CN106803251B - The apparatus and method of aortic coaractation pressure difference are determined by CT images - Google Patents
The apparatus and method of aortic coaractation pressure difference are determined by CT images Download PDFInfo
- Publication number
- CN106803251B CN106803251B CN201710021582.5A CN201710021582A CN106803251B CN 106803251 B CN106803251 B CN 106803251B CN 201710021582 A CN201710021582 A CN 201710021582A CN 106803251 B CN106803251 B CN 106803251B
- Authority
- CN
- China
- Prior art keywords
- aorta
- image
- module
- value
- dimensional
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30048—Heart; Cardiac
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a kind of device and method for determining aortic coaractation pressure difference by CT images.The inventive system comprises: data read module, two dimension slicing module, segmentation module, three-dimensional reconstruction module, aortic coaractation discrimination module, characteristic extracting module, categorization module, result display module.Step of the invention includes: 1, reads CT data, 2, two dimension slicing processing is carried out to CT data, 3, divide aorta images, 4, aorta diameter ratio is measured, 5, discriminate whether that there are constrictions, 6, feature is extracted from three-dimensional aorta model, 7, classify to aortic coaractation degree, 8, pressure difference at display aortic coaractation.The present invention classifies to CT images morphological feature using machine learning algorithm, and accurate aortic coaractation pressure difference can be directly obtained by CT images, improves accuracy of the CT images in aortic coaractation diagnosis.
Description
Technical field
The invention belongs to electronic technology fields, further relate to medical image processing, artificial intelligence and machine learning skill
One of art field determines aortic coaractation pressure by computer tomography CT (Computed Tomography) image
The device and method of difference.The present invention is for aortic coaractation is diagnosed in clinical medicine when, the feelings quantified to pressure difference are needed
Shape realizes using the device for determining pressure difference at aortic coaractation and quantitatively obtains aortic coaractation both ends by CT images
Pressure difference.
Background technique
Aortic coaractation is a kind of common deformity in congenital heart disease, clinically diagnoses aortic coaractation by CT at present
Mainly by carrying out qualitative estimation to aortic coaractation severity from morphological feature, aorta is accurately assessed to realize
Constriction severity is needed to be had using Pressure wire and invasively protrudes into intralesional, and the pressure difference of rear and front end, this measurement are measured
Method not only has biggish damage to patient, and clinical manipulation difficulty is larger, while expense is also more expensive.
Institute of Semiconductors,Academia Sinica its application patent document " a kind of intracardiac pressure seal wire of optical fiber " (publication number:
CN105054916A, application number: CN201510609653.4, the applying date: on November 18th, 2015) in disclose a kind of optical fiber heart
Internal pressure seal wire.The device includes optical fiber and the pressure measuring cavity that is fixed on fiber end face, realizes and seal wire is protruded into inside of human body surveys
Measure pressure difference.Deficiency existing for the device is, it is necessary to device be stretched to inside of human body, the method for invasive side pressure has one to patient
Fixed damage, and expense is costly.
Matthew J.Budoff, A Shittu, S Roy, in paper " the Use of cardiovascular that it is delivered
computed tomography in the diagnosis and management of coarctation of the
Aorta " is described in (Journal of Thoracic&Cardiovascular Surgery, 2013,146 (1): 229-32)
By the method for CT images assessment aortic coaractation.This method carries out CT scan to the lesions position of patient first, and by CT images
It is directed into the image system of computer, secondly, the feature in CT images is read from computer, as: the ratio of constriction, constriction institute
Position etc., realize the qualitative evaluation to aortic coaractation.It is disadvantageous in that, can not quantitatively assess existing for this method
The degree of aortic coaractation cannot be obtained the pressure difference of aortic coaractation, the gold of this differentiation aortic coaractation by CT images
Standard.
Patent document " coronary blood based on cardiac CT image of the Suzhou Run Xin medical science and technology Co., Ltd in its application
Stream deposit score calculation method " (publication number: CN106023202A, application number: CN201610339892.7, the applying date: 2016
May 20) in disclose a kind of method that degree of stenosis is quantitatively determined by CT image.This method extracts the heart first
Then flesh image does edge detection to coronary artery volume data to coronary artery Accurate Segmentation, generate coronary artery triangle gridding
Model finally realizes the qualitative assessment to degree of stenosis.Shortcoming existing for this method is targeted blood vessel
Only coronary artery, coronary artery physiological structure and aorta gap are big, and coronary artery stenosis is often due to thrombus etc. blocks
And formed, and aortic coaractation is the change due to blood vessel physiological structure, the above method is not suitable for being used to assess aortic coaractation.
In conclusion existing method can only carry out "ball-park" estimate to aortic coaractation situation by CT images, to
The method that the pressure difference at both ends can only take invasive measurement at aortic coaractation, it is larger to the damage of patient, and expense compared with
For valuableness.
Summary of the invention
When the purpose of the present invention is for diagnosing aortic coaractation in clinical medicine, the feelings quantified to pressure difference are needed
Shape provides a kind of device and method for directly obtaining constriction pressure difference by CT images, can noninvasive, quantitative, rapidly basis
CT images assess the severity of aortic coaractation, provide strong auxiliary for the diagnosing and treating of doctor, and can reduce disease
The treatment cost of people.
The thinking for realizing the object of the invention is the aorta vessel being partitioned into the preoperative CT image of patients with congenital heart diseases, and will
It carries out three-dimensional visualization reconstruction, carries out aortic coaractation judgement to the obtained model of reconstruction, to be determined as the data of constriction into
Onestep extraction feature, and active rugosity is obtained by within the scope of Feature Mapping to specific pressure difference using the method for machine learning
The pressure difference at narrow place provides reference for clinician.
To achieve the above object, the inventive system comprises data read module, two dimension slicing module, segmentation modules, three
Dimension rebuilds module, aortic coaractation discrimination module, characteristic extracting module, categorization module, result display module, in which:
The data read module, for reading in the original thoracic CT data that format is .dcm or .raw；
The two dimension slicing module, for the original thoracic CT data read in be each mapped to different gray-value pixels
The two-dimensional slice image matrix of point；
The segmentation module, for sectioning image of the selection comprising descending aorta end from two-dimensional slice image matrix
Matrix, selection includes the sectioning image matrix of descending aorta end, raw centered on the central point of selected sectioning image matrix
At n*n centimetres of square-shaped frame, the size of n is between image array length 1/5th between a quarter；Using S [i,
J]=I [p, q] × G [0.25] formula, the gray value at each pixel coordinate of smooth rear image is calculated, two-dimensional slice image is removed
Noise extracts the descending aorta tail vein edge in square-shaped frame, obtains vascular wall edge graph using medical image software
Picture；The gray value for all internal points that vascular wall edge image is surrounded is set to 255, forms an internal blood vessel administrative division map
Picture utilizesFormula calculates the mass center of internal blood vessel area image；It is loaded into marked blood vessel
Lowest level sectioning image matrix except the sectioning image matrix of region determines a seed point element on sectioning image matrix,
The abscissa of the seed point element is equal with the mass center abscissa in upper layer sectioning image matrix internal blood vessel region, ordinate with it is upper
The mass center ordinate in layer sectioning image matrix internal blood vessel region is equal, using the edge of vascular wall image as boundary, with seed point
Element coordinate determines that seed point carries out the growth of 8 neighborhood adaptive regions, obtains the active vascular inner region on sectioning image matrix
Image；Wherein, it is located at the gray value of the pixel at [i, j] after S [i, j] indicates smooth in image, I [p, q] indicates that two dimension is cut
It is located at the gray value of the pixel at [p, q] in picture, G [0.25] indicates that standard deviation is 0.25 Gaussian function, and M indicates blood
The center of mass point of pipe images of interior regions, ∫ indicate integration operation, and f (x, y) is indicated at internal blood vessel area image pixel (x, y)
Gray value；
The three-dimensional reconstruction module, for the active vascular inner region image on sectioning image matrix to be imported into three-dimensional
Three-dimensional volume drawing, the three-dimensional aorta model after being drawn are carried out in reconstruction software；
The aortic coaractation discrimination module, for measuring primary active at interval of 1mm in three-dimensional aorta model
The adjacent aorta diameter measured twice is done compare respectively by arteries and veins diameter, and ratio result is stored in diameter than in statistical form, if
Diameter is all larger than 0.8 than the value in statistical form, then there is constriction, otherwise, constriction is not present；
The characteristic extracting module, for three-dimensional aorta model to be imported Medical Image Processing software, by software meter
Value of the greatest gradient dmax value of obtained aorta vessel as feature 1；Aortic coaractation is located at aorta ascendens position
When mark be assigned a value of 0, mark when aortic coaractation is located at descending aorta position is assigned a value of -1, and aortic coaractation is located at
Mark when arch of aorta position is assigned a value of 1, by value of the mark as feature 2 after assignment；Three-dimensional aorta model is imported
Medical Image Processing software, diameter is as the value of feature 3 at the most constriction for the aorta vessel that software measurement is obtained；It will be three-dimensional
Aorta model imports Medical Image Processing software, and diameter and descending aorta are straight at the most constriction that software is measured and is calculated
Value of the ratio of diameter as feature 4；Utilize R1=π R2/ (SQRT ((H × W)/3600) formula calculates aortic coaractation area
Than using calculated result as the value of feature 5；Using R2=D/SQRT, (SQRT ((H × W)/3600) formula calculates active rugosity
Narrow ratio, using calculated result as the value of feature 6；Wherein, R1 indicates the constriction area ratio of aorta images, and π indicates pi,
R indicates radius at aorta images most constriction ,/indicating divide operations, SQRT indicates evolution operation, and H indicates Patient height, W table
Show patient's weight, R2 indicates the constriction ratio of aorta images, diameter at most constriction in D expression aorta images；
The categorization module, for 6 characteristic values to be input in main contracting pressure difference model, the output of main contracting pressure difference model with
Pressure difference at the corresponding aortic coaractation of 6 characteristic values；
The result display module, for showing pressure difference at aortic coaractation that main contracting pressure difference model obtains.
The method of the present invention includes following steps:
(1) CT data are read:
Data read module reads in the original thoracic CT data that format is .dcm or .raw；
(2) two dimension slicing processing is carried out to CT data:
Two dimension slicing module by the original thoracic CT data of reading, cut by the two dimension for being each mapped to different gray-value pixel points
Picture matrix；
(3) divide aorta images:
(3a) divides module from two-dimensional slice image matrix, and selection includes the sectioning image matrix of descending aorta end,
Centered on the central point of selected sectioning image matrix, n*n centimetres of square-shaped frame is generated, the size of n is between image array
/ 5th of length are between a quarter；
(3b) utilizes Gaussian smoothing formula, calculates the gray value at each pixel coordinate of smooth rear image, and removal two dimension is cut
Picture noise extracts the descending aorta tail vein edge in square-shaped frame, obtains vascular wall side using medical image software
Edge image；
The gray value for all internal points that vascular wall edge image is surrounded is set to 255 by (3c), and composition one is intravascular
Portion's area image calculates the mass center of internal blood vessel area image using Image Moments formula；
(3d) is loaded into the lowest level sectioning image matrix except marked angiosomes sectioning image matrix, utilizes aorta
Sectioning image dividing method obtains the active vascular inner region image on sectioning image matrix；
(3e) judges whether loaded all sectioning image matrixes, if so, thening follow the steps (4), otherwise, executes step
(3d)；
(4) aorta diameter ratio is measured:
Active vascular inner region image on sectioning image matrix it is soft to be imported into three-dimensional reconstruction by (4a) three-dimensional reconstruction module
Three-dimensional volume drawing, the three-dimensional aorta model after being drawn are carried out in part；
(4b) measures an aorta diameter in three-dimensional aorta model, at interval of 1mm；
(4c), which respectively does the adjacent aorta diameter measured twice, to be compared, and ratio result deposit diameter is compared statistical form
In；
(5) aortic coaractation discrimination module judges diameter than whether there is the value less than 0.8 in statistical form, if so, holding
Row step (6) otherwise executes step (7)；
(6) feature is extracted from three-dimensional aorta model:
Three-dimensional aorta model is imported Medical Image Processing software, software is calculated by (6a) characteristic extracting module
Aorta vessel value of the greatest gradient dmax value as feature 1；
Mark when aortic coaractation is located at aorta ascendens position by (6b) is assigned a value of 0, and aortic coaractation is located at drop master
Mark when artery position is assigned a value of -1, and mark when aortic coaractation is located at arch of aorta position is assigned a value of 1, after assignment
Value of the mark as feature 2；
Three-dimensional aorta model is imported Medical Image Processing software by (6c), the aorta vessel that software measurement is obtained
Value of the diameter as feature 3 at most constriction；
Three-dimensional aorta model is imported Medical Image Processing software, the most constriction that software is measured and is calculated by (6d)
Locate value of the ratio of diameter and descending aorta diameter as feature 4；
(6e) utilize constriction area ratio formula, calculate aorta images aortic coaractation area ratio, using calculated result as
The value of feature 5；
(6f) utilizes constriction ratio formula, aorta images constriction ratio is calculated, using calculated result as the value of feature 6；
(7) classify to aortic coaractation degree:
6 characteristic values are input in main contracting pressure difference model by (7a) categorization module；
(7b) main contracting pressure difference model exports the pressure difference at aortic coaractation corresponding with 6 characteristic values；
(8) result display module shows pressure difference at aortic coaractation that main contracting pressure difference model obtains.
The present invention has the advantage that compared with prior art
First, since the device of the invention uses CT image segmentation module, CT image characteristics extraction module, CT image classification
Module realizes that CT images directly obtain pressure difference at accurate aortic coaractation, overcomes prior art seal wire manometric method
Big to patient's damage, clinical manipulation difficulty is big, the more expensive deficiency of expense, such that the present invention is noninvasive, rapidly measures
Pressure difference reduces clinical manipulation difficulty, saves diagnosis cost.
Second, due to method of the invention using the aorta divided in original thoracic CT data, from three-dimensional aorta mould
Feature is extracted in type, is classified to aortic coaractation degree, is realized and is extracted CT images feature, obtains accurate aorta
Pressure difference at constriction overcomes the prior art and is only capable of estimating from CT morphological feature is qualitative to the progress of aortic coaractation severity
It surveys, the deficiency that can not be quantitatively evaluated, so that the present invention substantially increases CT images for the accuracy of diagnosis aortic coaractation.
Detailed description of the invention
Fig. 1 is the block diagram of apparatus of the present invention；
Fig. 2 is the flow chart of the method for the present invention；
Fig. 3 is the flow chart for dividing aorta images in the method for the present invention；
Fig. 4 is segmentation sectioning image matrix result figure in the method for the present invention, and wherein black line identified areas image is after dividing
Obtained active vascular inner region image；
Fig. 5 is the three-dimensional reconstruction being sliced after dividing in the method for the present invention, three-dimensional aorta model after obtained drafting.
Specific embodiment
The present invention will be described in detail with reference to the accompanying drawing.
Referring to attached drawing 1, the device of the invention is clearly and completely described.
The inventive system comprises data read module, two dimension slicing module, segmentation modules, three-dimensional reconstruction module, active
The narrow discrimination module of rugosity, characteristic extracting module, categorization module, result display module.
Data read module, for reading in the original thoracic CT data that format is .dcm or .raw.
Two dimension slicing module, for the original thoracic CT data read in be each mapped to the two of different gray-value pixel points
Tie up sectioning image matrix.
Divide module, for selecting the sectioning image matrix comprising descending aorta end from two-dimensional slice image matrix,
Selection includes the sectioning image matrix of descending aorta end, centered on the central point of selected sectioning image matrix, generates n*n
Centimetre square-shaped frame, the size of n is between image array length 1/5th between a quarter；Utilize S [i, j]=I
[p, q] × G [0.25] formula calculates the gray value at each pixel coordinate of smooth rear image, removes two-dimensional slice image noise,
Using medical image software, the descending aorta tail vein edge in square-shaped frame is extracted, vascular wall edge image is obtained；By blood
The gray value for all internal points that tube wall edge image is surrounded is set to 255, forms an internal blood vessel area image, utilizesFormula calculates the mass center of internal blood vessel area image；It is loaded into marked angiosomes slice
Lowest level sectioning image matrix except image array determines a seed point element, the seed point on sectioning image matrix
The abscissa of element is equal with the mass center abscissa in upper layer sectioning image matrix internal blood vessel region, ordinate and upper layer slice map
As the mass center ordinate in matrix internal blood vessel region is equal, using the edge of vascular wall image as boundary, with seed point element coordinate
It determines that seed point carries out the growth of 8 neighborhood adaptive regions, obtains the active vascular inner region image on sectioning image matrix；Its
In, the gray value of the pixel after S [i, j] indicates smooth in image at [i, j], I [p, q] is indicated in two-dimensional slice image
The gray value of pixel at [p, q], G [0.25] indicate that standard deviation is 0.25 Gaussian function, and M indicates internal blood vessel area
The center of mass point of area image, ∫ indicate integration operation, and f (x, y) indicates the gray value at internal blood vessel area image pixel (x, y).
Three-dimensional reconstruction module, it is soft for the active vascular inner region image on sectioning image matrix to be imported into three-dimensional reconstruction
Three-dimensional volume drawing, the three-dimensional aorta model after being drawn are carried out in part.
Aortic coaractation discrimination module, for it is straight to measure an aorta at interval of 1mm in three-dimensional aorta model
The adjacent aorta diameter measured twice is done compare respectively by diameter, and ratio result is stored in diameter than in statistical form, if diameter
It is all larger than 0.8 than the value in statistical form, then there is constriction, otherwise, constriction is not present.
Software is calculated for three-dimensional aorta model to be imported Medical Image Processing software for characteristic extracting module
Aorta vessel value of the greatest gradient dmax value as feature 1；Aortic coaractation is located to mark when aorta ascendens position
Knowledge is assigned a value of 0, and mark when aortic coaractation is located at descending aorta position is assigned a value of -1, and aortic coaractation is located at aorta
Mark when bending position is assigned a value of 1, by value of the mark as feature 2 after assignment；Three-dimensional aorta model is imported into medicine figure
As processing software, diameter is as the value of feature 3 at the most constriction for the aorta vessel that software measurement is obtained；By three-dimensional aorta
Model imports Medical Image Processing software, the ratio of diameter and descending aorta diameter at the most constriction that software is measured and is calculated
It is worth the value as feature 4；Utilize R1=π R2/ (SQRT ((H × W)/3600) formula calculates aortic coaractation area ratio, will count
Calculate value of the result as feature 5；Using R2=D/SQRT (SQRT ((H × W)/3600) formula, calculate aortic coaractation ratio,
Using calculated result as the value of feature 6；Wherein, R1 indicates the constriction area ratio of aorta images, and π indicates that pi, R indicate master
Radius at arterial images most constriction ,/indicating divide operations, SQRT indicates evolution operation, and H indicates that Patient height, W indicate patient's body
Weight, R2 indicate the constriction ratio of aorta images, diameter at most constriction in D expression aorta images.
Categorization module, for 6 characteristic values to be input in main contracting pressure difference model, main contracting pressure difference model output and 6 spies
Pressure difference at the corresponding aortic coaractation of value indicative.
Result display module, for showing pressure difference at aortic coaractation that main contracting pressure difference model obtains.
It is described in further detail referring to 2 pairs of methods of the invention of attached drawing.
Step 1, CT images are read.
Data read module reads in the original thoracic CT data that format is .dcm or .raw.
Step 2, two dimension slicing processing is carried out to CT data.
Two dimension slicing module cuts the two dimension that the original thoracic CT data of reading is each mapped to different gray-value pixel points
Original thoracic CT data is mapped as two-dimensional slice image matrix according to gray value using 3DMed software by picture matrix.
Step 3, divide aorta images.
The first step divides module from two-dimensional slice image matrix, and selection includes the sectioning image square of descending aorta end
Battle array generates n*n centimetres of square-shaped frame, the size of n is between image moment centered on the central point of selected sectioning image matrix
/ 5th of array length degree are between a quarter.
Second step calculates the gray value at each pixel coordinate of smooth rear image, removal two dimension using Gaussian smoothing formula
Sectioning image noise extracts the descending aorta tail vein edge in square-shaped frame, obtains vascular wall using medical image software
Edge image.
The Gaussian smoothing formula is as follows:
S [i, j]=I [p, q] × G [0.25]
Wherein, it is located at the gray value of the pixel at [i, j] after S [i, j] indicates smooth in image, I [p, q] indicates two dimension
It is located at the gray value of the pixel at [p, q] in sectioning image, G [0.25] indicates that standard deviation is 0.25 Gaussian function.
The gray value for all internal points that vascular wall edge image is surrounded is set to 255, forms a blood vessel by third step
Images of interior regions calculates the mass center of internal blood vessel area image using Image Moments formula.
The Image Moments formula is as follows:
Wherein, M indicates the center of mass point of internal blood vessel area image, and ∫ indicates integration operation, and f (x, y) indicates internal blood vessel area
Gray value at area image pixel (x, y).
4th step, the lowest level sectioning image matrix being loaded into except marked angiosomes sectioning image matrix, is being loaded into
Sectioning image matrix on determine a seed point element, the matter of abscissa and upper layer sectioning image matrix internal blood vessel region
Heart abscissa is equal, and ordinate is equal with the mass center ordinate in upper layer sectioning image matrix internal blood vessel region, is obtained with third step
The edge of the vascular wall image arrived is boundary, determines that seed point carries out the growth of 8 neighborhood adaptive regions with seed point element coordinate,
Obtain the active vascular inner region image on sectioning image matrix.
5th step judges whether loaded all sectioning image matrixes, if so, thening follow the steps 4, otherwise, executes the 4th
Step.
Step 4, aorta diameter ratio is measured.
Active vascular inner region image on sectioning image matrix is imported into three-dimensional reconstruction software by three-dimensional reconstruction module
Carry out three-dimensional volume drawing, the three-dimensional aorta model after being drawn.
In three-dimensional aorta model, an aorta diameter is measured at interval of 1mm.
The adjacent aorta diameter measured twice is done respectively and is compared, and by ratio result deposit diameter than in statistical form.
Step 5, aortic coaractation discrimination module, judge diameter than in statistical form with the presence or absence of value less than 0.8, if so,
(6) are thened follow the steps, otherwise, are executed step (7).
Step 6, feature is extracted from three-dimensional aorta model.
Three-dimensional aorta model is imported Medical Image Processing software, the active that software is calculated by characteristic extracting module
Value of the greatest gradient dmax value of arteries and veins blood vessel as feature 1.
Mark when aortic coaractation is located at aorta ascendens position is assigned a value of 0, and aortic coaractation is located at descending aorta
Mark when position is assigned a value of -1, and mark when aortic coaractation is located at arch of aorta position is assigned a value of 1, by the mark after assignment
Know the value as feature 2.
Three-dimensional aorta model is imported into Medical Image Processing software, the most contracting for the aorta vessel that software measurement is obtained
Value of the narrow place's diameter as feature 3.
It is straight at the most constriction that software is measured and is calculated by three-dimensional aorta model importing Medical Image Processing software
Value of the ratio of diameter and descending aorta diameter as feature 4.
Using constriction area ratio formula, aorta images aortic coaractation area ratio is calculated, using calculated result as feature 5
Value.
The constriction area ratio formula is as follows:
R1=π R2/(SQRT((H×W)/3600)
Wherein, R1 indicates the constriction area ratio of aorta images, and π indicates that pi, R indicate at aorta images most constriction
Radius ,/indicating divide operations, SQRT indicates evolution operation, and H indicates that Patient height, W indicate patient's weight.
Using constriction ratio formula, aorta images constriction ratio is calculated, using calculated result as the value of feature 6.
The constriction ratio formula is as follows:
R2=D/SQRT (SQRT ((H × W)/3600)
Wherein, R2 indicates the constriction ratio of aorta images, diameter at most constriction in D expression aorta images.
Step 7, classify to aortic coaractation degree:
The case data that 500 aortic coaractation patients are acquired from hospital database form training set, every case load
According to the CT images comprising patient, height, weight, four kinds of data of pressure difference at aortic coaractation.
Using the method in step 6, six features of patient are quantitatively calculated.
The label label value of the training data is determined according to the pressure difference at patient sustainer constriction, specifically, pressure difference is from 0-
5mmHg, 6-10mmHg, 10-15mmHg ... 95-100mmHg, label label value are respectively 1,2,3 ... 20.
It using the feature of all cases and corresponding label value as training set, is input in classifier, all 500 number of cases
After importing, operation classifier obtains disaggregated model model.
100 cases are chosen as test set, six features of every an example test patient is extracted, is input to classifier
Model model is classified, and obtained result is compared with the actual pressure difference of the case, the accuracy rate of statistical classification, quasi-
True rate then shows that the model has good availability 80% or more.
6 characteristic values are input in main contracting pressure difference model by categorization module.
Main contracting pressure difference model exports the pressure difference at aortic coaractation corresponding with 6 characteristic values, specifically according to master
The output category result of contracting pressure difference model determines specific pressure difference, and output result indicates the patient sustainer constriction pressure difference for 0
For 0-5mmHg, exporting indicates that the patient sustainer constriction pressure difference is 5-10mmHg for 1, and so on, export indicates to suffer from for 20
The aortic coaractation pressure difference of person is 95-100mmHg.
Step 8, result display module shows pressure difference at aortic coaractation that main contracting pressure difference model obtains.
Claims (1)
1. a kind of device for determining aortic coaractation pressure difference by CT images, including data read module, two dimension slicing module,
Divide module, three-dimensional reconstruction module, aortic coaractation discrimination module, characteristic extracting module, categorization module, result display module,
Wherein:
The data read module, for reading in the original thoracic CT data that format is .dcm or .raw；
The two dimension slicing module, for the original thoracic CT data read in be each mapped to different gray-value pixel points
Two-dimensional slice image matrix；
The segmentation module, for sectioning image square of the selection comprising descending aorta end from two-dimensional slice image matrix
Battle array, selection includes the sectioning image matrix of descending aorta end, centered on the central point of selected sectioning image matrix, is generated
N*n centimetres of square-shaped frame, the size of n is between image array length 1/5th between a quarter；It utilizes S [i, j]
=I [p, q] × G [0.25] formula, calculates the gray value at each pixel coordinate of smooth rear image, and removal two-dimensional slice image is made an uproar
Sound extracts the descending aorta tail vein edge in square-shaped frame, obtains vascular wall edge image using medical image software；
The gray value for all internal points that vascular wall edge image is surrounded is set to 255, forms an internal blood vessel area image, benefit
WithFormula calculates the mass center of internal blood vessel area image；Marked angiosomes are loaded into cut
Lowest level sectioning image matrix except picture matrix determines a seed point element, the seed on sectioning image matrix
The abscissa of point element is equal with the mass center abscissa in upper layer sectioning image matrix internal blood vessel region, and ordinate and upper layer are sliced
The mass center ordinate in image array internal blood vessel region is equal, using the edge of vascular wall image as boundary, with seed point element seat
It marks and determines that seed point carries out the growth of 8 neighborhood adaptive regions, obtain the active vascular inner region image on sectioning image matrix；Its
In, the gray value of the pixel after S [i, j] indicates smooth in image at [i, j], I [p, q] is indicated in two-dimensional slice image
The gray value of pixel at [p, q], G [0.25] indicate that standard deviation is 0.25 Gaussian function, and M indicates internal blood vessel area
The center of mass point of area image, ∫ indicate integration operation, and f (x, y) indicates the gray value at internal blood vessel area image pixel (x, y)；
The three-dimensional reconstruction module, for the active vascular inner region image on sectioning image matrix to be imported into three-dimensional reconstruction
Three-dimensional volume drawing, the three-dimensional aorta model after being drawn are carried out in software；
The aortic coaractation discrimination module, for it is straight to measure an aorta at interval of 1mm in three-dimensional aorta model
The adjacent aorta diameter measured twice is done compare respectively by diameter, and ratio result is stored in diameter than in statistical form, if diameter
It is all larger than 0.8 than the value in statistical form, then there is constriction, otherwise, constriction is not present；
The characteristic extracting module calculates software for three-dimensional aorta model to be imported Medical Image Processing software
Value of the greatest gradient dmax value of the aorta vessel arrived as feature 1；When aortic coaractation is located at aorta ascendens position
Mark is assigned a value of 0, and mark when aortic coaractation is located at descending aorta position is assigned a value of -1, and aortic coaractation is located at actively
Mark when arcus haemalis position is assigned a value of 1, by value of the mark as feature 2 after assignment；Three-dimensional aorta model is imported into medicine
Image processing software, diameter is as the value of feature 3 at the most constriction for the aorta vessel that software measurement is obtained；By three-dimensional active
Arteries and veins model imports Medical Image Processing software, diameter and descending aorta diameter at the most constriction that software is measured and is calculated
Value of the ratio as feature 4；Utilize R1=π R2/ (SQRT ((H × W)/3600) formula calculates aortic coaractation area ratio, will
Value of the calculated result as feature 5；Using R2=D/SQRT, (SQRT ((H × W)/3600) formula calculates aortic coaractation ratio
Rate, using calculated result as the value of feature 6；Wherein, R1 indicates the constriction area ratio of aorta images, and π indicates pi, R table
Show radius at aorta images most constriction ,/indicating divide operations, SQRT indicates evolution operation, and H indicates that Patient height, W indicate to suffer from
Person's weight, R2 indicate the constriction ratio of aorta images, diameter at most constriction in D expression aorta images；
The categorization module, for 6 characteristic values to be input in main contracting pressure difference model, main contracting pressure difference model is exported and 6
Pressure difference at the corresponding aortic coaractation of characteristic value；
The result display module, for showing pressure difference at aortic coaractation that main contracting pressure difference model obtains.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710021582.5A CN106803251B (en) | 2017-01-12 | 2017-01-12 | The apparatus and method of aortic coaractation pressure difference are determined by CT images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710021582.5A CN106803251B (en) | 2017-01-12 | 2017-01-12 | The apparatus and method of aortic coaractation pressure difference are determined by CT images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106803251A CN106803251A (en) | 2017-06-06 |
| CN106803251B true CN106803251B (en) | 2019-10-08 |
Family
ID=58985649
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710021582.5A Active CN106803251B (en) | 2017-01-12 | 2017-01-12 | The apparatus and method of aortic coaractation pressure difference are determined by CT images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106803251B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109360214B (en) * | 2018-10-05 | 2021-07-13 | 数坤(北京)网络科技股份有限公司 | Multi-channel network prediction optimization method for coronary artery segmentation |
| CN109886973B (en) * | 2019-01-25 | 2021-01-08 | 杭州晟视科技有限公司 | Blood vessel extraction method and device and computer readable storage medium |
| CN109978888B (en) * | 2019-02-18 | 2023-07-28 | 平安科技（深圳）有限公司 | Image segmentation method, device and computer readable storage medium |
| CN110021053B (en) * | 2019-04-16 | 2020-05-15 | 河北医科大学第二医院 | Image positioning method and device based on coordinate transformation, storage medium and equipment |
| CN115769252A (en) * | 2020-06-29 | 2023-03-07 | 苏州润迈德医疗科技有限公司 | Method and storage medium for obtaining aorta based on deep learning |
| EP3975115A1 (en) * | 2020-09-25 | 2022-03-30 | Siemens Healthcare GmbH | Method and system for analyzing a plurality of interconnected blood vessels |
| CN112330708B (en) * | 2020-11-24 | 2024-04-23 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, device, storage medium and electronic equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913432A (en) * | 2016-04-12 | 2016-08-31 | 妙智科技（深圳）有限公司 | Aorta extracting method and aorta extracting device based on CT sequence image |
-
2017
- 2017-01-12 CN CN201710021582.5A patent/CN106803251B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913432A (en) * | 2016-04-12 | 2016-08-31 | 妙智科技（深圳）有限公司 | Aorta extracting method and aorta extracting device based on CT sequence image |
Non-Patent Citations (4)
| Title |
|---|
| Ambulatory Blood Pressure Monitoring and Left Ventricular Mass and Function After Successful Surgical Repair of Coarctation of the Aorta;Janine Leandro MD et al;《Jouranl of the American College of Cardiology》;19920731;第20卷(第1期);197-204页 * |
| Surgical and Endovascular Repair of Aortic Coarctation:Normal Findings and Appearance of Complications on CT Angiography and MR Angiography;Ming-Chen Paul Shih et al;《American Journal of Roentgenology》;20060930;第187卷(第3期);302-312页 * |
| Virtual interventions for image-based blood flow computation;Guanglei Xiong et al;《Computer-Aided Design》;20120131;第44卷(第1期);3-14页 * |
| 超声心动图与CT血管成像诊断主动脉缩窄的价值对比研究;任兴香等;《中华医学超声杂志（电子版）》;20140930;第11卷(第9期);732-736页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106803251A (en) | 2017-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106803251B (en) | The apparatus and method of aortic coaractation pressure difference are determined by CT images | |
| US11896416B2 (en) | Method for calculating coronary artery fractional flow reserve on basis of myocardial blood flow and CT images | |
| EP3723038B1 (en) | Fast calculation method and system employing plaque stability index of medical image sequence | |
| CN106023202B (en) | CFR fraction computational methods based on cardiac CT image | |
| CN105184086B (en) | For the method and system that improved Hemodynamics calculates in coronary artery | |
| US11901081B2 (en) | Method for calculating index of microcirculatory resistance based on myocardial blood flow and CT image | |
| US7567696B2 (en) | System and method for detecting the aortic valve using a model-based segmentation technique | |
| CN110120033A (en) | Based on improved U-Net neural network three-dimensional brain tumor image partition method | |
| CN106659400A (en) | Apparatus for determining a fractional flow reserve value | |
| Shekhar et al. | Three-dimensional segmentation of luminal and adventitial borders in serial intravascular ultrasound images | |
| WO2005023086B1 (en) | Systems, methods, and computer program products for analysis of vessel attributes for diagnosis, disease staging, and surgical planning | |
| CN110120031B (en) | Method and device for obtaining blood flow reserve fraction of blood vessel | |
| Olender et al. | A domain enriched deep learning approach to classify atherosclerosis using intravascular ultrasound imaging | |
| CN102171725B (en) | Brain ventricle analysis | |
| KR101530352B1 (en) | Cfd modeling and analysis method based on material properties | |
| CN113470060A (en) | Coronary artery multi-angle curved surface reconstruction visualization method based on CT image | |
| Babin et al. | Robust segmentation methods with an application to aortic pulse wave velocity calculation | |
| WO2014163334A1 (en) | Method for modeling and analyzing computational fluid dynamics on basis of material properties | |
| Niu et al. | Improving automated tissue characterization in optical coherence tomography by melding attenuation compensation with deep learning | |
| CN115546123A (en) | Prediction system for rupture risk of abdominal aortic aneurysm based on machine learning | |
| CN111292285A (en) | Automatic screening method for diabetes mellitus based on naive Bayes and support vector machine | |
| EP4174762A1 (en) | Deep learning-based aortic image acquisition system | |
| US10332255B2 (en) | Method for assessing stenosis severity in a lesion tree through stenosis mapping | |
| CN115115735A (en) | Rapid calculation system and method for endothelium dynamic strain based on multi-phase coronary CT radiography | |
| EP3995081A1 (en) | Diagnosis assisting program |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20171218 Address after: Xi'an City, Shaanxi province Taibai Road 710071 No. 2 Applicant after: Xidian University Applicant after: Guangdong Prov. Cardiovascular disease Inst. Address before: Taibai Road Shaanxi Yanta District 710071 city of Xi'an province No. 2 Applicant before: Xidian University |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |