CN106803251A - The apparatus and method of aortic coaractation pressure differential are determined by CT images - Google Patents

Abstract

The invention discloses a device and a method for determining the pressure difference at the constriction of the aorta from CT images. The device of the present invention includes: a data reading module, a two-dimensional slicing module, a segmentation module, a three-dimensional reconstruction module, aortic coarctation discrimination module, feature extraction module, classification module and result display module. The steps of the present invention include: 1. reading CT data, 2. performing two-dimensional slice processing on the CT data, 3. segmenting the aorta image, 4. measuring the aorta diameter ratio, 5. judging whether there is constriction, 6. Extract features from the three-dimensional aortic model, 7. Classify the degree of aortic coarctation, and 8. Display the pressure difference at the coarctation of the aorta. The invention classifies the morphological features of CT images by using a machine learning algorithm, and can directly obtain a relatively accurate pressure difference value at the coarctation of the aorta from the CT images, thereby improving the accuracy of the CT images in the diagnosis of coarctation of the aorta.

Description

Device and method for determining pressure difference at coarctation of aorta from CT images

technical field

The invention belongs to the field of electronic technology, and further relates to a device and method for determining the pressure difference at the coarctation of the aorta from computed tomography CT (Computed Tomography) images in the technical fields of medical image processing, artificial intelligence and machine learning. The present invention aims at the situation that the pressure difference needs to be quantified when diagnosing coarctation of the aorta in clinical medicine, and realizes quantitatively obtaining the pressure difference at both ends of the coarctation of the aorta from CT images by using the device for determining the pressure difference at the coarctation of the aorta .

Background technique

Coarctation of the aorta is a common malformation in congenital heart disease. At present, CT diagnosis of coarctation of the aorta is mainly based on the qualitative assessment of the severity of coarctation of the aorta based on morphological features. To achieve an accurate assessment To determine the severity of coarctation of the aorta, it is necessary to use a pressure guide wire to invasively penetrate into the lesion to measure the pressure difference between the front and rear ends. relatively expensive.

The Institute of Semiconductors of the Chinese Academy of Sciences discloses an optical fiber in the patent document "A Fiber Optic Cardiac Pressure Guidewire" (publication number: CN105054916A, application number: CN201510609653.4, application date: November 18, 2015). Intracardiac guidewire. The device includes an optical fiber and a pressure measuring cavity fixed on the end face of the optical fiber, and realizes that the guide wire is inserted into the human body to measure the pressure difference. The disadvantage of this device is that the device must be inserted into the inside of the human body, and the method of invasive lateral pressure will cause certain damage to the patient, and the cost is relatively expensive.

Matthew J.Budoff, A Shittu, S Roy, described in their published paper "Use of cardiac computed tomography in the diagnosis and management of coarctation of theaorta" (Journal of Thoracic & Cardiovascular Surgery, 2013,146(1):229-32) A method for assessing coarctation of the aorta from CT images. In this method, a CT scan is first performed on the lesion of the patient, and the CT image is imported into the computer imaging system. Secondly, the features on the CT image are read from the computer, such as the ratio of narrowing, the position of the narrowing, etc. Enables qualitative assessment of coarctation of the aorta. The disadvantage of this method is that the degree of aortic coarctation cannot be quantitatively evaluated, and the pressure difference at the coarctation of the aorta, which is the gold standard for judging aortic coarctation, cannot be obtained from CT images.

Suzhou Runxin Medical Technology Co., Ltd. applied for the patent document "Calculation method of coronary blood flow reserve fraction based on cardiac CT image" (public number: CN106023202A, application number: CN201610339892.7, application date: May 20, 2016 ) discloses a method for quantitatively determining the degree of coronary artery stenosis from CT images. The method first extracts the myocardial image, accurately segments the coronary arteries, and then performs edge detection on the coronary artery volume data to generate a triangular mesh model of the coronary arteries, and finally realizes the quantitative evaluation of the degree of coronary artery stenosis. The disadvantage of this method is that the blood vessels targeted are only the coronary arteries, the physiological structure of the coronary arteries differs greatly from that of the aorta, and coronary artery stenosis is often formed due to blockage such as thrombus, while aortic stenosis is due to the physiological structure of the blood vessels. Changes, the above method is not suitable for the assessment of aortic coarctation.

To sum up, the existing methods can only roughly estimate the coarctation of the aorta from CT images. To obtain the pressure difference between the two ends of the coarctation of the aorta, only invasive measurement can be used, which will harm the patient. Larger and more expensive.

Contents of the invention

The purpose of the present invention is to provide a device and method for directly obtaining the pressure difference at the coarctation from CT images, which can be non-invasive, quantitative, and rapid Accurately evaluate the severity of aortic coarctation based on CT images, which provides powerful assistance for doctors' diagnosis and treatment, and can reduce the treatment cost of patients.

The idea of realizing the object of the present invention is to segment out the aortic vessel in the preoperative CT image of the patient with congenital heart disease, and carry out three-dimensional visualization reconstruction, and determine the coarctation of the aorta on the model obtained by reconstruction, and further analyze the data determined as coarctation. Features are extracted, and machine learning methods are used to map the features to a specific range of differential pressure values to obtain the differential pressure value at the coarctation of the aorta, which provides reference for clinicians.

To achieve the above object, the device of the present invention includes a data reading module, a two-dimensional slicing module, a segmentation module, a three-dimensional reconstruction module, aortic coarctation discrimination module, feature extraction module, classification module, and result display module, wherein:

The data reading module is used to read in the original chest CT data whose format is .dcm or .raw;

The two-dimensional slicing module is used to map the read-in original chest CT data into two-dimensional sliced image matrices of pixels with different gray values;

The segmentation module is used to select a slice image matrix including the end of the descending aorta from the two-dimensional slice image matrix, select a slice image matrix including the end of the descending aorta, and take the center point of the selected slice image matrix as the center, Generate a square frame of n*n centimeters, the size of n is between one-fifth and one-fourth of the length of the image matrix; use the formula S[i,j]=I[p,q]×G[0.25] , calculate the gray value at the coordinates of each pixel in the smoothed image, remove the noise of the two-dimensional slice image, use the medical image software to extract the edge of the blood vessel at the end of the descending aorta in the square frame, and obtain the edge image of the blood vessel wall; the edge image of the blood vessel wall The gray value of all the surrounding points is set to 255 to form an image of the inner area of the blood vessel, using The formula calculates the centroid of the image of the inner area of the blood vessel; loads the slice image matrix of the lowest layer other than the slice image matrix of the marked blood vessel area, determines a seed point element on the slice image matrix, and the abscissa of the seed point element is consistent with the slice image of the upper layer The abscissa of the center of mass of the inner area of the matrix blood vessel is equal, and the ordinate is equal to the ordinate of the center of mass of the inner area of the blood vessel in the upper slice image matrix. The edge of the blood vessel wall image is used as the boundary, and the seed point is determined by the element coordinates of the seed point to perform 8 Neighborhood adaptive areas grow to obtain the image of the aortic region on the sliced image matrix; where, S[i,j] represents the gray value of the pixel at [i,j] in the image after smoothing, and I[p,q] represents the two The gray value of the pixel at [p,q] in the three-dimensional slice image, G[0.25] represents a Gaussian function with a standard deviation of 0.25, M represents the centroid point of the image of the inner region of the blood vessel, ∫ represents the integral operation, f(x , y) represents the gray value at the image pixel point (x, y) of the inner region of the blood vessel;

The three-dimensional reconstruction module is used to import the image of the aortic region on the slice image matrix into three-dimensional reconstruction software for three-dimensional volume rendering, and obtain a three-dimensional aortic model after rendering;

The aortic coarctation discrimination module is used to measure the aortic diameter at intervals of 1 mm in the three-dimensional aortic model, compare the aortic diameters measured twice adjacently, and store the ratio result in the diameter In the ratio statistical table, if the values in the diameter ratio statistical table are greater than 0.8, there is narrowing, otherwise, there is no narrowing;

The feature extraction module is used to import the three-dimensional aorta model into medical image processing software, and use the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1; when the coarctation of the aorta is located at the position of the ascending aorta Assign the logo of 0 to 0, assign the logo of the coarctation of the aorta to the position of the descending aorta as -1, assign the logo of the coarctation of the aorta to the position of the aortic arch to 1, and use the assigned logo as the value of feature 2; Import the three-dimensional aortic model into the medical image processing software, and use the diameter of the narrowest part of the aortic vessel measured by the software as the value of feature 3; import the three-dimensional aortic model into the medical image processing software, and use the software to measure and calculate the most The ratio of the diameter of the constriction to the diameter of the descending aorta was taken as the value of feature 4; the ratio of aortic coarctation area was calculated using the formula R1=πR ² /(SQRT((H×W)/3600), and the calculation result was taken as feature 5 The value of R2=D/SQRT(SQRT((H×W)/3600) formula to calculate the aortic coarctation ratio, and use the calculation result as the value of feature 6; wherein, R1 represents the coarctation area ratio of the aortic image , π represents the circumference ratio, R represents the radius of the most constricted part of the aortic image, / represents the division operation, SQRT represents the square root operation, H represents the patient's height, W represents the patient's weight, R2 represents the constriction ratio of the aortic image, D represents the aorta The diameter of the narrowest point in the arterial image;

The classification module is used to input 6 eigenvalues into the main systolic pressure difference model, and the main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues;

The result display module is used to display the pressure difference value at the coarctation of the aorta obtained from the main systolic pressure difference model.

Method of the present invention comprises the steps:

(1) Read CT data:

The data reading module reads in the original chest CT data in the format of .dcm or .raw;

(2) Perform two-dimensional slice processing on CT data:

The two-dimensional slicing module maps the read-in original chest CT data into two-dimensional sliced image matrices of pixels with different gray values;

(3) Segment the aortic image:

(3a) The segmentation module selects the slice image matrix containing the end of the descending aorta from the two-dimensional slice image matrix, and takes the center point of the selected slice image matrix as the center to generate a square frame of n*n centimeters, and the size of n is between Between one-fifth and one-fourth of the length of the image matrix;

(3b) Using the Gaussian smoothing formula, calculate the gray value at the coordinates of each pixel point of the smoothed image, remove the noise of the two-dimensional slice image, and use the medical image software to extract the edge of the descending aorta end vessel in the square frame to obtain the edge of the vessel wall image;

(3c) Set the gray value of all internal points surrounded by the edge image of the blood vessel wall to 255 to form an image of the internal area of the blood vessel, and use the Image Moments formula to calculate the centroid of the image of the internal area of the blood vessel;

(3d) loading the lowest slice image matrix other than the slice image matrix of the marked blood vessel area, and using the aorta slice image segmentation method to obtain the image of the aortic intravascular area on the slice image matrix;

(3e) judging whether all slice image matrices have been loaded, if so, perform step (4), otherwise, perform step (3d);

(4) Measure the aortic diameter ratio:

(4a) The three-dimensional reconstruction module imports the image of the inner region of the aorta on the slice image matrix into the three-dimensional reconstruction software for three-dimensional volume rendering, and obtains the three-dimensional aortic model after rendering;

(4b) In the three-dimensional aorta model, the aortic diameter was measured every 1mm interval;

(4c) compare the aortic diameters of two adjacent measurements respectively, and store the ratio results in the diameter ratio statistics table;

(5) Whether the aortic coarctation discrimination module judges whether there is a value less than 0.8 in the diameter ratio statistical table, if so, then perform step (6), otherwise, perform step (7);

(6) Extract features from the three-dimensional aorta model:

(6a) The feature extraction module imports the three-dimensional aorta model into the medical image processing software, and uses the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1;

(6b) Assign a value of 0 to the flag when the coarctation of the aorta is located at the position of the ascending aorta, assign a value of -1 to the flag when the coarctation of the aorta is located at the position of the descending aorta, and assign a value to the flag when the coarctation of the aorta is located at the position of the aortic arch is 1, and the assigned identifier is used as the value of feature 2;

(6c) Import the three-dimensional aorta model into the medical image processing software, and use the diameter of the narrowest part of the aortic vessel measured by the software as the value of feature 3;

(6d) Import the three-dimensional aorta model into the medical image processing software, and use the ratio of the diameter of the narrowest part measured and calculated by the software to the diameter of the descending aorta as the value of feature 4;

(6e) Using the narrowing area ratio formula to calculate the aortic narrowing area ratio of the aortic image, and use the calculation result as the value of feature 5;

(6f) using the narrowing ratio formula to calculate the narrowing ratio of the aortic image, and use the calculation result as the value of feature 6;

(7) Classify the degree of aortic coarctation:

(7a) The classification module inputs 6 eigenvalues into the main compression difference model;

(7b) The main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues;

(8) The result display module displays the pressure difference value at the coarctation of the aorta obtained from the main systolic pressure difference model.

Compared with the prior art, the present invention has the following advantages:

First, since the device of the present invention adopts a CT image segmentation module, a CT image feature extraction module, and a CT image classification module, the CT image can directly obtain a more accurate pressure difference value at the coarctation of the aorta, which overcomes the problem of the prior art guide wire measurement. The pressure method has the disadvantages of great damage to the patient, difficult clinical operation, and high cost, so that the present invention can measure the pressure difference non-invasively and quickly, which reduces the difficulty of clinical operation and saves the cost of diagnosis.

Second, since the method of the present invention adopts the method of segmenting the aorta in the original chest CT data, extracting features from the three-dimensional aortic model, and classifying the degree of aortic coarctation, the extraction of CT image features is realized, and a more accurate aortic coarctation can be obtained. The pressure difference value at the stenosis overcomes the deficiency that the existing technology can only qualitatively estimate the severity of aortic coarctation from the CT morphological features, but cannot quantitatively evaluate it, so that the present invention greatly improves the use of CT images in diagnosing aortic coarctation. accuracy.

Description of drawings

Fig. 1 is the block diagram of device of the present invention;

Fig. 2 is a flow chart of the inventive method;

Fig. 3 is the flowchart of segmentation aortic image in the method of the present invention;

Fig. 4 is the segmented slice image matrix result figure in the method of the present invention, wherein the black line mark region image is the region image in the aortic tube obtained after segmentation;

Fig. 5 is the three-dimensional reconstruction of the segmented slice in the method of the present invention, and the obtained three-dimensional aortic model after rendering.

detailed description

The present invention will be described in detail below in conjunction with the accompanying drawings.

Referring to accompanying drawing 1, the device of the present invention is clearly and completely described.

The device of the present invention includes a data reading module, a two-dimensional slicing module, a segmentation module, a three-dimensional reconstruction module, an aortic coarctation discrimination module, a feature extraction module, a classification module and a result display module.

The data reading module is used to read in the original chest CT data in the format of .dcm or .raw.

The two-dimensional slice module is used to map the read-in raw chest CT data into two-dimensional slice image matrices of pixels with different gray values.

The segmentation module is used to select the slice image matrix comprising the end of the descending aorta from the two-dimensional slice image matrix, select the slice image matrix comprising the end of the descending aorta, and take the center point of the selected slice image matrix as the center to generate n* A square frame of n centimeters, the size of n is between one-fifth and one-fourth of the length of the image matrix; use the formula S[i,j]=I[p,q]×G[0.25] to calculate the smoothness The gray value at the coordinates of each pixel point in the final image is used to remove the noise of the two-dimensional slice image. Using medical image software, the edge of the blood vessel at the end of the descending aorta in the square frame is extracted to obtain the edge image of the blood vessel wall; the edge image surrounded by the edge of the blood vessel wall The gray value of all internal points is set to 255 to form an image of the internal area of the blood vessel, using The formula calculates the centroid of the image of the inner area of the blood vessel; loads the slice image matrix of the lowest layer other than the slice image matrix of the marked blood vessel area, determines a seed point element on the slice image matrix, and the abscissa of the seed point element is consistent with the slice image of the upper layer The abscissa of the center of mass of the inner area of the matrix blood vessel is equal, and the ordinate is equal to the ordinate of the center of mass of the inner area of the blood vessel in the upper slice image matrix. The edge of the blood vessel wall image is used as the boundary, and the seed point is determined by the element coordinates of the seed point to perform 8 Neighborhood adaptive areas grow to obtain the image of the aortic region on the sliced image matrix; where, S[i,j] represents the gray value of the pixel at [i,j] in the image after smoothing, and I[p,q] represents the two The gray value of the pixel at [p,q] in the three-dimensional slice image, G[0.25] represents a Gaussian function with a standard deviation of 0.25, M represents the centroid point of the image of the inner region of the blood vessel, ∫ represents the integral operation, f(x , y) represents the gray value at the image pixel point (x, y) of the inner area of the blood vessel.

The three-dimensional reconstruction module is used for importing the image of the inner region of the aorta on the slice image matrix into a three-dimensional reconstruction software for three-dimensional volume rendering, and obtaining a three-dimensional aortic model after rendering.

The aortic coarctation discrimination module is used to measure the aortic diameter at intervals of 1 mm in the three-dimensional aortic model, compare the aortic diameters of two adjacent measurements, and store the ratio results in the diameter ratio statistics table In , if the values in the diameter ratio statistical table are greater than 0.8, there is narrowing, otherwise, there is no narrowing.

The feature extraction module is used to import the three-dimensional aorta model into the medical image processing software, and use the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1; when the coarctation of the aorta is located at the position of the ascending aorta, the identification value is assigned When the aortic coarctation is located at the position of the descending aorta, the logo is assigned as -1, and when the aortic coarctation is located at the aortic arch, the logo is assigned a value of 1, and the assigned logo is used as the value of feature 2; the three-dimensional aorta The arterial model is imported into the medical image processing software, and the diameter of the narrowest part of the aortic vessel measured by the software is used as the value of feature 3; the three-dimensional aortic model is imported into the medical image processing software, and the most narrowed part measured and calculated by the software The ratio of the diameter to the diameter of the descending aorta is used as the value of feature 4; the ratio of aortic coarctation area is calculated using the formula R1=πR ² /(SQRT((H×W)/3600), and the calculation result is used as the value of feature 5; Use the formula R2=D/SQRT(SQRT((H×W)/3600) to calculate the aortic coarctation ratio, and use the calculation result as the value of feature 6; where, R1 represents the coarctation area ratio of the aortic image, and π represents PI, R represents the radius of the narrowest part of the aorta image, / represents the division operation, SQRT represents the square root operation, H represents the patient's height, W represents the patient's weight, R2 represents the constriction ratio of the aorta image, D represents the center of the aorta image The diameter of the narrowest point.

The classification module is used to input the 6 eigenvalues into the main systolic pressure difference model, and the main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues.

The result display module is used to display the pressure difference value at the aortic constriction obtained by the main systolic pressure difference model.

The method of the present invention is described in further detail with reference to accompanying drawing 2.

Step 1, read CT images.

The data reading module reads in the original chest CT data in the format of .dcm or .raw.

Step 2, perform two-dimensional slice processing on the CT data.

The two-dimensional slicing module maps the read-in original chest CT data into two-dimensional slice image matrices of pixels with different gray values, and uses 3DMed software to map the original chest CT data into two-dimensional slice image matrices according to the gray values.

Step 3, segment the aortic image.

In the first step, the segmentation module selects the slice image matrix containing the end of the descending aorta from the two-dimensional slice image matrix, and generates a square frame of n*n cm centered on the center point of the selected slice image matrix, and the size of n is Between one-fifth and one-fourth the length of the image matrix.

The second step is to use the Gaussian smoothing formula to calculate the gray value at the coordinates of each pixel point of the smoothed image, remove the noise of the two-dimensional slice image, and use medical image software to extract the edge of the descending aorta end vessel in the square frame to obtain the vessel wall edge image.

The Gaussian smoothing formula is as follows:

S[i,j]=I[p,q]×G[0.25]

Among them, S[i,j] represents the gray value of the pixel at [i,j] in the smoothed image, and I[p,q] represents the pixel at [p,q] in the two-dimensional slice image The gray value of G[0.25] represents a Gaussian function with a standard deviation of 0.25.

In the third step, the gray value of all internal points surrounded by the edge image of the vessel wall is set to 255 to form an image of the inner region of the vessel, and the centroid of the image of the inner region of the vessel is calculated using the Image Moments formula.

The Image Moments formula is as follows:

Among them, M represents the centroid point of the image of the inner region of the blood vessel, ∫ represents the integration operation, and f(x, y) represents the gray value at the pixel point (x, y) of the image of the inner region of the blood vessel.

The fourth step is to load the lowest slice image matrix other than the slice image matrix of the marked blood vessel area, and determine a seed point element on the loaded slice image matrix, whose abscissa is the same as the centroid of the inner area of the blood vessel in the upper slice image matrix. The coordinates are equal, the ordinate is equal to the centroid ordinate of the inner region of the blood vessel in the upper slice image matrix, and the edge of the vessel wall image obtained in the third step is used as the boundary, and the seed point is determined by the element coordinates of the seed point to perform 8-neighborhood adaptive region growth, Obtain an image of the aortic intravascular area on the slice image matrix.

The fifth step is to judge whether all slice image matrices have been loaded, if so, go to step 4, otherwise, go to step 4.

Step 4, measure the aortic diameter ratio.

The three-dimensional reconstruction module imports the image of the inner region of the aorta on the slice image matrix into the three-dimensional reconstruction software for three-dimensional volume rendering, and obtains the three-dimensional aortic model after rendering.

In the three-dimensional aortic model, the aortic diameter was measured at intervals of 1 mm.

The aortic diameters of two adjacent measurements are compared respectively, and the ratio results are stored in the diameter ratio statistics table.

Step 5, the aortic coarctation discrimination module judges whether there is a value less than 0.8 in the diameter ratio statistical table, if yes, execute step (6), otherwise, execute step (7).

Step 6, feature extraction from the 3D aorta model.

The feature extraction module imports the three-dimensional aorta model into the medical image processing software, and uses the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1.

Assign a value of 0 when the coarctation of the aorta is located in the ascending aorta, assign a value of -1 when the coarctation of the aorta is located in the descending aorta, and assign a value of 1 when the coarctation of the aorta is located in the aortic arch. Use the assigned identifier as the value of feature 2.

The three-dimensional aorta model was imported into the medical image processing software, and the diameter of the narrowest part of the aortic vessel measured by the software was used as the value of feature 3.

The three-dimensional aorta model was imported into the medical image processing software, and the ratio of the diameter of the narrowest part measured and calculated by the software to the diameter of the descending aorta was used as the value of feature 4.

Using the narrowing area ratio formula, calculate the aortic narrowing area ratio of the aortic image, and use the calculation result as the value of feature 5.

The formula for the narrowing area ratio is as follows:

R1＝πR ² /(SQRT((H×W)/3600)

Among them, R1 represents the constriction area ratio of the aortic image, π represents the circumference ratio, R represents the radius of the most constricted part of the aortic image, / represents the division operation, SQRT represents the square root operation, H represents the patient’s height, and W represents the patient’s weight.

Using the narrowing ratio formula, calculate the narrowing ratio of the aortic image, and use the calculation result as the value of feature 6.

The formula for the narrowing ratio is as follows:

R2＝D/SQRT(SQRT((H×W)/3600)

Among them, R2 represents the constriction ratio of the aortic image, and D represents the diameter of the most constricted part in the aortic image.

Step 7, classify the degree of aortic coarctation:

The case data of 500 patients with coarctation of the aorta were collected from the hospital database to form a training set. The data of each case included the patient's CT image, body length, weight, and pressure difference at the coarctation of the aorta.

Using the method in Step 6, quantitatively calculate the six characteristics of the patient.

The label value of the training data is determined according to the pressure difference at the coarctation of the aorta of the patient. Specifically, the pressure difference is from 0-5mmHg, 6-10mmHg, 10-15mmHg...95-100mmHg, and the label values are 1, 2, 3…20.

The characteristics and corresponding label values of all cases are used as the training set and input into the classifier. After all 500 cases of data are imported, run the classifier to obtain the classification model model.

Select 100 cases as the test set, extract six features of each test patient, input them into the model model of the classifier for classification, compare the obtained results with the actual pressure difference of the case, and count the classification accuracy and accuracy Above 80% indicates that the model has good usability.

The classification module inputs 6 eigenvalues into the main compression difference model.

The main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues. Specifically, the specific pressure difference value is determined according to the output classification results of the main systolic pressure difference model. The output result is 0, indicating that the patient The pressure difference of aortic coarctation is 0-5mmHg, the output of 1 means that the patient's aortic pressure difference is 5-10mmHg, and so on, the output of 20 means that the patient's aortic pressure difference is 95-100mmHg.

In step 8, the result display module displays the pressure difference value at the coarctation of the aorta obtained from the main systolic pressure difference model.

Claims (7)

1. A device for determining the pressure difference at the coarctation of the aorta from CT images, including a data reading module, a two-dimensional slice module, a segmentation module, a three-dimensional reconstruction module, aortic coarctation discrimination module, feature extraction module, classification module, The results show modules where: The data reading module is used to read in the original chest CT data whose format is .dcm or .raw; The two-dimensional slicing module is used to map the read-in original chest CT data into two-dimensional sliced image matrices of pixels with different gray values; The segmentation module is used to select a slice image matrix including the end of the descending aorta from the two-dimensional slice image matrix, select a slice image matrix including the end of the descending aorta, and take the center point of the selected slice image matrix as the center, Generate a square frame of n*n centimeters, the size of n is between one-fifth and one-fourth of the length of the image matrix; use the formula S[i,j]=I[p,q]×G[0.25] , calculate the gray value at the coordinates of each pixel in the smoothed image, remove the noise of the two-dimensional slice image, use the medical image software to extract the edge of the blood vessel at the end of the descending aorta in the square frame, and obtain the edge image of the blood vessel wall; the edge image of the blood vessel wall The gray value of all the surrounding points is set to 255 to form an image of the inner area of the blood vessel, using The formula calculates the centroid of the image of the inner area of the blood vessel; loads the slice image matrix of the lowest layer other than the slice image matrix of the marked blood vessel area, determines a seed point element on the slice image matrix, and the abscissa of the seed point element is consistent with the slice image of the upper layer The abscissa of the center of mass of the inner area of the matrix blood vessel is equal, and the ordinate is equal to the ordinate of the center of mass of the inner area of the blood vessel in the upper slice image matrix. The edge of the blood vessel wall image is used as the boundary, and the seed point is determined by the element coordinates of the seed point to perform 8 Neighborhood adaptive areas grow to obtain the image of the aortic region on the sliced image matrix; where, S[i,j] represents the gray value of the pixel at [i,j] in the image after smoothing, and I[p,q] represents the two The gray value of the pixel at [p,q] in the three-dimensional slice image, G[0.25] represents a Gaussian function with a standard deviation of 0.25, M represents the centroid point of the image of the inner region of the blood vessel, ∫ represents the integral operation, f(x , y) represents the gray value at the image pixel point (x, y) of the inner region of the blood vessel; The three-dimensional reconstruction module is used to import the image of the aortic region on the slice image matrix into three-dimensional reconstruction software for three-dimensional volume rendering, and obtain a three-dimensional aortic model after rendering; The aortic coarctation discrimination module is used to measure the aortic diameter at intervals of 1 mm in the three-dimensional aortic model, compare the aortic diameters measured twice adjacently, and store the ratio result in the diameter In the ratio statistical table, if the values in the diameter ratio statistical table are greater than 0.8, there is narrowing, otherwise, there is no narrowing; The feature extraction module is used to import the three-dimensional aorta model into medical image processing software, and use the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1; when the coarctation of the aorta is located at the position of the ascending aorta Assign the value of 0 to the flag of the aortic coarctation at the position of the descending aorta, assign a value of -1 to the flag of the coarctation of the aorta at the position of the aortic arch, and assign a value of 1 to the flag of the coarctation of the aorta at the position of the aortic arch, and use the assigned flag as the value of feature 2; Import the three-dimensional aortic model into the medical image processing software, and use the diameter of the narrowest part of the aortic vessel measured by the software as the value of feature 3; import the three-dimensional aortic model into the medical image processing software, and use the software to measure and calculate the most The ratio of the diameter of the constriction to the diameter of the descending aorta was taken as the value of feature 4; the ratio of aortic coarctation area was calculated using the formula R1=πR ² /(SQRT((H×W)/3600), and the calculation result was taken as feature 5 The value of R2=D/SQRT(SQRT((H×W)/3600) formula to calculate the aortic coarctation ratio, and use the calculation result as the value of feature 6; wherein, R1 represents the coarctation area ratio of the aortic image , π represents the circumference ratio, R represents the radius of the most constricted part of the aortic image, / represents the division operation, SQRT represents the square root operation, H represents the patient's height, W represents the patient's weight, R2 represents the constriction ratio of the aortic image, D represents the aorta The diameter of the narrowest point in the arterial image; The classification module is used to input 6 eigenvalues into the main systolic pressure difference model, and the main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues; The result display module is used to display the pressure difference value at the coarctation of the aorta obtained from the main systolic pressure difference model. 2. A method for determining the pressure difference at the coarctation of the aorta by CT images, comprising the steps of: (1) Read CT data: The data reading module reads in the original chest CT data in the format of .dcm or .raw; (2) Perform two-dimensional slice processing on CT data: The two-dimensional slicing module maps the read-in original chest CT data into two-dimensional sliced image matrices of pixels with different gray values; (3) Segment the aortic image: (3a) The segmentation module selects the slice image matrix containing the end of the descending aorta from the two-dimensional slice image matrix, and takes the center point of the selected slice image matrix as the center to generate a square frame of n*n centimeters, and the size of n is between Between one-fifth and one-fourth of the length of the image matrix; (3b) Using the Gaussian smoothing formula, calculate the gray value at the coordinates of each pixel point of the smoothed image, remove the noise of the two-dimensional slice image, and use the medical image software to extract the edge of the descending aorta end vessel in the square frame to obtain the edge of the vessel wall image; (3c) Set the gray value of all internal points surrounded by the edge image of the blood vessel wall to 255 to form an image of the internal area of the blood vessel, and calculate the centroid of the image of the internal area of the blood vessel by using the Image Moments formula; (3d) loading the lowest slice image matrix other than the slice image matrix of the marked blood vessel area, and using the aorta slice image segmentation method to obtain the image of the aortic intravascular area on the slice image matrix; (3e) judging whether all slice image matrices have been loaded, if so, perform step (4), otherwise, perform step (3d); (4) Measure the aortic diameter ratio: (4a) The three-dimensional reconstruction module imports the image of the inner region of the aorta on the slice image matrix into the three-dimensional reconstruction software for three-dimensional volume rendering, and obtains the three-dimensional aortic model after rendering; (4b) In the three-dimensional aorta model, the aortic diameter was measured every 1mm interval; (4c) compare the aortic diameters of two adjacent measurements respectively, and store the ratio results in the diameter ratio statistics table; (5) Whether the aortic coarctation discrimination module judges whether there is a value less than 0.8 in the diameter ratio statistical table, if so, then perform step (6), otherwise, perform step (7); (6) Extract features from the three-dimensional aorta model: (6a) The feature extraction module imports the three-dimensional aorta model into the medical image processing software, and uses the maximum gradient dmax value of the aortic vessel calculated by the software as the value of feature 1; (6b) Assign a value of 0 to the flag when the coarctation of the aorta is located at the position of the ascending aorta, assign a value of -1 to the flag when the coarctation of the aorta is located at the position of the descending aorta, and assign a value to the flag when the coarctation of the aorta is located at the position of the aortic arch is 1, and the assigned identifier is used as the value of feature 2; (6c) Import the three-dimensional aorta model into the medical image processing software, and use the diameter of the narrowest part of the aortic vessel measured by the software as the value of feature 3; (6d) Import the three-dimensional aorta model into the medical image processing software, and use the ratio of the diameter of the narrowest part measured and calculated by the software to the diameter of the descending aorta as the value of feature 4; (6e) Using the narrowing area ratio formula to calculate the aortic narrowing area ratio of the aortic image, and use the calculation result as the value of feature 5; (6f) using the narrowing ratio formula to calculate the narrowing ratio of the aortic image, and use the calculation result as the value of feature 6; (7) Classify the degree of aortic coarctation: (7a) The classification module inputs 6 eigenvalues into the main compression difference model; (7b) The main systolic pressure difference model outputs the pressure difference value at the coarctation of the aorta corresponding to the 6 eigenvalues; (8) The result display module displays the pressure difference value at the coarctation of the aorta obtained from the main systolic pressure difference model. 3. the method for determining the pressure difference at the coarctation of the aorta by CT image according to claim 2, is characterized in that, the Gaussian smoothing formula described in step (3b) is as follows: S[i,j]=I[p,q]×G[0.25] Among them, S[i,j] represents the gray value of the pixel at [i,j] in the smoothed image, and I[p,q] represents the pixel at [p,q] in the two-dimensional slice image The gray value of G[0.25] represents a Gaussian function with a standard deviation of 0.25. 4. the method for determining the pressure difference at the coarctation of the aorta by the CT image according to claim 2, is characterized in that, the Image Moments formula described in the step (3c) is as follows: $\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&infin;</span>}}^{\mathrm{<span\; class="notranslate">\; \&infin;</span>}}{<span\; class="notranslate">\; \&Integral;</span>}_{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&infin;</span>}}^{\mathrm{<span\; class="notranslate">\; \&infin;</span>}}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}$ Among them, M represents the centroid point of the image of the inner region of the blood vessel, ∫ represents the integration operation, and f(x, y) represents the gray value at the pixel point (x, y) of the image of the inner region of the blood vessel. 5. the method for determining the pressure difference at the coarctation of the aorta by the CT image according to claim 2, is characterized in that, the steps of the aortic slice image segmentation method described in the step (3d) are as follows: The first step is to determine a seed point element on the slice image matrix. The abscissa of the seed point element is equal to the abscissa of the center of mass of the inner region of the blood vessel in the upper slice image matrix, and the ordinate is equal to the ordinate of the centroid of the inner region of the blood vessel in the upper slice image matrix. equal; In the second step, the edge of the blood vessel wall image is used as the boundary, and the seed point is determined by the element coordinates of the seed point to perform 8-neighborhood adaptive region growth. 6. the method for determining the pressure difference at the coarctation of the aorta by CT image according to claim 2, is characterized in that, the narrowing area ratio formula described in step (6e) is as follows: R1＝πR ² /(SQRT((H×W)/3600) Among them, R1 represents the constriction area ratio of the aortic image, π represents the circumference ratio, R represents the radius of the most constricted part of the aortic image, / represents the division operation, SQRT represents the square root operation, H represents the patient’s height, and W represents the patient’s weight. 7. the method for determining the pressure difference at the coarctation of the aorta by CT image according to claim 2, is characterized in that, the coarctation ratio formula described in step (6f) is as follows: R2＝D/SQRT(SQRT((H×W)/3600) Among them, R2 represents the constriction ratio of the aortic image, and D represents the diameter of the most constricted part in the aortic image.