CN110313990B - Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery - Google Patents
Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery Download PDFInfo
- Publication number
- CN110313990B CN110313990B CN201910563810.0A CN201910563810A CN110313990B CN 110313990 B CN110313990 B CN 110313990B CN 201910563810 A CN201910563810 A CN 201910563810A CN 110313990 B CN110313990 B CN 110313990B
- Authority
- CN
- China
- Prior art keywords
- features
- bridge
- shear stress
- wall shear
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B34/00—Computer-aided surgery; Manipulators or robots specially adapted for use in surgery
- A61B34/10—Computer-aided planning, simulation or modelling of surgical operations
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B34/00—Computer-aided surgery; Manipulators or robots specially adapted for use in surgery
- A61B34/10—Computer-aided planning, simulation or modelling of surgical operations
- A61B2034/101—Computer-aided simulation of surgical operations
- A61B2034/105—Modelling of the patient, e.g. for ligaments or bones
Landscapes
- Health & Medical Sciences (AREA)
- Surgery (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Robotics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
A method for establishing a model for predicting the permeability of a bridge vessel based on wall shear stress image characteristics in a heart bypass surgery belongs to the field of model establishment. The method comprises the steps of carrying out digital processing on bridge blood vessel instant blood flow waveforms measured in the operation, enabling the bridge blood vessel instant blood flow waveforms to be assigned to a three-dimensional model of the bridging operation as boundary conditions, obtaining a wall surface shear stress cloud picture of an anastomotic site through a finite element method, extracting color features and texture features of an image, then carrying out dimension reduction processing on the features by using a principal component analysis method, and then using the features after dimension reduction for constructing a prediction model based on a support vector machine. The method can be used for helping doctors and patients to know the effect of the operation, and determining the strategy of further operation or the scheme of postoperative review.
Description
The technical field is as follows:
the invention belongs to the technical field of model establishment, and essentially belongs to the field of machine learning. In particular to a method for establishing a model for predicting the permeability of a bridge vessel based on wall shear stress image characteristics in a heart bypass surgery.
Background art:
coronary bypass surgery (CABG) is a common surgical procedure for treating coronary heart disease, and the main problem is that the bridge vessels are at risk of failure after the operation. According to statistics, the failure rate of the vein-bridge blood vessel in the early postoperative period is 15-30%, and the failure rate after 10 years reaches 50%. The 10-year patency rate of the artery-bridge vessels is 95%, and the 15-year patency rate of the artery-bridge vessels is 88%. After the bridge vessel fails, the patient can suffer myocardial ischemia again, and can be life-threatening in severe cases. Therefore, for each specific patient, how to predict the permeability of the bridged vessel is of great significance. Can help patients avoid possible adverse cardiovascular events in the future and can also help doctors plan the next treatment plan.
Currently, common prediction methods are based on clinical data obtained by instant blood flow measurement, such as average flow, pulsatility index, diastolic flow ratio, and the like, and learners also obtain some new features based on flow waveforms by using a signal processing mode, such as a ratio of fundamental waves to harmonic waves of the waveforms after fourier transform. However, no one has proposed to construct a predictive model using features based on hemodynamics.
According to research, hemodynamics is a critical factor affecting the permeability of the bridged vessels. The failure of the bridged vessels is mainly caused by atherosclerosis and intimal hyperplasia of the vessels, and poor hemodynamic factors are considered to be the most important factors in their occurrence and development. Especially for the wall shear stress, the wall shear stress is too low or too high, and the wall shear stress gradient is too high, which can cause the permeability of the bridge vessel to be reduced. After the model is calculated, the wall shear stress cloud picture can visually reflect the magnitude and the gradient of the wall shear stress. The bridge vessel anastomosis is the most main occurrence area of atherosclerosis and intimal hyperplasia, so that the wall shear stress characteristic at the anastomosis is taken into consideration, and the method has important significance for improving the prediction accuracy of the model.
The invention content is as follows:
compared with the prior art, the method has the advantage that a wall shear stress factor which is closely related to permeability is introduced, so that the accuracy of the constructed prediction model is improved under the condition that the number of samples is limited.
The technical scheme is as follows:
a method for establishing a predicted bridge vascular permeability model based on wall shear stress image characteristics in a heart bypass surgery is characterized by comprising the following steps:
(1) digitalizing the waveform picture of the intraoperative instant blood flow measurement;
(2) constructing a three-dimensional model of the bypass operation;
(3) extracting and reducing the dimension of wall shear stress image features of an anastomotic stoma part in the three-dimensional model of the bypass surgery;
(4) and (5) constructing a bridge vascular permeability prediction model.
Wherein the step (1) specifically comprises the following steps:
1.1 intercepting a waveform picture of each bridge blood vessel from a postoperative instant blood flow measurement report, and paying attention to avoid intercepting character parts in the report as much as possible;
1.2 processing the colorful waveform picture into a gray picture, selecting a certain gray value range, and distinguishing the waveform from background colors and coordinate axes in the picture;
1.3 setting a starting point moment and an end point moment for an abscissa in the picture, and solving a time step represented by each pixel point on the abscissa;
1.4 for each time step, calculating the number of pixel points spaced between the waveform and the abscissa, setting the maximum value of the waveform, and solving the flow value represented by each longitudinal pixel point, so that a bridge blood vessel flow sequence based on the time step length can be obtained;
1.5 after the waveform is stabilized, selecting a flow waveform of one period, and extracting a flow sequence of the flow waveform to carry out the next work. See, for example, fig. 1.
Wherein, the step (2) of bridging the three-dimensional model of the operation comprises the following specific parameters:
2.1 constructing an ideal model of the bypass surgery, and respectively simulating a coronary artery part and a bridge blood vessel part by using a straight circular tube, wherein the intersection angle of the bridge blood vessel and the coronary artery is 45 degrees;
2.2 the length of the blood vessel of the bridge is set to be 120mm, the length of the coronary vessel is set to be 100mm, the front end of an anastomotic stoma in the coronary is 10mm, and the rear end of the anastomotic stoma is 90 mm; the front and back are consistent with the blood flow direction;
2.3 the diameter of the coronary artery is set to be 3mm, and the diameter of the bridge vessel is set according to the caliber of an ultrasonic probe used by the instant blood flow measuring instrument;
2.4 based on the center position of the anastomotic stoma, cutting a square with the side length of 15mm, wherein the area in the square is the area of the wall shear stress image of the anastomotic stoma to be extracted;
2.5 applying the flow waveform of one cycle extracted in step 1.5 as an inlet flow boundary condition to the entrance of the bridge vessel, and setting 0mmHg as an outlet pressure boundary condition at the outlet of the coronary artery. See, for example, fig. 2.
Wherein the step (3) specifically comprises the following steps:
3.1, equally dividing a period into 10 moments, and extracting a wall shear stress cloud chart of the anastomotic stoma part at each moment, wherein the range of the cloud chart is fixed at 0-1.2 Pa. See, for example, fig. 3.
3.2 extracting color features and texture features of the wall shear stress cloud chart at each moment respectively. For color features, a color gray histogram method is used to find the features. The image is decomposed into three color channels of R, G and B, and for each color channel, six characteristics of mean value, variance, standard deviation, skewness, kurtosis, energy and entropy are obtained. And for texture features, a gray level co-occurrence matrix method is used for solving the features. Converting the color image into a gray image, setting four scanning directions of the gray co-occurrence matrix as (0,1), (-1,1), (-1,0), (-1, -1), and obtaining the gray co-occurrence matrix in the four directions. And solving four characteristics of contrast, correlation, energy and homogeneity of each gray level co-occurrence matrix. In summary, there are 34 features for each time, and each bridge vessel extracts ten times, and there are 340 features in total, which are arranged in sequence.
3.3 using principal component analysis method to reduce the dimension of the extracted 340 features, finally obtaining 19 features after dimension reduction processing for each bridge vessel, wherein the 19 features can represent more than 90% of information content.
Wherein the step (4) comprises the following steps:
4.1, constructing a bridge vascular permeability prediction model by using a support vector machine, selecting a Radial Basis Function (RBF) kernel function, searching for an optimal penalty coefficient (C) and a kernel function radius (g), and establishing the bridge vascular permeability prediction model by using the wall shear stress image characteristics of the anastomotic stoma part;
4.2 when finding the optimal penalty coefficient and kernel function radius pair (C, g) mentioned in step 2.1, using the method of grid search and cross validation, detecting the value of each C and g one by one according to a certain step length (which can be selected according to the requirement, without a certain limit) in a certain range, calculating the average accuracy of the (C, g) pair searched in each step by using a k-fold cross validation method, and finally taking the (C, g) pair with the highest classification accuracy;
4.3 the bridge vessel data as training sample includes instant blood flow measurement data and postoperative real Computer Tomography (CTA) data, marking the training sample as unobstructed or invalid according to the postoperative CTA data, training the support vector machine by using the image characteristics of the tangential stress of the anastomotic orifice wall after dimensionality reduction, and obtaining the prediction model of the bridge vessel permeability.
The method can well establish a predicted bridge vascular permeability model, and the predicted bridge vascular permeability model established according to the method has high accuracy and can be used for helping doctors and patients to know the operation effect and determining the strategy of further operation or the scheme of postoperative reexamination.
Description of the drawings:
FIG. 1: a digitized schematic of an instantaneous blood flow waveform;
FIG. 2: the three-dimensional model of the bypass operation is characterized in that a square frame part in the drawing is a wall surface shear stress cloud picture extraction area;
FIG. 3: wall shear stress cloud pictures extracted at different times.
The specific implementation mode is as follows:
the present invention will be further illustrated with reference to the following examples, but the present invention is not limited to the following examples.
Example 1
And (3) intercepting the waveform of each blood bridge in the instant blood flow waveform picture, carrying out digital processing on the waveform, and constructing a three-dimensional model conforming to the blood bridge according to the type of the blood bridge and the size of the ultrasonic probe. The digitized flow waveform was added to the entrance of the bridge vessel, a pressure of 0mmHg was added to the exit of the coronary as a boundary condition, and calculations were performed using finite element software.
And after calculation, extracting image characteristics of a wall surface shear stress cloud picture of the bridge vessel anastomotic stoma part, reducing dimensions, and constructing a prediction model by using a support vector machine with the image characteristics as characteristics. And (3) selecting an RBF kernel function, optimizing the (C, g) coefficients by using a grid search and cross validation method, and selecting the optimal (C, g) as the coefficient of the prediction model (see the steps of the invention content specifically).
Two types of prediction models are respectively constructed by using common clinical characteristics and wall shear stress image characteristics as characteristics, and the average accuracy, average sensitivity and average specificity of the two types of prediction models under the optimal (C, g) condition are obtained by a cross validation method so as to compare the performance of the prediction models. The results are shown in the following table:
from the results, it is known that, by using the wall shear stress image features as the features, the accuracy of the prediction model is improved, and the improvement of the sensitivity is more remarkable.
The method selects 61 bridge vessel data of 37 patients to test, and collects the intraoperative instant blood flow waveform picture data and postoperative CTA (CTA) review result of each bridge vessel. According to the rechecking result, 21 of the bridge vessels are failed, and 40 of the bridge vessels are unobstructed.
Claims (1)
1. A method for establishing a predicted bridge vascular permeability model based on wall shear stress image characteristics in a heart bypass surgery is characterized by comprising the following steps:
(1) extracting and reducing the dimension of wall shear stress image features of an anastomotic stoma part in the three-dimensional model of the bypass surgery;
(2) constructing a bridge vascular permeability prediction model;
wherein the step (1) specifically comprises the following steps:
(1.1) equally dividing a period into 10 moments, and extracting a wall shear stress cloud chart of an anastomotic stoma part at each moment, wherein the range of the cloud chart is fixed at 0-1.2 Pa;
(1.2) respectively extracting color features and texture features of the wall shear stress cloud chart at each moment;
for color features, a color gray histogram method is applied to solve the features;
decomposing the image into three color channels of R, G and B, and solving six characteristics of mean value, variance, standard deviation, skewness, kurtosis, energy and entropy of each color channel; for the texture features, a gray level co-occurrence matrix method is used for solving the features; converting the color image into a gray image, setting four scanning directions of a gray co-occurrence matrix as (0,1), (-1,1), (-1,0), (-1, -1), and solving the gray co-occurrence matrix in the four directions; for each gray level co-occurrence matrix, solving four characteristics of contrast, correlation, energy and homogeneity; in summary, there are 34 features at each time, and each bridge vessel extracts ten times, and there are 340 features in total, which are arranged in sequence;
(1.3) reducing the dimensions of the extracted 340 features by using a principal component analysis method, and finally obtaining 19 features subjected to dimension reduction processing for each blood bridge, wherein the 19 features can represent more than 90% of information content;
wherein the step (2) comprises the following steps:
(2.1) constructing a bridge vascular permeability prediction model by using a support vector machine, selecting a radial basis kernel function, searching for an optimal penalty coefficient (C) and a kernel function radius (g), and establishing the bridge vascular permeability prediction model by using the wall shear stress image characteristics of the anastomotic stoma part;
(2.2) when the optimal penalty coefficient and kernel function radius pair (C, g) mentioned in the step 2.1 are searched, detecting the value of each C and g one by one according to a certain step length in a certain range by using a method of grid searching and cross validation, calculating the average accuracy of the (C, g) pair searched in each step by using a method of k-fold cross validation, and finally taking the (C, g) pair with the highest classification accuracy;
(2.3) the bridge vessel data serving as the training sample comprises instant blood flow measurement data and real computer tomography data after operation, the training sample is marked as unobstructed or invalid according to the real computer tomography data after operation, and the support vector machine is trained by using the image characteristics of the tangential stress of the wall surface of the anastomotic stoma after dimension reduction to obtain a prediction model of the permeability of the bridge vessel.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910563810.0A CN110313990B (en) | 2019-06-26 | 2019-06-26 | Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910563810.0A CN110313990B (en) | 2019-06-26 | 2019-06-26 | Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110313990A CN110313990A (en) | 2019-10-11 |
CN110313990B true CN110313990B (en) | 2021-05-14 |
Family
ID=68120500
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910563810.0A Active CN110313990B (en) | 2019-06-26 | 2019-06-26 | Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110313990B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113609787A (en) * | 2021-08-27 | 2021-11-05 | 中国核动力研究设计院 | Electric valve fault diagnosis method, terminal equipment and readable storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104116563A (en) * | 2014-07-02 | 2014-10-29 | 北京工业大学 | Hemodynamic fast planning method for personalized coronary artery bypass graft surgery |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10482215B2 (en) * | 2014-10-22 | 2019-11-19 | Indiana University Research And Technology Corporation | Unified computational method and system for patient-specific hemodynamics |
DE102014222804B4 (en) * | 2014-11-07 | 2023-12-28 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Device and method for determining a wall shear stress and system for detecting arteriosclerosis |
CN106384114B (en) * | 2016-09-22 | 2019-09-27 | 哈尔滨理工大学 | A kind of adapting to image binarization method based on support vector machines |
CN109637657B (en) * | 2018-11-27 | 2022-04-15 | 北京工业大学 | Method for establishing prediction of vascular permeability bridge model in heart bypass surgery based on hemodynamics model |
CN109740692A (en) * | 2019-01-13 | 2019-05-10 | 胡燕祝 | A kind of target classifying method of the logistic regression based on principal component analysis |
-
2019
- 2019-06-26 CN CN201910563810.0A patent/CN110313990B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104116563A (en) * | 2014-07-02 | 2014-10-29 | 北京工业大学 | Hemodynamic fast planning method for personalized coronary artery bypass graft surgery |
Non-Patent Citations (3)
Title |
---|
《冠脉搭桥手术移植管即时通透性的血流动力学研究》;毛伯䶮,刘有军,王文馨,李鲍;《医用生物力学》;20180831;第33卷;324 * |
Hemodynamic analysis of sequential graft from right coronary system to left coronary system;Wenxin Wang;Boyan Mao;Haoran Wang;Xueying Geng;Xi Zhao;Huixia Z;《BioMedical Engineering OnLine》;20161231;第2卷(第15期);545-555页 * |
冠状动脉搭桥手术规划的血流动力学研究;赵夕;《中国博士学位论文全文数据库(电子期刊)》;20180630;1-50 * |
Also Published As
Publication number | Publication date |
---|---|
CN110313990A (en) | 2019-10-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109637657B (en) | Method for establishing prediction of vascular permeability bridge model in heart bypass surgery based on hemodynamics model | |
Li et al. | Progress in non-invasive detection of liver fibrosis | |
CN107945878B (en) | Method for constructing hepatic vein pressure gradient calculation model based on radiology group | |
Cornelis et al. | An overview of the applied definitions and diagnostic methods to assess exercise oscillatory ventilation—a systematic review | |
CN101732040A (en) | Non-invasive multipath pulse wave detection device, system and analytical system | |
WO2021259395A3 (en) | Method and apparatus for obtaining myocardial bridge image, and electronic device and storage medium | |
Atkinson | Shear rate normalization is not essential for removing the dependency of flow-mediated dilation on baseline artery diameter: past research revisited | |
CN110313990B (en) | Method for establishing prediction bridge vascular permeability model based on wall shear stress image characteristics in heart bypass surgery | |
Jian et al. | Diagnostic value of two dimensional shear wave elastography combined with texture analysis in early liver fibrosis | |
CN109841278A (en) | A method of the screening of coronary disease disease is carried out with angiocarpy mark and rote learning operation method | |
CN117100246A (en) | Health state analysis method based on dynamic bioelectric signals and related products | |
CN115579128A (en) | Multi-model feature-enhanced disease screening system | |
CN114176532B (en) | Clinical verification method for determining cfPWV parameters and application system thereof | |
Botha et al. | Comparing the associations of clinic vs. ambulatory blood pressure with subclinical organ damage in young healthy adults: the African-PREDICT study | |
CN115719329A (en) | Method and system for fusing RA ultrasonic modal synovial membrane scores based on deep learning | |
CN109461139B (en) | Liver cancer quantitative analysis method based on dynamic MRI information fusion | |
TWI685854B (en) | Liver fibrosis assessment model, liver fibrosis assessment system and liver fibrosis assessment method | |
CN108498088A (en) | A kind of instrument of detection and analysis cerebral artery function and state | |
CN117954083B (en) | Preeclampsia prediction method based on machine learning algorithm | |
CN110786839A (en) | Method, device, equipment and medium for generating instantaneous waveform-free ratio | |
JP2010068955A (en) | Ultrasonic diagnostic apparatus and data processing program of ultrasonic diagnostic apparatus | |
CN110706803A (en) | Method, device, readable medium and electronic equipment for determining myocardial fibrosis | |
CN111192244A (en) | Method and system for determining tongue characteristics based on key points | |
US11540782B2 (en) | Method, apparatus and computer program for measuring and analysis of blood pressure using photoplethysmography | |
CN114550942B (en) | Liver significant fibrosis prediction model and construction method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |