CN117198510A - Vascular simulation intervention evaluation system based on imaging - Google Patents
Vascular simulation intervention evaluation system based on imaging Download PDFInfo
- Publication number
- CN117198510A CN117198510A CN202311363032.3A CN202311363032A CN117198510A CN 117198510 A CN117198510 A CN 117198510A CN 202311363032 A CN202311363032 A CN 202311363032A CN 117198510 A CN117198510 A CN 117198510A
- Authority
- CN
- China
- Prior art keywords
- blood vessel
- vascular
- sequence
- neural network
- blood
- 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.)
- Granted
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 40
- 238000004088 simulation Methods 0.000 title claims abstract description 38
- 238000003384 imaging method Methods 0.000 title claims abstract description 18
- 230000002792 vascular Effects 0.000 title claims description 83
- 210000004204 blood vessel Anatomy 0.000 claims abstract description 170
- 238000013528 artificial neural network Methods 0.000 claims abstract description 40
- 230000003840 blood vessel health Effects 0.000 claims abstract description 34
- 230000000306 recurrent effect Effects 0.000 claims abstract description 28
- 238000013136 deep learning model Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000013507 mapping Methods 0.000 claims abstract description 4
- 230000006438 vascular health Effects 0.000 claims abstract description 4
- 230000017531 blood circulation Effects 0.000 claims description 49
- 239000013598 vector Substances 0.000 claims description 25
- 238000005516 engineering process Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 20
- 238000013527 convolutional neural network Methods 0.000 claims description 14
- 238000005259 measurement Methods 0.000 claims description 13
- 238000000034 method Methods 0.000 claims description 12
- 238000010008 shearing Methods 0.000 claims description 12
- 210000004369 blood Anatomy 0.000 claims description 11
- 239000008280 blood Substances 0.000 claims description 11
- 239000012530 fluid Substances 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000002059 diagnostic imaging Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 230000003993 interaction Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 230000015654 memory Effects 0.000 claims description 2
- 230000004044 response Effects 0.000 claims description 2
- 230000036541 health Effects 0.000 abstract description 8
- 230000008859 change Effects 0.000 abstract description 3
- 238000000605 extraction Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 4
- 230000000877 morphologic effect Effects 0.000 description 4
- 208000019553 vascular disease Diseases 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 208000027418 Wounds and injury Diseases 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000000004 hemodynamic effect Effects 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 238000002583 angiography Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000010297 mechanical methods and process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000012798 spherical particle Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The application discloses a blood vessel simulation intervention evaluation system based on imaging, which comprises an image acquisition unit, a reconstruction unit, a hydrodynamic simulation unit, a training unit, a prediction unit, a mapping unit and an evaluation unit, and relates to the technical field of blood vessel simulation; the technical key points are as follows: predicting the sequence of the blood vessel condition parameters on the blood vessel path through the deep learning model, realizing the efficient prediction of the blood vessel condition parameters by utilizing the mutual coordination among the data stream neural network, the recurrent neural network and the conditional random field model, and considering the context information and the time sequence relationship of the blood vessel path so as to improve the accuracy of the prediction, comprehensively considering the change of each parameter of the relevant blood vessel when evaluating the blood vessel health state, and ensuring the generation of the blood vessel health state evaluation value Bphz i In evaluating the vascular health state of the blood vessel by Bphz i After the comparison with the preset threshold value, the health condition of the corresponding section of blood vessel can be accurately and efficiently obtained.
Description
Technical Field
The application relates to the technical field of vascular simulation, in particular to a vascular simulation intervention evaluation system based on imaging.
Background
With the increasing incidence of cardiovascular diseases year by year, relevant parameters of blood vessels and blood are important reference indexes for disease diagnosis, and when the parameters of the blood vessels and the blood are acquired, common technical means mainly include invasive quantitative measurement by using instruments, non-invasive measurement, reconstruction of a vascular geometric model of a patient by using a medical image sequence of the blood vessels, simulation of blood flow by using a computational fluid mechanical method with proper physiological boundary conditions and parameters in calculation of the model, non-invasive measurement by using machine learning, and prediction of blood vessel condition parameters at a certain point or each point on the blood vessels by mainly using a learning network in isolation.
Taking FFR prediction as an example, a traditional FFR prediction system based on machine learning and a learning network generally comprises a plurality of modules, including a feature extraction model, an FFR prediction model and an FFR smoothing post-processing model, wherein the feature extraction model generally adopts certain fixed feature extraction algorithms, the FFR prediction model and the FFR smoothing post-processing model need to be trained independently, and each model works independently;
the problems in the existing deployment schemes are: the invasive measurement requires wound access to cause a certain injury to a human body, the existing non-invasive measurement adopts computational simulation of a computational dynamics method, a large amount of computational burden is generated, the computational time is long, the virtual non-invasive measurement is difficult to meet the requirements of clinical environments, the application of the virtual non-invasive measurement in the clinical environments is seriously hindered, the existing non-invasive measurement using machine learning is performed independently, each parameter is different, the parameters cannot complement each other during training, the objective function of single module training is deviated from the overall performance of the system, and the trained network often cannot reach the optimal performance of the overall system.
Disclosure of Invention
(one) solving the technical problems
Aiming at the defects of the prior art, the application provides a vascular simulation intervention evaluation system based on imaging, which solves a plurality of key problems in the rapid calculation of a vascular condition parameter sequence through an end-to-end training model by utilizing the advantages of a recurrent neural network in sequence learning and a convolutional neural network in image learning, comprises improving the calculation speed and reducing the manual intervention of feature extraction, and realizes a complete solution for the rapid and accurate calculation of the vascular condition parameter sequence on a vascular path.
(II) technical scheme
In order to achieve the above purpose, the application is realized by the following technical scheme:
an imaging-based vascular simulation intervention assessment system, comprising:
an image acquisition unit that acquires a medical image sequence of a blood vessel tree;
a reconstruction unit for reconstructing a geometric model of the vessel tree based on the medical image sequence, extracting a vessel path and a central line of a vessel thereon from the geometric model, and intercepting the image block sequence along the central line of the vessel on the vessel path;
the hydrodynamic simulation unit is used for performing computational hydrodynamic model simulation based on a geometric model of the blood vessel tree so as to acquire a corresponding blood vessel condition parameter sequence on the blood vessel path;
the training unit takes the corresponding vascular condition parameter sequence on the vascular path as training data, and acquires the corresponding vascular condition parameter sequence from the measurement assembly as training data to train the constructed deep learning model;
the prediction unit predicts the sequence of the vascular condition parameters on the vascular path by using a trained deep learning model based on the intercepted image block sequence, and the deep learning model is formed by sequentially connecting a data stream neural network, a recurrent neural network and a conditional random field model in series;
a mapping unit that maps the sequence of the predicted vascular condition parameters on the predicted vascular paths back into a vascular tree including the vascular paths, and obtains a vascular condition parameter sequence segment of the overlapping portion in the vascular tree based on the vascular condition parameter sequence segments of the predicted vascular condition parameter sequences on the respective vascular paths overlapping in the vascular tree at the overlapping portion;
the evaluation unit extracts the blood vessel condition value of each blood vessel section on the same time node T based on the blood vessel condition parameter sequence section of the overlapped part, builds a data analysis model by taking the difference value between the blood vessel condition value and the corresponding initial value as a parameter, and generates a blood vessel health condition evaluation value Bphz i And the blood vessel health state evaluation value Bphz i And comparing the blood vessel health status with a preset threshold value, and predicting the blood vessel health status according to a comparison result.
Further, the step of reconstructing the geometrical model of the vessel tree in the reconstruction unit is as follows:
s101, acquiring geometric shape information of a blood vessel from a medical image sequence, wherein the geometric shape information at least comprises a blood vessel center line and a path, and the geometric shape information of the blood vessel is acquired by using an image processing technology;
s102, constructing a geometric model of a blood vessel tree according to geometric shape information by utilizing a 3D reconstruction algorithm technology;
s103, in the geometric model of the blood vessel tree, the boundary of the blood vessel is extracted from the medical image sequence through an image processing technology, and the blood vessel is represented by using a tree structure according to the branching and connecting relation of the blood vessel, so that the blood vessel tree is further established, and the geometric model is required to be smoothed after the boundary of the blood vessel is extracted from the medical image sequence.
Further, the sequence of image blocks in the reconstruction unit comprises a number of sequences of 2D image blocks and a sequence of 3D image blocks.
Further, the vascular condition parameter sequence in the hydrodynamic simulation unit comprises a vascular radius, a blood flow velocity and a blood flow shear stress;
by reconstructing a geometric model of a blood vessel tree and performing computational fluid dynamics simulation, the blood vessel radius value of each blood vessel segment at different times can be obtained, the blood flow velocity of each blood vessel segment at different times can be obtained, and the blood flow shearing stress of each blood vessel segment at different times can be obtained, so that a blood vessel condition parameter sequence can be obtained.
Further, the measurement assembly used in the training unit includes a medical imaging device and a physiological signal device.
Further, in the prediction unit, the data flow neural network adopts a 2D convolutional neural network and a 3D convolutional neural network to respectively learn a plurality of 2D image block sequences and 2D image blocks and 3D image blocks in the 3D image block sequences on the vascular path, each convolutional neural network respectively returns a vector, and the vectors returned by each convolutional neural network are connected into a vector and then transmitted to the recurrent neural network;
the recurrent neural network adopts a two-way long-short-term memory recurrent neural network, which is provided with LSTM layers in two directions, one is used for processing forward sequence data, the other is used for processing reverse sequence data, after receiving vectors from the data stream neural network, the recurrent neural network processes the vectors so as to consider the context information of a blood vessel path, the recurrent neural network captures the time sequence relation of the blood vessel path by correlating the image block sequences adjacent to each other, an output vector is returned, and the vector is used as the input of a conditional random field model;
after the conditional random field model receives the output vector of the recurrent neural network, the conditional random field model is used in combination with the recurrent neural network to take into account prior knowledge and constraints in the observation sequence, in the conditional random field model, the context information of the observation sequence is used to infer the required tag sequence, which is achieved by combining the output vector of the recurrent neural network with the features of the observation sequence, and the conditional random field model takes into account the interactions between the features and the continuity of the observation sequence to generate the required tag sequence.
Further, the blood vessel condition value in the evaluation unit includes a blood vessel radius value Gb i Vr of blood flow velocity i No. of blood flow shear stress i And the blood vessel condition value and the blood vessel health state evaluation value Bphz i I in (a) represents the sequential order numbers of the blood vessel sections from top to bottom, i=1, 2, …, n, wherein n is a positive integer.
Further, a blood vessel health state evaluation value Bphz is generated i The steps of (a) are as follows:
s201, parameter acquisition: the parameters include a blood vessel radius value Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i A difference from an initial value of blood flow shear stress No;
s202, parameter pretreatment: carrying out dimensionless treatment on the calculated parameters;
s203, generating a blood vessel health state evaluation value Bphz according to the preprocessed parameters i The formula is as follows:
wherein alpha, beta and gamma are respectively vascular radius values Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i The preset proportionality coefficient of the difference value between the initial value No of the blood flow shearing stress and the initial value No of the blood flow shearing stress, and alpha > beta > gamma > 0,G are constant correction coefficients.
Further, the blood vessel health state evaluation value Bphz i The comparison with the preset threshold value results in the following:
if it is the blood vessel health state evaluation value Bphz i If the threshold value is exceeded, no response is made;
if it is the blood vessel health state evaluation value Bphz i If the threshold value is not exceeded, an early warning is sent out, and stroboscopic red light is sent out at the position corresponding to the blood vessel segment on the geometric model.
(III) beneficial effects
The application provides a blood vessel simulation intervention evaluation system based on imaging, which has the following beneficial effects:
1. the method comprises the steps of obtaining an image block sequence on a blood vessel path, predicting a sequence of blood vessel condition parameters on the blood vessel path based on the obtained image block sequence on the blood vessel path by using a trained deep learning model, realizing efficient prediction of the blood vessel condition parameters by using the interaction among a data flow neural network, a recurrent neural network and a conditional random field model in the deep learning model, and considering the context information and time sequence relation of the blood vessel path, thereby improving the prediction accuracy, solving the problems of slow speed and low precision in the calculation of the blood vessel condition parameter sequence and the requirement of excessive manual intervention in the feature extraction in the prior art, and ensuring the optimal performance of the whole system;
2. when the blood vessel health state is evaluated, the change of each parameter of the related blood vessel is comprehensively considered, and the generation of the blood vessel health state evaluation value Bphz is ensured i Not only can the specific segment blood vessel be subjected to targeted health assessment, but also all blood vessel segments in the human body can be subjected to overall assessment, and the blood vessel health state assessment value Bphz is obtained i After the comparison with the preset threshold value, the health condition of the corresponding section of blood vessel can be accurately and efficiently obtained, the position of the abnormal blood vessel section can be effectively distinguished through visual display on the geometric model, the follow-up medical staff can conveniently conduct targeted examination, and the practicability of the overall evaluation system design is embodied.
Drawings
FIG. 1 is a schematic diagram of a system for image-based vascular simulation intervention assessment in accordance with the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Referring to fig. 1, the present application provides an imaging-based vascular simulation intervention evaluation system, which includes:
an image acquisition unit that acquires a medical image sequence of a blood vessel tree;
the technologies adopted in the medical image sequence for acquiring the blood vessel tree by using the imaging method comprise an MRA technology, a CTA technology, a DSA technology and an ultrasonic technology, and the corresponding technologies are selected according to actual requirements;
it should be noted that: MRA is a noninvasive medical imaging examination, can clearly display the morphological structure of human blood vessels, provides an important basis for diagnosing vascular diseases, CTA is a computer tomography technique, can clearly display the morphological structure of human blood vessels, provides an important basis for diagnosing vascular diseases, DSA is a digital subtraction angiography technique, can clearly display the morphological structure of human blood vessels, provides an important basis for diagnosing vascular diseases, and ultrasound is a noninvasive medical imaging examination, can clearly display the morphological structure of human blood vessels, and provides an important basis for diagnosing vascular diseases;
the medical image sequence acquired by the image acquisition unit comprises a plurality of 2D image block sequences and 3D image block sequences, which are used for reflecting the morphology and structure of the vessel tree, and the image block sequences in the reconstruction unit are used for subsequent analysis and processing.
A reconstruction unit reconstructing a geometric model of the vessel tree based on the medical image sequence acquired from the image acquisition unit, extracting a vessel path and a center line of a vessel thereon from the geometric model, and intercepting an image block sequence along the extracted center line of the vessel on the vessel path, the image block sequence including a plurality of 2D image block sequences and a 3D image block sequence;
the steps of reconstructing the geometric model of the vessel tree are as follows:
s101, acquiring geometric shape information of a blood vessel from a medical image sequence, wherein the geometric shape information at least comprises a blood vessel central line and a path, and the geometric shape information also comprises the size, the shape and the position of the blood vessel; acquisition of the geometric information of the blood vessel is achieved by using computer vision and image processing techniques, such as edge detection, region filling and surface reconstruction;
s102, constructing a geometric model of a blood vessel tree according to geometric shape information by utilizing a 3D reconstruction algorithm technology;
the 3D reconstruction algorithm technology comprises a spherical particle method and a contour tracing method, and a geometric modeling technology of parameterization and spline interpolation can be adopted as required when a geometric model is constructed;
s103, in a geometric model of the vessel tree, extracting the boundary of the vessel from the medical image sequence by an image processing technology, and representing by using a tree structure according to the branching and connection relation of the vessel so as to finish further establishment of the vessel tree, for example, using nodes and edges to represent branching and connection points of the vessel; the adopted image processing technology comprises edge detection and region growing, and detail processing, such as smoothing and grid optimization, is needed to be carried out on the geometric model after the boundary of the blood vessel is extracted from the medical image sequence so as to improve the accuracy and quality of the reconstructed geometric model;
specifically, the step of acquiring a plurality of 2D image block sequences and 3D image block sequences on a blood vessel path includes: acquiring a plurality of 2D image block sequences and 3D image block sequences of a vessel tree, reconstructing a geometric model of the vessel tree based on the acquired plurality of 2D image block sequences and 3D image block sequences of the vessel tree, extracting a vessel path and a central line of a vessel on the vessel path from the geometric model of the vessel tree, and intercepting the plurality of 2D image block sequences and 3D image block sequences along the central line of the vessel on the extracted vessel path.
The hydrodynamic simulation unit is used for performing computational hydrodynamic model simulation based on the geometric model of the blood vessel tree reconstructed by the reconstruction unit so as to obtain a corresponding blood vessel condition parameter sequence on the blood vessel path, wherein the blood vessel condition parameter sequence is used for reflecting the hemodynamic characteristics and states;
wherein the vascular condition parameter sequence comprises a vascular radius, a blood flow velocity and a blood flow shear stress;
radius of blood vessel: the blood vessel radius can reflect the thickness degree of the blood vessel, is one of important parameters for evaluating the health condition of the blood vessel, and can acquire the blood vessel radius value of each blood vessel section at different times by reconstructing a geometric model of a blood vessel tree and performing computational fluid dynamics simulation so as to obtain a blood vessel condition parameter sequence;
blood flow velocity: the blood flow velocity can reflect the flow condition of blood in blood vessels, is one of important parameters for evaluating the health condition of the blood vessels, and can acquire the blood flow velocity of each blood vessel segment at different times by reconstructing a geometric model of a blood vessel tree and performing computational fluid dynamics simulation so as to obtain a blood vessel condition parameter sequence.
Shear stress of blood flow: the blood flow shear stress can reflect the interaction between blood and the blood vessel wall, is one of important parameters for evaluating the health condition of the blood vessel, and can be obtained at different times by reconstructing a geometric model of a blood vessel tree and performing computational fluid dynamics simulation, so as to obtain a blood vessel condition parameter sequence;
specifically, the step of obtaining the corresponding vascular condition parameter sequence includes: acquiring a plurality of 2D image block sequences and 3D image block sequences of a blood vessel tree, reconstructing a geometric model of the blood vessel tree based on the acquired plurality of 2D image block sequences and 3D image block sequences of the blood vessel tree, and performing computational fluid dynamics simulation on the reconstructed geometric model of the blood vessel tree to obtain a corresponding blood vessel condition parameter sequence.
The training unit takes the corresponding vascular condition parameter sequence on the vascular path acquired from the hydrodynamic simulation unit as training data for training the constructed deep learning model, and acquires the corresponding vascular condition parameter sequence from the measurement assembly as training data for verifying and optimizing the performance of the deep learning model;
the measuring component comprises medical imaging equipment such as CT, MRI, ultrasonic wave and physiological signal equipment such as electrocardiogram, sphygmomanometer, pulse oximeter and ultrasonic blood flow measuring instrument;
medical imaging equipment: acquiring blood vessel image data of a patient through medical image equipment, and then processing and analyzing the data by utilizing an image processing technology, extracting parameters such as the geometric shape, the structure, the hemodynamics and the like of blood vessels to form a blood vessel condition parameter sequence, wherein the image processing technology at least comprises image enhancement and denoising, image segmentation, feature extraction and three-dimensional reconstruction;
physiological signal device: the physiological signal data of a patient are acquired through physiological signal equipment, then the data are processed and analyzed by a signal processing technology, parameters such as blood flow velocity, blood flow shearing stress and the like of a blood vessel are extracted, and a blood vessel condition parameter sequence is formed, wherein the signal processing technology at least comprises signal filtering and denoising, signal characteristic extraction, time domain and frequency domain analysis and nonlinear analysis.
The prediction unit predicts the sequence of the vascular condition parameters on the vascular path by using a trained deep learning model based on a plurality of 2D image block sequences and 3D image block sequences on the vascular path intercepted by the reconstruction unit, and the deep learning model is formed by sequentially connecting a data flow neural network, a recurrent neural network and a conditional random field model in series;
the data flow neural network adopts a 2D Convolutional Neural Network (CNN) and a 3D convolutional neural network to respectively learn a plurality of 2D image block sequences and 2D image blocks and 3D image blocks in the 3D image block sequences on a blood vessel path, each convolutional neural network respectively returns a vector, and the vectors returned by each convolutional neural network are connected into a vector and then transmitted to the recurrent neural network;
recurrent Neural Networks (RNNs) include two-way long and short term memory recurrent neural networks (BiLSTM), which is a special recurrent neural network with two-way LSTM layers, one for processing forward sequence data and the other for processing reverse sequence data, after receiving vectors from the data stream neural network, the BiLSTM network will process them to take into account the vascular path context information, by correlating the sequence of image blocks that are adjacent one after the other, the BiLSTM network will be able to capture the vascular path timing relationship, the BiLSTM network will return an output vector that will serve as an input to the conditional random field model;
a conditional random field model (CRF) which is a statistical learning model used in conjunction with a recurrent neural network to take into account prior knowledge and constraints in the observation sequence, after receiving the output vector of the BiLSTM network, in which the most probable tag sequence will be inferred using the context information of the observation sequence, by combining the output vector of the BiLSTM network with the features of the observation sequence, the CRF model will take into account the interactions between the features and the continuity of the observation sequence to generate the most accurate tag sequence;
it should be noted that: the observation sequence is known data input into the deep learning model and is used for training the deep learning model and guiding prediction, and the sequence of the blood vessel condition parameters is output obtained after the observation sequence is processed by the deep learning model, namely the blood vessel condition parameter sequence predicted by the model according to the observation sequence; in the deep learning model, the vascular path tag sequence can be used as an output variable of the model to guide the prediction and learning of the model.
A mapping unit that maps the sequence of the predicted vascular condition parameters on the predicted vascular paths back into a vascular tree including the vascular paths, and obtains a vascular condition parameter sequence segment of the overlapping portion in the vascular tree based on the vascular condition parameter sequence segments of the predicted vascular condition parameter sequences on the respective vascular paths overlapping in the vascular tree at the overlapping portion;
by adopting the technical scheme: the method comprises the steps of acquiring an image block sequence on a blood vessel path, predicting a sequence of blood vessel condition parameters on the blood vessel path by using a trained deep learning model based on the acquired image block sequence on the blood vessel path, realizing efficient prediction of the blood vessel condition parameters by using the data flow neural network, the recurrent neural network and the conditional random field model in the deep learning model, and considering the context information and the time sequence relationship of the blood vessel path, thereby improving the prediction accuracy, solving the problems of slow speed, low precision and excessive manual intervention in the process of feature extraction in the traditional calculation of the blood vessel condition parameter sequence, and ensuring the optimal performance of the whole system.
An evaluation unit for extracting a blood vessel condition value of each blood vessel segment at the same time node T based on the blood vessel condition parameter sequence segments of the overlapping portions, including a blood vessel radius value Gb i Vr of blood flow velocity i No. of blood flow shear stress i Taking the difference value between the blood vessel state value and the corresponding initial value as a parameter, constructing a data analysis model, and generating a blood vessel health state evaluation value Bphz i And the blood vessel health state evaluation value Bphz i Comparing the blood vessel health status with a preset threshold value, and predicting the health status of the blood vessel according to a comparison result;
wherein each vessel segment represents a vessel segment with bifurcation points at both ends and consistent overall vessel radius values, the same time node T can be selected according to practical needs, for example, the time node T, bphz after 24h from the initial value i I represents the sequential ordering numbers of the blood vessel sections from top to bottom, i.e. the body is from head to foot, i=1, 2, …, n, where n is a positive integer;
generating a vascular health status assessment value Bphz i The steps of (a) are as follows:
s201, parameter acquisition: the parameters include a blood vessel radius value Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i A difference from an initial value of blood flow shear stress No; it should be noted that: the initial value Gb of the blood vessel radius, the initial flow velocity Vr of the blood and the initial value No of the blood flow shearing stress are the blood vessel condition parameter sequences obtained from the hydrodynamic simulation unit;
specifically, the blood vessel radius value and the initial value thereof, the blood flow velocity and the initial flow velocity, and the blood flow shearing stress and the initial value thereof can be directly obtained from the measuring assembly.
S202, parameter pretreatment: performing dimensionless treatment on the calculated parameters to remove units;
s203, generating a blood vessel health state evaluation value Bphz according to the preprocessed parameters i The formula is as follows:
wherein alpha, beta and gamma are respectively vascular radius values Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i The preset proportionality coefficient of the difference value between the blood flow shearing stress initial value No and the blood flow shearing stress initial value No is that alpha is larger than beta is larger than gamma is larger than 0, alpha+beta+gamma=1.534, G is a constant correction coefficient, the specific value of the constant correction coefficient can be adjusted and set by a user or generated by fitting an analysis function, and the value of G is 1.320;
it should be noted that: a person skilled in the art collects a plurality of groups of sample data and sets a corresponding preset scaling factor for each group of sample data; substituting the preset proportionality coefficient, which can be the preset proportionality coefficient and the acquired sample data, into a formula, forming a ternary once equation set by any three formulas, screening the calculated coefficient, taking an average value, and obtaining a value; the magnitude of the coefficient is a specific numerical value obtained by quantizing each parameter, so that the subsequent comparison is convenient, the magnitude of the coefficient depends on the number of sample data and the corresponding preset proportional coefficient preliminarily set by a person skilled in the art for each group of sample data, that is, the coefficient is preset according to the actual practice, so long as the proportional relation between the parameter and the quantized numerical value is not influenced, and the above description is also adopted for the preset proportional coefficient and the constant correction coefficient described in other formulas.
Blood vessel health state evaluation value Bphz i The comparison with the preset threshold value results in the following:
if it is the blood vessel health state evaluation value Bphz i Exceeding a threshold value, indicating vascular health of the corresponding segment, the system not responding;
if it is the blood vessel health state evaluation value Bphz i If the threshold value is not exceeded, the system gives out early warning, and gives out stroboscopic red light at the position corresponding to the blood vessel segment on the geometric model so as to prompt the staff to check the position, wherein the stroboscopic red light is onlyThe early warning mode can be realized by any other mode, and the early warning effect can be realized only by ensuring.
By adopting the technical scheme: when the blood vessel health state is evaluated, the change of each parameter of the related blood vessel is comprehensively considered, and the generation of the blood vessel health state evaluation value Bphz is ensured i Not only can the specific segment blood vessel be subjected to targeted health assessment, but also all blood vessel segments in the human body can be subjected to overall assessment, and the blood vessel health state assessment value Bphz is obtained i After the comparison with the preset threshold value, the health condition of the corresponding section of blood vessel can be accurately and efficiently obtained, the position of the abnormal blood vessel section can be effectively distinguished through visual display on the geometric model, the follow-up medical staff can conveniently conduct targeted examination, and the practicability of the overall evaluation system design is embodied.
In the application, the related formulas are all the numerical calculation after dimensionality removal, the formula is a formula for acquiring a large amount of data to perform software simulation to obtain the latest real situation, the formula is set by a person skilled in the art according to the actual situation, and for a preset threshold value, the threshold value is set according to the acquisition of historical data, and corresponding adjustment can be performed according to the actual needs.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application.
Claims (9)
1. An imaging-based vascular simulation intervention assessment system, comprising:
an image acquisition unit that acquires a medical image sequence of a blood vessel tree;
the reconstruction unit is used for reconstructing a geometric model of the blood vessel tree based on the medical image sequence, extracting a blood vessel path and the central line of the blood vessel on the blood vessel path from the geometric model, and intercepting an image block sequence along the central line of the blood vessel on the blood vessel path, and is characterized in that:
the hydrodynamic simulation unit is used for performing computational hydrodynamic model simulation based on a geometric model of the blood vessel tree so as to acquire a corresponding blood vessel condition parameter sequence on the blood vessel path;
the training unit takes the corresponding vascular condition parameter sequence on the vascular path as training data, and acquires the corresponding vascular condition parameter sequence from the measurement assembly as training data to train the constructed deep learning model;
the prediction unit predicts the sequence of the vascular condition parameters on the vascular path by using a trained deep learning model based on the intercepted image block sequence, and the deep learning model is formed by sequentially connecting a data stream neural network, a recurrent neural network and a conditional random field model in series;
a mapping unit that maps the sequence of the predicted vascular condition parameters on the predicted vascular paths back into a vascular tree including the vascular paths, and obtains a vascular condition parameter sequence segment of the overlapping portion in the vascular tree based on the vascular condition parameter sequence segments of the predicted vascular condition parameter sequences on the respective vascular paths overlapping in the vascular tree at the overlapping portion;
the evaluation unit extracts the blood vessel condition value of each blood vessel section on the same time node T based on the blood vessel condition parameter sequence section of the overlapped part, builds a data analysis model by taking the difference value between the blood vessel condition value and the corresponding initial value as a parameter, and generates a blood vessel health condition evaluation value Bphz i And the blood vessel health state evaluation value Bphz i And comparing the blood vessel health status with a preset threshold value, and predicting the blood vessel health status according to a comparison result.
2. The imaging-based vascular simulation intervention assessment system of claim 1, wherein: the step of reconstructing the geometric model of the vessel tree in the reconstruction unit is as follows:
s101, acquiring geometric shape information of a blood vessel from a medical image sequence, wherein the geometric shape information at least comprises a blood vessel center line and a path, and the geometric shape information of the blood vessel is acquired by using an image processing technology;
s102, constructing a geometric model of a blood vessel tree according to geometric shape information by utilizing a 3D reconstruction algorithm technology;
s103, in the geometric model of the blood vessel tree, the boundary of the blood vessel is extracted from the medical image sequence through an image processing technology, and the blood vessel is represented by using a tree structure according to the branching and connecting relation of the blood vessel, so that the blood vessel tree is further established, and the geometric model is required to be smoothed after the boundary of the blood vessel is extracted from the medical image sequence.
3. The imaging-based vascular simulation intervention assessment system of claim 2, wherein: the sequence of image blocks in the reconstruction unit comprises several sequences of 2D image blocks and a sequence of 3D image blocks.
4. The imaging-based vascular simulation intervention assessment system of claim 3, wherein: the vascular condition parameter sequence in the hydrodynamic simulation unit comprises a vascular radius, a blood flow speed and a blood flow shearing stress;
by reconstructing a geometric model of a blood vessel tree and performing computational fluid dynamics simulation, the blood vessel radius value of each blood vessel segment at different times can be obtained, the blood flow velocity of each blood vessel segment at different times can be obtained, and the blood flow shearing stress of each blood vessel segment at different times can be obtained, so that a blood vessel condition parameter sequence can be obtained.
5. The imaging-based vascular simulation intervention assessment system of claim 1, wherein: the measurement components used in the training unit include a medical imaging device and a physiological signal device.
6. The imaging-based vascular simulation intervention assessment system of claim 4, wherein: in the prediction unit, the data flow neural network adopts a 2D convolutional neural network and a 3D convolutional neural network to respectively learn a plurality of 2D image block sequences on a blood vessel path and 2D image blocks and 3D image blocks in the 3D image block sequences, each convolutional neural network respectively returns a vector, and the vectors returned by each convolutional neural network are connected into a vector and then transmitted to the recurrent neural network;
the recurrent neural network adopts a two-way long-short-term memory recurrent neural network, which is provided with LSTM layers in two directions, one is used for processing forward sequence data, the other is used for processing reverse sequence data, after receiving vectors from the data stream neural network, the recurrent neural network processes the vectors so as to consider the context information of a blood vessel path, the recurrent neural network captures the time sequence relation of the blood vessel path by correlating the image block sequences adjacent to each other, an output vector is returned, and the vector is used as the input of a conditional random field model;
after the conditional random field model receives the output vector of the recurrent neural network, the conditional random field model is used in combination with the recurrent neural network to take into account prior knowledge and constraints in the observation sequence, in the conditional random field model, the context information of the observation sequence is used to infer the required tag sequence, which is achieved by combining the output vector of the recurrent neural network with the features of the observation sequence, and the conditional random field model takes into account the interactions between the features and the continuity of the observation sequence to generate the required tag sequence.
7. The imaging-based vascular simulation intervention assessment system of claim 6, wherein: the vascular condition values in the evaluation unit comprise vascular radius values Gb i Vr of blood flow velocity i No. of blood flow shear stress i And the blood vessel condition value and the blood vessel health state evaluation value Bphz i I in (i) represents the sequential order number of each vessel segment from top to bottom, i=1, 2,..and n, wherein n is a positive integer.
8. The imaging-based vascular simulation intervention assessment system of claim 7, wherein: generating a vascular health status assessment value Bphz i The steps of (a) are as follows:
s201, parameter acquisition: the parameters include a blood vessel radius value Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i A difference from an initial value of blood flow shear stress No;
s202, parameter pretreatment: carrying out dimensionless treatment on the calculated parameters;
s203, generating a blood vessel health state evaluation value Bphz according to the preprocessed parameters i The formula is as follows:
wherein alpha, beta and gamma are respectively vascular radius values Gb i Difference from initial value Gb of blood vessel radius, blood flow velocity Vr i Difference from the initial flow velocity Vr of blood and the shear stress No of blood flow i The preset proportionality coefficient of the difference value between the initial value No of the blood flow shearing stress and the initial value No of the blood flow shearing stress, and alpha > beta > gamma > 0,G are constant correction coefficients.
9. The imaging-based vascular simulation stem of claim 8A pre-evaluation system, characterized by: blood vessel health state evaluation value Bphz i The comparison with the preset threshold value results in the following:
if it is the blood vessel health state evaluation value Bphz i If the threshold value is exceeded, no response is made;
if it is the blood vessel health state evaluation value Bphz i If the threshold value is not exceeded, an early warning is sent out, and stroboscopic red light is sent out at the position corresponding to the blood vessel segment on the geometric model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311363032.3A CN117198510B (en) | 2023-10-20 | 2023-10-20 | Vascular simulation intervention evaluation system based on imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311363032.3A CN117198510B (en) | 2023-10-20 | 2023-10-20 | Vascular simulation intervention evaluation system based on imaging |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117198510A true CN117198510A (en) | 2023-12-08 |
| CN117198510B CN117198510B (en) | 2024-09-06 |
Family
ID=88990757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311363032.3A Active CN117198510B (en) | 2023-10-20 | 2023-10-20 | Vascular simulation intervention evaluation system based on imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117198510B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830848A (en) * | 2018-05-25 | 2018-11-16 | 深圳科亚医疗科技有限公司 | The device and system of the sequence of the vascular condition parameter on blood vessel are determined using computer |
| CN109979593A (en) * | 2018-09-24 | 2019-07-05 | 北京昆仑医云科技有限公司 | Prediction technique, the prediction technique of the candidate stenosis of blood flow paths, blood vessel stenosis prediction meanss of the healthy radius of blood flow paths |
| CN110853029A (en) * | 2017-11-15 | 2020-02-28 | 深圳科亚医疗科技有限公司 | Method, system, and medium for automatically predicting blood flow characteristics based on medical images |
| CN111680447A (en) * | 2020-04-21 | 2020-09-18 | 深圳睿心智能医疗科技有限公司 | Blood flow characteristic prediction method, blood flow characteristic prediction device, computer equipment and storage medium |
| CN114596311A (en) * | 2022-04-22 | 2022-06-07 | 深圳科亚医疗科技有限公司 | Blood vessel function evaluation method and blood vessel function evaluation device based on blood vessel image |
-
2023
- 2023-10-20 CN CN202311363032.3A patent/CN117198510B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110853029A (en) * | 2017-11-15 | 2020-02-28 | 深圳科亚医疗科技有限公司 | Method, system, and medium for automatically predicting blood flow characteristics based on medical images |
| CN108830848A (en) * | 2018-05-25 | 2018-11-16 | 深圳科亚医疗科技有限公司 | The device and system of the sequence of the vascular condition parameter on blood vessel are determined using computer |
| CN109979593A (en) * | 2018-09-24 | 2019-07-05 | 北京昆仑医云科技有限公司 | Prediction technique, the prediction technique of the candidate stenosis of blood flow paths, blood vessel stenosis prediction meanss of the healthy radius of blood flow paths |
| CN111680447A (en) * | 2020-04-21 | 2020-09-18 | 深圳睿心智能医疗科技有限公司 | Blood flow characteristic prediction method, blood flow characteristic prediction device, computer equipment and storage medium |
| WO2021213124A1 (en) * | 2020-04-21 | 2021-10-28 | 深圳睿心智能医疗科技有限公司 | Blood flow feature prediction method and apparatus, computer device, and storage medium |
| CN114596311A (en) * | 2022-04-22 | 2022-06-07 | 深圳科亚医疗科技有限公司 | Blood vessel function evaluation method and blood vessel function evaluation device based on blood vessel image |
Non-Patent Citations (1)
| Title |
|---|
| 康皓贝;孙耀宗;陈杰;: "基于深度卷积网络的冠脉血管分割方法", 电子技术与软件工程, no. 06, 15 March 2020 (2020-03-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117198510B (en) | 2024-09-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10249048B1 (en) | Method and system for predicting blood flow features based on medical images | |
| EP3723038B1 (en) | Fast calculation method and system employing plaque stability index of medical image sequence | |
| US20210358634A1 (en) | Systems and methods for image processing to determine blood flow | |
| CN110866914B (en) | Evaluation method, system, equipment and medium for cerebral aneurysm hemodynamic index | |
| CN104736046B (en) | System and method for number evaluation vascular system | |
| CN107427268A (en) | Method and system for the blood flow reserve fraction based on pure geometry machine learning | |
| CN108665449B (en) | Deep learning model and device for predicting blood flow characteristics on blood flow vector path | |
| CN108735270A (en) | Blood flow reserve score acquisition methods, device, system and computer storage media based on dimensionality reduction model | |
| KR20160031026A (en) | Systems and methods for estimating blood flow characteristics from vessel geometry and physiology | |
| CN111588367A (en) | Heart rate detection method and device and computer readable storage medium | |
| CN115861172A (en) | Wall motion estimation method and device based on self-adaptive regularized optical flow model | |
| CN113077435A (en) | Aneurysm analysis method and device based on hemodynamics | |
| CN117198510B (en) | Vascular simulation intervention evaluation system based on imaging | |
| JP7446277B2 (en) | Colocalization detection of retinal perfusion and optic disc deformation | |
| US10332255B2 (en) | Method for assessing stenosis severity in a lesion tree through stenosis mapping | |
| Mehta et al. | Heart Disease Diagnosis using Deep Learning | |
| Abdalla | Evaluating Retinal Blood Vessels’ Abnormal Tortuosity in Digital Image Fundus Submitted in fulfilment of the requirements for the Degree of MSc by research |
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 |