CN113611423A - Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model - Google Patents
Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model Download PDFInfo
- Publication number
- CN113611423A CN113611423A CN202110817557.4A CN202110817557A CN113611423A CN 113611423 A CN113611423 A CN 113611423A CN 202110817557 A CN202110817557 A CN 202110817557A CN 113611423 A CN113611423 A CN 113611423A
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- maximum principal
- predicting
- principal strain
- 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.)
- Pending
Links
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 38
- 208000014674 injury Diseases 0.000 title claims abstract description 24
- 230000008733 trauma Effects 0.000 title claims abstract description 18
- 238000004088 simulation Methods 0.000 claims abstract description 29
- 210000004556 brain Anatomy 0.000 claims abstract description 20
- 238000013507 mapping Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 13
- 210000003128 head Anatomy 0.000 claims description 24
- 210000005013 brain tissue Anatomy 0.000 claims description 17
- 210000001638 cerebellum Anatomy 0.000 claims description 11
- 210000000877 corpus callosum Anatomy 0.000 claims description 11
- 210000000133 brain stem Anatomy 0.000 claims description 8
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 210000001061 forehead Anatomy 0.000 claims description 5
- 210000003455 parietal bone Anatomy 0.000 claims description 4
- 229910052742 iron Inorganic materials 0.000 claims description 3
- 210000003582 temporal bone Anatomy 0.000 claims description 3
- 210000004720 cerebrum Anatomy 0.000 claims description 2
- 230000008859 change Effects 0.000 claims description 2
- 210000000103 occipital bone Anatomy 0.000 claims description 2
- 238000002790 cross-validation Methods 0.000 claims 1
- 239000002023 wood Substances 0.000 claims 1
- 208000029028 brain injury Diseases 0.000 description 9
- 230000006378 damage Effects 0.000 description 8
- 208000027418 Wounds and injury Diseases 0.000 description 6
- 206010019196 Head injury Diseases 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 238000011158 quantitative evaluation Methods 0.000 description 5
- 208000030886 Traumatic Brain injury Diseases 0.000 description 4
- 230000000451 tissue damage Effects 0.000 description 4
- 231100000827 tissue damage Toxicity 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 239000007799 cork Substances 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 208000033986 Device capturing issue Diseases 0.000 description 1
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000006931 brain damage Effects 0.000 description 1
- 231100000874 brain damage Toxicity 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- FKLFBQCQQYDUAM-UHFFFAOYSA-N fenpiclonil Chemical compound ClC1=CC=CC(C=2C(=CNC=2)C#N)=C1Cl FKLFBQCQQYDUAM-UHFFFAOYSA-N 0.000 description 1
- 239000011121 hardwood Substances 0.000 description 1
- 230000035876 healing Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 230000029663 wound healing Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/41—Medical
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Public Health (AREA)
- Computer Graphics (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Geometry (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses a craniocerebral maximum principal strain prediction method of blunt trauma based on a convolutional neural network model, which comprises the following steps: step 1: carrying out finite element simulation on the hitting head of the blunt instrument; step 2: extracting the maximum principal strain of the cranium and the brain in the simulation data; and step 3: extracting a speed curve of the blunt instrument along the direction X, Y, Z in the global coordinate system in the simulation data; cutting and filling the speed curve, and mapping and scaling the speed curve to obtain a mapping chart; the mapping graph comprises an X axial speed, a Y axial speed and a Z axial speed; and 4, step 4: constructing a convolutional neural network for training and predicting the influence of the striking head of a blunt instrument on the maximum main strain of the cranium; and 5: dividing the maximum principal strain of the mapping chart and the cranium into input and output to train the convolutional neural network, and selecting an optimal model; step 6: and using the optimal model for predicting the maximum main strain of the cranium and the brain.
Description
Technical Field
The invention relates to the technical field of deep learning and finite element simulation calculation, in particular to a rapid quantitative evaluation method for craniocerebral injury caused by hitting head with a stick based on a convolutional neural network and a finite element technology.
Background
The craniocerebra is the most important life center of the human body, and the fatality rate and disability rate after injury are extremely high. Brain damage is one of the most major public health problems in the world. According to the world health organization data, over 4000 million people worldwide each year suffer from mild brain injury. In forensic identification, blunt blow is one of the major factors causing craniocerebral injury. Wherein, the club blunt instrument has the highest ratio. However, the current quantitative evaluation method for blunt-induced brain injury is very limited. Along with the development of injury biomechanics and the combination and application of the injury biomechanics and a finite element technology, the biomechanics research and judicial identification by means of a finite element method are applied more and more. Especially in the study of craniocerebral injuries. However, the finite element simulation experiment requires a large amount of computation time and a high-performance workstation, and requires a certain professional knowledge of an operator. This clearly limits the spread and application of finite element techniques. To solve this problem, some simplified physical models for predicting brain injury indicators are proposed. These models take as input and output, respectively, the head kinematics data, the corresponding finite element simulation results (brain tissue maximum principal strain). And determining an optimal model by adjusting parameters, and realizing quantitative evaluation on the craniocerebral injury. Compared with the traditional injury standard, such as a brain injury evaluation criterion (HIC), a head impact force (HIP), an injury threshold (GAMBIT) based on a generalized acceleration model brain, and the like, the complexity and the accuracy of the model are improved. However, the prior art has the following disadvantages:
1. finite element simulation experiments require a large amount of computing time and high performance workstations and require operators to have certain expertise.
2. For larger strain impact and more complicated healing factors, the prediction accuracy of the simplified model is greatly reduced. This is due to the failure to capture significant non-linearity at higher impact severity.
3. The prior art model can only carry out overall quantitative evaluation on brain tissues and cannot carry out analysis and evaluation on local brain tissues.
Disclosure of Invention
In view of the defects of the prior art, the invention aims to provide a method for predicting the craniocerebral maximum principal strain of a clubber blunt trauma based on a convolutional neural network model, and solves the problems of a finite element technology and a head simplified physical model.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a method for predicting the maximum principal strain of the cranium and brain of blunt trauma based on a convolutional neural network model is characterized by comprising the following steps:
step 1: carrying out finite element simulation on the hitting head of the blunt instrument;
step 2: extracting the maximum principal strain of the cranium and the brain in the simulation data;
and step 3: extracting a speed curve of the blunt instrument along the direction X, Y, Z in the global coordinate system in the simulation data; cutting and filling the speed curve, and mapping and scaling the speed curve to obtain a mapping chart; the mapping graph comprises an X axial speed, a Y axial speed and a Z axial speed;
and 4, step 4: constructing a plurality of convolutional neural networks for training and predicting the influence of the striking head of the blunt instrument on the maximum main strain of the cranium;
and 5: dividing the mapping chart and the maximum principal strain of the cranium into input and output to train the convolutional neural network, and selecting an optimal model;
step 6: and using the optimal model for predicting the maximum main strain of the cranium and the brain.
In the invention, the finite element simulation initial conditions are that a blunt instrument with a certain initial speed strikes different parts of the head, including the forehead, the left temporal bone, the left back parietal bone and the occiput.
In the present invention, the maximum principal strains of the brain, corpus callosum, cerebellum and brainstem are extracted.
In step 3 of the invention, the speed curve is translated to have consistent initial time; the curve after the blunt instrument was removed from contact with the head in the simulation was cut and filled up to 2ms at the speed of the last point after cutting.
In the invention, a convolution neural network is constructed for training and predicting the influence of the striking head of the stick on the maximum main strain of the cranium.
The structure of the convolutional neural network model in the invention comprises 3 convolutional pooling modules (the convolutional pooling module comprises a layer of convolutional layer, the sizes of convolutional kernels of which are 1 × 1, 3 × 3 and 5 × 5 respectively, a layer of pooling layer connects output values of 3 convolutional channels in parallel and pools), 1 convolutional layer (the size of convolutional kernels is 3 × 3), and 4 fully-connected layers (the number of nodes of four layers is 1024, 512, 256 and 1 respectively). The input is a blunt velocity picture with the size of 150 multiplied by 3, and the output is the maximum principal strain of brain tissue.
In the invention, a ten-fold cross verification method is adopted for a convolutional neural network training strategy; evaluating the prediction result of the convolutional neural network by adopting a mean square error, an absolute error and goodness of fit; the change range of the goodness of fit is between 0 and 1; MAE and MSE are two common indexes for measuring the precision of the predicted value, the smaller the value is, the higher the precision of the prediction is, but the MSE is greatly influenced by the abnormal value; therefore, the MSE, MAE and R are considered together2And selecting a convolutional neural network for the optimal value, and finally using the optimal convolutional neural network model for predicting the maximum main strain of the brain tissue.
The invention also provides a craniocerebral maximum principal strain prediction system of blunt trauma based on a convolutional neural network model, which comprises the following steps: a memory and a processor; the memory has stored thereon a computer program which, when executed by the processor, implements the aforementioned inventive method.
The invention also proposes a computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the aforementioned inventive method.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a method for predicting the maximum principal strain of a craniocerebral of a clubber blunt trauma based on a convolutional neural network model, which comprises the steps of synthesizing pictures of measured speeds of a clubber along X, Y, Z three directions and inputting the pictures into a trained and verified convolutional neural network to obtain the maximum principal strain of a brain, a corpus callosum, a cerebellum and a brainstem. The damage degree of different brain tissues can be judged by utilizing the maximum principal strain damage criterion. Compared with the finite element technology, the invention reduces the requirement on hardware equipment, and changes the task finished on a high-end workstation by the finite element technology into the task finished on a common notebook computer. In addition, finite element simulations typically take several hours or even tens of hours. However, the method only needs a few seconds or even tens of milliseconds for predicting the brain injury, and the efficiency is remarkably improved. Compared with a simplified physical model and a traditional brain injury standard, the method improves the complexity and the accuracy of the model. At the same time, the invention still maintains higher accuracy for larger strain impact and more complicated wound healing factors.
Drawings
FIG. 1 is a technical flow diagram.
Fig. 2 is a convolutional neural network structure.
FIG. 3 shows the results of the convolutional neural network prediction on the brain, corpus callosum, cerebellum and brainstem.
Fig. 4 is a speed picture of the blunt edge X, Y, Z in example 1.
Fig. 5 is a speed picture of the blunt edge X, Y, Z in example 2.
Fig. 6 is a speed picture of the blunt edge X, Y, Z in example 3.
Detailed Description
The present invention will be described in further detail with reference to the following specific examples and the accompanying drawings. The procedures, conditions, experimental methods and the like for carrying out the present invention are general knowledge and common general knowledge in the art except for the contents specifically mentioned below, and the present invention is not particularly limited.
The invention discloses a craniocerebral maximum principal strain prediction method of a clubber blunt trauma based on a convolutional neural network algorithm, which adopts a finite element method to simulate and calculate biomechanical parameters of brain tissue damage of different clubbers at different hitting speeds; extracting the maximum main strain of brain tissue and the speed of the stick along the direction X, Y, Z under a space coordinate system, processing the obtained stick speed curve, and mapping and scaling the processed stick speed curve into a picture; and selecting the convolutional neural network as a brain tissue damage prediction model, and obtaining the convolutional neural network which can be finally applied to reality through training and selection. The evaluation method for the influence of the stick striking on the maximum main strain of the cranium based on the convolutional neural network algorithm reduces the requirement on hardware equipment, improves the calculation efficiency, and realizes the rapid quantitative evaluation of the cranium injury compared with the traditional evaluation method. Can provide reference data for forensic identification and craniocerebral injury mechanism analysis.
The invention provides a method for predicting the craniocerebral maximum principal strain of a clubber blunt trauma based on a convolutional neural network model, and a technical flow chart is shown in figure 1. The specific implementation comprises the following steps:
(1) finite element simulation of the blunt instrument striking head: the finite element simulation initial conditions are set as that the club blunt instrument with certain initial speed strikes different parts of the head, namely the forehead, the left temporal bone, the left back parietal bone and the occipital bone; in the case of brain injury caused by blunt instrument striking, the wooden stick and the iron stick are respectively positioned at the first and the second parts, and the blunt instruments of the other types have lower proportion and are not common; the invention adopts 18 kinds of wooden sticks and 10 kinds of iron sticks as the hurting tools of the finite element simulation experiment of the hitting head of the blunt instrument, and the specific parameters are shown in the table 1; adopting Tianjin science and technology university 50 in finite element simulation experimentthPercentile adult head finite element models with detailed anatomical structures including brains, cerebellums, corpus callosum, brainstem, etc.; when the simulation experiment is carried out, the stick strikes five different parts of the head at the speeds of 10m/s, 15 m/s, 20m/s, and 700 groups of striking experiments of 18 kinds of sticks, 5 speeds and 5 parts are carried out;
TABLE 1 Blunt construction parameters
(2) Extracting finite element simulation data: extracting Maximum Principal Strains (MPS) of the brain, the corpus callosum, the cerebellum and the brainstem as a basis for judging the brain tissue damage by utilizing the finite element simulation experiment obtained in the step (1); in order to protect the integrity of the input information of the convolutional neural network, the speed curve of the stick along the direction X, Y, Z under a global coordinate system is extracted in a simulation experiment; because, compared with the composite speed of the stick, the speeds along the three X, Y, Z directions can reflect the stick information of the blunt instrument during the striking process;
(3) data processing: in the step (2), because the initial distances of the stick and the head finite element model are different during simulation, the time points of starting striking are different when the speed curve of the stick is reflected; different hitting speeds and different types of stick contact time with the head are different, and in addition, the convolutional neural network needs a fixed input size; therefore, we need to further process the extracted speed curve; first, the velocity profile is translated so that it has a consistent start time; secondly, cutting out a curve of the simulation medium in which the stick is not contacted with the head, and filling the curve at the last point of the cut curve for 2ms, wherein the acceleration of the filled speed curve is zero, so that the final result is not influenced; finally, in order to reduce the size of the image input by the convolutional neural network and reduce the characteristics, the speed curve is mapped and scaled, and the obtained mapping graph is divided into an upper part, a middle part and a lower part which are respectively an X axial speed, a Y axial speed and a Z axial speed;
(4) constructing a convolutional neural network: constructing a convolutional neural network for training and predicting the influence of the striking head of the stick on the maximum main strain of the cranium;
(5) training and selecting a convolutional neural network: according to the method, the speed synthesis pictures of the stick obtained in the step (3) along X, Y, Z three directions and the maximum brain tissue principal strain extracted in the step (2) are respectively used as input and output training convolutional neural networks; convolutional neural network training to prevent overfitting due to small data sizeThe strategy adopts a ten-fold cross verification method; the estimation of the prediction result of the convolutional neural network adopts Mean Square Error (MSE), absolute error (MAE) and goodness of fit (R, R)2);R2The fitting goodness of the model is reflected, the variation range is between 0 and 1, and the closer to 1, the better the fitting effect of the model is; MAE and MSE are two common indexes for measuring the precision of the predicted value, the smaller the value is, the higher the precision of the prediction is, but the MSE is greatly influenced by the abnormal value; therefore, the MSE, MAE and R are considered together2And selecting a convolutional neural network for the optimal value, and finally using the optimal convolutional neural network model for predicting the maximum main strain of the brain tissue. Specifically, the method comprises the following steps:
700 pieces of maximum principal strain data of the cerebrum, the corpus callosum, the cerebellum and the brainstem are respectively obtained in the step (2), 700 processed speed pictures are correspondingly obtained in the step (3), 600 sticks obtained in the step (3) are synthesized into a picture along the X, Y, Z speed, the maximum main strain of brain tissue extracted in the step (2) is respectively used as input and output to train and optimize a convolutional neural network, the convolutional neural network structure in the figure 2 is finally selected through training and optimization, the convolutional neural network is used for predicting the maximum main strain of the brain tissue, the figure 3 is the prediction result of the convolutional neural network on the brain, the corpus callosum, the cerebellum and the brainstem, according to the result, the convolutional neural network can be suitable for predicting brain tissue damage data.
Examples
1) The maximum principal strain is predicted when a round stick of 400mm long and 55mm diameter made of hardwood strikes the left forehead at 10 m/s.
FIG. 4 shows the velocity curve of the stick along direction X, Y, Z, after the mapping scaling, and the introduction of the above-mentioned selected convolutional neural network, the prediction results of the brain, cerebellum and corpus callosum are 0.515, 0.532 and 0.395, respectively. The real values are 0.501, 0.529, 0.430, respectively.
2) The maximum principal strain is predicted when a round stick of cork 400mm long and 40mm in diameter strikes the left posterior parietal bone at 20 m/s.
FIG. 5 shows the velocity curve of the round stick along direction X, Y, Z, after the mapping and scaling, the prediction results of the brain, cerebellum and corpus callosum are 0.682, 0.453 and 0.175 respectively after the round stick is introduced into the selected convolutional neural network. The real values are 0.661, 0.459, and 0.182, respectively.
3) The maximum principal strain is predicted when a round stick of 400mm long and 20mm diameter made of cork strikes the left forehead at 30 m/s.
FIG. 6 shows the velocity curve of the round stick along direction X, Y, Z, after the mapping scaling, the prediction results of the brain, cerebellum and corpus callosum are 0.769, 0.408 and 0.080 respectively after the round stick is introduced into the selected convolutional neural network. The actual values were 0.763, 0.423, and 0.089, respectively.
The method for predicting the maximum principal strain of the craniocerebral of the clubber blunt trauma based on the convolutional neural network model improves the calculation efficiency and reduces the requirements on hardware equipment. Compared with the traditional evaluation method, the method improves the complexity and the prediction precision of the model. Can provide brain injury diagnosis value for forensic identification and clinical treatment.
The protection of the present invention is not limited to the above embodiments. Variations and advantages that may occur to those skilled in the art may be incorporated into the invention without departing from the spirit and scope of the inventive concept, which is set forth in the following claims.
Claims (10)
1. A method for predicting the maximum principal strain of the cranium and brain of blunt trauma based on a convolutional neural network model is characterized by comprising the following steps:
step 1: carrying out finite element simulation on the hitting head of the blunt instrument;
step 2: extracting the maximum principal strain of the cranium and the brain in the simulation data;
and step 3: extracting a speed curve of the blunt instrument along the direction X, Y, Z in the global coordinate system in the simulation data; cutting and filling the speed curve, and mapping and scaling the speed curve to obtain a mapping chart; the mapping graph comprises an X axial speed, a Y axial speed and a Z axial speed;
and 4, step 4: constructing a convolutional neural network for training and predicting the influence of the striking head of a blunt instrument on the maximum main strain of the cranium;
and 5: dividing the maximum principal strain of the mapping chart and the cranium into input and output to train the convolutional neural network, and selecting an optimal model;
step 6: and using the optimal model for predicting the maximum main strain of the cranium and the brain.
2. The method of claim 1, wherein the initial condition of finite element simulation is that a blunt object with a certain initial velocity strikes different parts of the head.
3. The convolutional neural network model-based craniocerebral maximum principal strain prediction method for blunt trauma as claimed in claim 1, wherein the maximum principal strains of the cerebrum, corpus callosum, cerebellum and brainstem are extracted.
4. The method for predicting the maximum principal strain of the cranium and brain of the blunt trauma based on the convolutional neural network model as claimed in claim 1, wherein in step 3, the velocity curve is translated to have a consistent starting time; the curve after the blunt instrument was removed from contact with the head in the simulation was cut and filled up to 2ms at the speed of the last point after cutting.
5. The method for predicting the craniocerebral maximum principal strain of blunt trauma based on the convolutional neural network model as claimed in claim 1, wherein the convolutional neural network model is constructed for training and predicting the influence of a stick striking head on the craniocerebral maximum principal strain;
the structure of the convolutional neural network model comprises 3 convolutional pooling modules, 1 convolutional layer and 4 full-connection layers; the input is a blunt instrument speed picture, and the output is the maximum principal strain of brain tissue.
6. The method of claim 5, wherein the convolutional neural network model-based craniocerebral maximum principal strain prediction method for blunt trauma comprises a convolutional layer, and the convolutional kernel size of the convolutional layer is 1 × 1, 3 × 3 or 5 × 5; a pooling layer for connecting the output values of the 3 convolution channels in parallel and pooling;
the convolution kernel size of the convolution layer is 3 x 3;
the number of nodes of the four layers of the full connection layer is 1024, 512, 256 and 1 respectively;
the blunter velocity picture size is 150 × 150 × 3.
7. The method for predicting the craniocerebral maximum principal strain of blunt trauma based on the convolutional neural network model as claimed in claim 1, wherein the convolutional neural network training strategy adopts a ten-fold cross-validation method; evaluating the prediction result of the convolutional neural network by adopting a mean square error, an absolute error and goodness of fit; the change range of the goodness of fit is between 0 and 1; taking MSE, MAE and R into account2Selecting a convolutional neural network according to the optimal value of the optimal value, and finally using the optimal convolutional neural network model for predicting the maximum main strain of the brain tissue.
8. The method for predicting the craniocerebral maximum principal strain of blunt trauma based on a convolutional neural network model as set forth in claim 1, wherein the types of the sticks used in the simulation experiment of step (1) include iron and wood, the striking position of the head includes forehead, left temporal bone, left posterior parietal bone and occipital bone, and the speed of the stick is 10, 15, 20, 25, 30 m/s.
9. A brain maximum principal strain prediction system of blunt trauma based on a convolutional neural network model, comprising: a memory and a processor;
the memory has stored thereon a computer program which, when executed by the processor, implements the method of any of claims 1-8.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110817557.4A CN113611423A (en) | 2021-07-20 | 2021-07-20 | Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110817557.4A CN113611423A (en) | 2021-07-20 | 2021-07-20 | Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113611423A true CN113611423A (en) | 2021-11-05 |
Family
ID=78337959
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110817557.4A Pending CN113611423A (en) | 2021-07-20 | 2021-07-20 | Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113611423A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140081601A1 (en) * | 2010-12-07 | 2014-03-20 | Wayne State University | Model-based helmet design to reduce concussions |
CN111009323A (en) * | 2019-11-12 | 2020-04-14 | 河北工业大学 | KNN-ANN-based prediction method for subdural hematoma injury |
CN111667580A (en) * | 2020-05-08 | 2020-09-15 | 司法鉴定科学研究院 | Injury mode reconstruction system and method for skull base fracture |
-
2021
- 2021-07-20 CN CN202110817557.4A patent/CN113611423A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140081601A1 (en) * | 2010-12-07 | 2014-03-20 | Wayne State University | Model-based helmet design to reduce concussions |
CN111009323A (en) * | 2019-11-12 | 2020-04-14 | 河北工业大学 | KNN-ANN-based prediction method for subdural hematoma injury |
CN111667580A (en) * | 2020-05-08 | 2020-09-15 | 司法鉴定科学研究院 | Injury mode reconstruction system and method for skull base fracture |
Non-Patent Citations (2)
Title |
---|
XIANGHAO ZHANG等: ""convolutional neural network for efcient estimation of regional brain strains"", 《SCIENTIFIC REPORTS》, pages 1 - 9 * |
高勇等: ""颅脑拳击损伤的参数化分析"", 《法医学杂志》, no. 3, pages 1 - 3 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107492099B (en) | Medical image analysis method, medical image analysis system, and storage medium | |
CN109508644B (en) | Facial paralysis grade evaluation system based on deep video data analysis | |
CN109409416A (en) | Feature vector dimension reduction method and medical image recognition method, apparatus and storage medium | |
KR20210021039A (en) | Image processing methods, devices, electronic devices and computer-readable storage media | |
CN111008974A (en) | Multi-model fusion femoral neck fracture region positioning and segmentation method and system | |
JP7189257B2 (en) | Method, apparatus and computer readable storage medium for detecting specific facial syndromes | |
Johnson et al. | Constrained topological optimization of a football helmet facemask based on brain response | |
CN113096137B (en) | Adaptive segmentation method and system for OCT (optical coherence tomography) retinal image field | |
CN112489769A (en) | Intelligent traditional Chinese medicine diagnosis and medicine recommendation system for chronic diseases based on deep neural network | |
Shim et al. | Rapid prediction of brain injury pattern in mTBI by combining FE analysis with a machine-learning based approach | |
CN112633416A (en) | Brain CT image classification method fusing multi-scale superpixels | |
CN112288645A (en) | Skull face restoration model construction method, restoration method and restoration system | |
CN116110597B (en) | Digital twinning-based intelligent analysis method and device for patient disease categories | |
CN113782184A (en) | Cerebral apoplexy auxiliary evaluation system based on facial key point and feature pre-learning | |
CN112508902A (en) | White matter high signal grading method, electronic device and storage medium | |
CN106844994B (en) | Choroidal neovascularization growth prediction method combining constitutive model and finite element | |
Du et al. | Efficient network construction through structural plasticity | |
Wu et al. | Instantaneous brain strain estimation for automotive head impacts via deep learning | |
Bourdet et al. | Deep learning methods applied to the assessment of brain injury risk | |
Krishna et al. | Automated detection of common maternal fetal ultrasound planes using deep feature fusion | |
CN112052843B (en) | Face key point detection method from coarse face to fine face | |
Tu et al. | DRPAN: a novel adversarial network approach for retinal vessel segmentation | |
Deck et al. | Deep learning method to assess brain injury risk | |
CN113611423A (en) | Method and system for predicting maximum principal strain of craniocerebral of blunt trauma based on convolutional neural network model | |
CN112613445A (en) | Face image generation method and device, computer equipment and storage medium |
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 |