detail-enhanced image by using fuzzy combination. Through this algorithm, the dark pixels in the image can be made darker and the bright pixels brighter, which is also just suitable for the global enhancement processing of liver ultrasound images, and can better meet the needs of subsequent morphological processing.

(S14) Binarize the enhanced ultrasonic image, and use an island removal algorithm to remove isolated noise points in the binary image to obtain a binary ultrasonic image.

In this step, first use the adaptive threshold algorithm to binarize the ultrasonic image after denoising and enhancement, because the cords in the liver parenchyma may cause some isolated noise points in the binarized image, so , before the morphological processing, it is necessary to set a certain threshold by the area method, and remove the isolated noise area smaller than the threshold, so as to obtain a more concise binary ultrasound image.

(S15) Process the binary ultrasound image by using morphological closing operation; after processing, the lower part of the binary ultrasound image is the liver image area, and the upper part is a single-color independent connected area. Morphological closing operations usually include dilation and erosion operations. The final binary ultrasound image is shown in Figure 5.

(S16) Search the processed binary ultrasound image based on the traversal algorithm, find the pixel point set corresponding to the liver capsule to form the predicted membrane of the liver capsule; fuse the predicted membrane into the original ultrasound image, and use gray difference The algorithm removes the pixels of pseudomembrane from the predicted membrane to obtain the real membrane of the liver capsule. This step needs to be divided into two parts: obtaining the predicted membrane of the liver capsule and obtaining the real membrane of the liver capsule. Obtaining the predicted membrane of the liver capsule needs to consider both the presence of hepatic ascites and the absence of hepatic ascites. specific:

As shown in Figure 5, searching for pixels corresponding to the liver capsule in an ultrasound image without hepatic ascites includes the following steps:

(S1601) Select a plurality of starting points on the top of the binary ultrasound image;

(S1602) Search downward from each starting point to find a pixel point opposite in color to the monochromatic independently connected region at the top of the binary ultrasound image, and use this pixel point as a pixel point corresponding to the liver capsule.

In the binary ultrasound image, the pixel value of the liver parenchyma below the image is 0, and the pixel value of the simply connected region above the image is 1. In a specific implementation process, each pixel point on the top edge of the binary ultrasound image is used as a starting point to traverse sequentially. During the traversal process for a certain starting point, traverse and search along the positive direction of the ordinate, when the gray value of the pixel is detected to be 0, stop the search of this column, and define this point as a point on the "predicted membrane" of the liver capsule _P1 ^* . Then, the point P ₁ is translated from the initial position along the x-axis with a step size of 1 pixel, and a total of m pixels of P ₁ , P ₂ , P ₃ ,...,P _m are sequentially obtained, and the above-mentioned traversal is repeated using them as the starting point According to the search method, the m pixel points P ₁ ^* , P ₂ ^* , P ₃ ^* ,...,P _m ^* that finally constitute the "predicted membrane" of the liver envelope can be obtained.

As shown in Figure 6, for an ultrasound image with hepatic ascites, finding the pixel corresponding to the liver capsule includes the following steps:

(S1611) Select a plurality of starting points at the top of the binary ultrasound image;

(S1612) Search downward from each starting point to find the pixel corresponding to the liver capsule;

In the process of searching downward from a certain starting point, look for the first pixel that is opposite in color to the monochromatic independently connected region at the top of the binary ultrasound image, and continue to search downward until it finds the pixel point that is opposite to the monochromatic independently connected region at the top of the ultrasound image. The pixel points with the same color in the independently connected regions are used as the pixel point corresponding to the liver capsule, and if the pixel point is not found, the starting point is skipped.

In the case of hepatic ascites, the selection method of each starting point is the same as that of the case without hepatic ascites. The difference between the two lies in the process of traversing each column of pixels. In the process of specific implementation, when performing traversal search in each column, the pixel gray level of the starting point and the simply connected region where it is located is 1. When the gray value of the pixel is detected to be 0, continue to search in the y-axis direction until If a pixel with a gray value of 1 is detected again, the pixel can be defined as a pixel on the liver capsule, and finally all retrieved pixels P _f1 ^* , P _f2 ^* , P _f3 ^* ,..., P _fm ^* constitutes the "predictive membrane" of the liver capsule with ultrasound images of ascitic fluid. In addition, because the liver capsule in the case of ascites usually has a break point, when the column where the break part is located is detected, the pixel with a gray value of 1 cannot be detected again after the pixel with a gray value of 0 is detected. pixel, then directly skip this column and continue the traversal search of the next column. In this case, there is no pixel of the predicted membrane in this column.

As shown in Figure 7, during the acquisition process of the prediction membrane of the liver capsule, the image has undergone complex processing, especially after expansion and erosion, the liver capsule that was originally discontinuous on the ultrasound image becomes continuous, so that the discontinuous liver capsule in the discontinuous place Form the pseudo-membrane shown in Figure 7c, so in the process of obtaining the real film, it is necessary to remove the pixels of the pseudo-membrane in the predicted film, and use the remaining pixels as the pixels of the real film, which specifically includes the following steps:

(S1621) Fusing each pixel of the expanded prediction membrane into the original ultrasound image;

(S1622) In the fused image, traverse the pixels of each column of the predicted membrane; during the process of traversing the pixels of a certain column of the predicted membrane, calculate the range of the predicted membrane of the liver capsule in the vertical direction of the pixel of this column The average value of several pixels within; if the average value is less than the false film threshold, the pixel point of the predicted film is the pixel point of the real film, otherwise the pixel point of the predicted film is not the pixel point of the real film. In this example,

In the specific implementation process, please refer to Figure 7, assuming that the pixel coordinates of the expanded "predicted membrane" area are (x, y), where x∈(1,m), when the expanded "predicted membrane" is fused to the original After the ultrasound image is on, look for the presence of "pseudomembrane" column by column sequentially starting from the first column. After expansion, it is predicted that the film will occupy a certain range in the vertical direction, and the expanded line width is the range of calculating the average value of pixels during the traversal process. According to the principle of formula 13, if the average pixel value of each column in the original ultrasound image within the y-value range of the "predicted membrane" is greater than the threshold value P, it proves that this is the real liver capsule, thereby saving the value of this column in the "predicted membrane" The original pixel value; on the contrary, if the average value in the range is less than the pseudo-membrane threshold P, it proves that the predicted liver capsule here is a "pseudo-membrane", so the pixel value of this column in "Predicted Membrane" should be set to 0. After traversing all columns of the entire image, the remaining pixels in the "predicted membrane" can form the final real liver capsule that reflects the original ultrasound image.

In the formula, G ₁ (i, j) represents the gray value of the liver capsule pixel at coordinates (i, j) in the “predicted membrane” image, and G ₂ (x, y) represents the original ultrasound image at coordinates (x, y) The gray value at the pixel point, G ₃ (i, j) represents the gray value of the pixel at the coordinate (i, j) in the real liver capsule.

In the description of this application, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", The orientation or positional relationship indicated by "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the application and simplifying the description, rather than indicating or implying the referred device Or elements must have a certain orientation, be constructed and operate in a certain orientation, and thus should not be construed as limiting the application.

Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. In the case of no conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims (6)

1. A liver cirrhosis disease stage identification method based on envelope line morphological feature analysis, characterized in that it comprises: (S1) Obtain the predicted membrane and real membrane of the liver capsule according to the ultrasonic image; (S2) Obtain the variance VoS of the subsection slope of the liver capsule, the coefficient of variation CV of the slope difference of adjacent segments, the number of times NoF of fluctuation changes from the predicted film; and obtain the number NoL of the line segments of the liver capsule from the real film; When obtaining the variance VoS of the segmental slope of the liver capsule, the predicted membrane of the liver capsule is divided into several segments according to a predetermined interval, and the slope of each segment is calculated separately, and the variance of the slope of each segment is calculated to obtain the score The variance VoS of the segment slope; In the process of obtaining the coefficient of variation CV of the slope difference of adjacent segments, calculate the slope difference between each adjacent segment of the predicted membrane, and obtain the mean value of each slope difference

And the standard deviation STD _Kd ; put the standard deviation STD _Kd at the mean value />

The coefficient of variation CV of the slope difference of adjacent segments can be obtained; In the process of obtaining the number of fluctuations NoF, calculate the absolute value of the slope difference between adjacent segments, and count the number of times the absolute value of the slope difference is greater than the fluctuation threshold and use this as the number of fluctuations NoF; (S3) The variance VoS of the segmental slope, the number of fluctuations NoF and the coefficient of variation CV of the slope difference of adjacent segments are input into the preliminary identification model as input features, and the two judgment types output by the preliminary identification model are "normal-early stage" and "mid-late"; (S4) carry out secondary classification according to the result of primary recognition; In secondary classification, If the result of preliminary recognition is "normal-early stage", the number of line segments NoL and the coefficient of variation CV of the slope difference of adjacent segments are input into the light recognition model as features, and the two recognition types output by the light recognition model are "normal". and "mild cirrhosis"; If the result of the preliminary recognition is "middle-late stage", the number of line segments NoL and the variance VoS of the segmental slope are input into the mid-late stage recognition model as features, and the two recognition types output by the middle-late stage recognition model are "moderate liver cirrhosis" and "severe cirrhosis". 2. a kind of liver cirrhosis disease staging recognition method based on envelope line morphological feature analysis according to claim 1, is characterized in that, in the process of obtaining the quantity NoL of line segment, counts the number of breakpoints in the real membrane, and according to The number of breakpoints determines the number NoL of line segments of the liver envelope. 3. a kind of liver cirrhosis disease staging identification method based on capsule line morphological feature analysis according to claim 1, is characterized in that, in the process of separating the prediction membrane of described liver capsule, each segment has the same number pixels; when calculating the slope of a segment, the slope between the two endpoints of the segment is used as the segment slope, or the slope of the line fitted by each pixel in the segment is used as the segment slope. 4. A method for identifying liver cirrhosis disease stages based on envelope line morphological feature analysis according to any one of claims 1 to 3, wherein the ultrasonic image is a superficial section image of the liver, which includes the upper edge of the liver liver capsule. 5. a kind of liver cirrhosis disease stage recognition method based on envelope line morphological feature analysis according to claim 1, is characterized in that, described preliminary recognition model is support vector machine model, adopts the mode of ten times of five-fold cross-validation Get trained. 6. A method for identifying liver cirrhosis disease stages based on envelope line morphological feature analysis according to claim 1, wherein both the mild recognition model and the mid-late recognition model are K-means clustering models , trained by ten times of five-fold cross-validation.

Images