CN113705591A - Readable storage medium, and support specification identification method and device - Google Patents
Readable storage medium, and support specification identification method and device Download PDFInfo
- Publication number
- CN113705591A CN113705591A CN202010430500.4A CN202010430500A CN113705591A CN 113705591 A CN113705591 A CN 113705591A CN 202010430500 A CN202010430500 A CN 202010430500A CN 113705591 A CN113705591 A CN 113705591A
- Authority
- CN
- China
- Prior art keywords
- stent
- target
- size
- image
- preset
- 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
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 238000003062 neural network model Methods 0.000 claims abstract description 37
- 238000004519 manufacturing process Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 27
- 230000003321 amplification Effects 0.000 claims description 15
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 15
- 238000005452 bending Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000001514 detection method Methods 0.000 abstract description 12
- 238000007689 inspection Methods 0.000 abstract description 4
- 210000004204 blood vessel Anatomy 0.000 description 7
- 238000011176 pooling Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 208000026106 cerebrovascular disease Diseases 0.000 description 2
- 238000002788 crimping Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 206010008118 cerebral infarction Diseases 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 210000004351 coronary vessel Anatomy 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 208000010125 myocardial infarction Diseases 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 210000002254 renal artery Anatomy 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a readable storage medium, a stent specification identification method and a stent specification identification device, which are characterized in that firstly, image production standard data sets of stents with various preset specifications are obtained, and a neural network model is trained by utilizing the standard data sets; and then acquiring an image of a target stent, loading the trained neural network model, and identifying the image of the target stent so as to evaluate the score value of the target stent corresponding to each preset specification. Therefore, the detection of the specification of the support can be automatically finished without manual intervention, and the problems of high false detection rate and low efficiency in manual support specification inspection are solved.
Description
Technical Field
The invention relates to the technical field of medical instrument detection, in particular to a readable storage medium, a stent specification identification method and a stent specification identification device.
Background
The stent is implanted in a blood vessel to open a narrow and blocked blood vessel, and is mainly used for cardiovascular and cerebrovascular diseases such as coronary arteriosclerosis, myocardial infarction, cerebral infarction and the like. The stent can effectively promote the normal flow of blood, promote the recovery of the remodeling function of the blood vessel, effectively prevent the elastic retraction of the blood vessel, and some stents can also prevent the secondary contraction and blockage of the blood vessel. Mainly comprises coronary artery stents, cerebral stents, renal artery stents, aortic stents, etc.
The clinical treatment effect of the stent is witnessed, and more patients select stent interventional therapy, so the importance of the quality of the stent is more and more concerned. The size detection of the stent is indispensable in the assembly link and is also particularly important. Once the large-size stent is loaded into the balloon to flow into the market after being crimped by the crimping machine, the stent can burst the blood vessel of a patient in the operation process. At present, the size and specification detection of the bracket is mainly manual spot check, and the observation and detection under a microscope are mainly performed. And judging whether the model is a large model, a small model or a medium model. However, manual spot check testing cannot guarantee the correct specification of each product, and manual inspection is limited by the physical condition and working state of quality inspectors. In a long time, manual testing has revealed very big drawback, because intensity of labour is big for judge the mistake to support size specification, thereby lead to the false retrieval rate high, if judge big specification support for little support or medium model support etc. greatly restrict production efficiency's improvement and product quality's promotion like this, lead to unqualified product to flow into subsequent production course of working, make very big accident.
Disclosure of Invention
The invention aims to provide a readable storage medium, a support specification identification method and a support specification identification device, which are used for solving the problems of high false detection rate and low efficiency when manual support specification inspection is carried out.
In order to solve the above technical problem, the present invention provides a stent specification identification method, including:
acquiring images of stents with different preset specifications to manufacture a standard data set, and training a neural network model by using the standard data set;
and acquiring an image of a target stent, loading the trained neural network model, and identifying the image of the target stent so as to evaluate the score value of the target stent corresponding to each preset specification.
Optionally, in the stent specification identification method, the stent specification identification method further includes:
setting a trustworthiness score threshold; and the number of the first and second groups,
after the score values of the target bracket corresponding to the preset specifications are obtained through evaluation, whether the score values exceed a credible score threshold value is judged;
if yes, obtaining and outputting the label corresponding to the preset specification.
Optionally, in the stent size identification method, the sum of the fractional values of the target stent corresponding to each of the preset sizes is 1.
Optionally, in the stent specification identification method, the method for acquiring images of stents with different preset specifications includes:
and judging the size of the stent in the acquired image according to the target characteristics of the stent, if the size is the preset size, reserving the stent, and if the size is not the preset size, discarding the stent.
Optionally, in the stent specification method, the target feature includes the number of the nodes of the stent and/or the bending angle of the connecting piece for connecting two adjacent nodes.
Optionally, in the stent specification method, the method for creating a standard data set includes:
and adjusting the sizes of the images of the supports with different preset specifications to a selected size, a selected name and a selected format, unifying the images and classifying the images according to the specifications.
Optionally, in the stent specification identification method, before identifying the image of the target stent, the stent specification identification method further includes:
adjusting the image of the target stent to the selected size.
Optionally, in the stent specification identifying method, before the training of the neural network model by using the standard data set, the method for training the neural network model further includes:
and judging whether the data volume of the standard data set is larger than a set value or not, and if not, performing data amplification on the standard data set.
The present invention also provides a stent specification identification device, comprising:
the image acquisition module is used for acquiring images of the stent, including an image of the stent for making a standard data set and an image of a target stent;
the standard data set manufacturing module is used for acquiring images of the stents with different preset specifications from the images of the stents acquired by the image acquisition module so as to manufacture the standard data set;
the model training and storing module is used for training a neural network model by utilizing the standard data set;
and the support specification identification module is used for loading the trained neural network model to identify the image of the target support so as to evaluate the score value of the target support corresponding to each preset specification.
Optionally, in the stent specification identification device, the stent specification identification device further includes: a trustworthiness score threshold setting module for setting a trustworthiness score threshold;
the support specification identification module is further used for judging that the score value exceeds a credible score threshold value after the score value of the target support corresponding to each preset specification is obtained through evaluation, and if the score value exceeds the credible score threshold value, obtaining and outputting the labels corresponding to the preset specifications.
Optionally, in the stent size identification device, a total of fractional values of the target stent corresponding to each of the preset sizes is 1.
Optionally, in the stent specification identification device, the method for acquiring images of stents with different preset specifications by the standard data set making module includes:
and judging the size of the stent in the acquired image according to the target characteristics of the stent, if the size is the preset size, reserving the stent, and if the size is not the preset size, discarding the stent.
Optionally, in the stent specification identifying device, the target feature includes the number of the nodes of the stent and/or the bending angle of the connecting piece for connecting two adjacent nodes.
Optionally, in the stent specification identifying device, the method for making the standard data set by the standard data set making module includes:
and adjusting the sizes of the images of the supports with different preset specifications to a selected size, a selected name and a selected format, unifying the images and classifying the images according to the specifications.
Optionally, in the stent specification identification device, the stent specification identification device further includes: an image resizing module to resize the image of the target stent to the selected size.
The invention also provides a readable storage medium, wherein a computer program is stored in the readable storage medium, and when the computer program is executed by a processor, the computer program realizes the stent specification identification method.
In summary, in the readable storage medium, the stent specification identification method and the device provided by the present invention, images of stents with different preset specifications are first obtained to make a standard data set, and a neural network model is trained by using the standard data set; and then acquiring an image of a target stent, loading the trained neural network model, and identifying the image of the target stent so as to evaluate the score value of the target stent corresponding to each preset specification. Therefore, the detection of the specification of the support can be automatically finished without manual intervention, and the problems of high false detection rate and low efficiency in manual support specification inspection are solved.
Drawings
FIG. 1 is a schematic structural view of a stent according to an embodiment of the present invention;
FIG. 2 is a flowchart of a method for identifying a stent specification according to an embodiment of the present invention;
FIG. 3 is a block diagram of a device for identifying specifications of a stent according to an embodiment of the present invention;
wherein the reference numerals are as follows:
11-section bar; 12-a connector; 100-an image acquisition module; 200-a standard data set making module; 300-a model training and saving module; 400-a stent specification identification module; 500-a trustworthiness score threshold setting module; 600-a data amplification module; 700-image resizing module.
Detailed Description
The readable storage medium, the method for identifying the stent specification and the device provided by the invention are further described in detail with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It is to be noted that the drawings are in a very simplified form and are not to precise scale, which is merely for the purpose of facilitating and distinctly claiming the embodiments of the present invention. Further, the structures illustrated in the drawings are often part of actual structures. In particular, the drawings may have different emphasis points and may sometimes be scaled differently.
Referring to fig. 1, the structural schematic diagram of a stent is shown, the stent includes a plurality of node rods 11 and a plurality of connecting pieces 12, the plurality of node rods 11 are sequentially arranged along an axial direction, and the connecting pieces 12 connect two adjacent node rods 11.
The number of the section rods 11 is different, so that the support presents different specifications, and the specification of the support also presents an increasing trend along with the increase of the number of the section rods 11. Since the adjacent link rods 11 are connected through the connecting member 12, the stent has a shape-changing degree of freedom in the axial direction, but as described above, once a large-sized stent is loaded into a balloon after being crimped by a crimping machine, the stent may burst a blood vessel of a patient during an operation, and thus the specification of the stent needs to be identified in advance.
In view of this, as shown in fig. 2, an embodiment of the present invention provides a method for identifying a stent specification, where the method includes the following steps:
s1, acquiring images of the stent with various preset specifications to manufacture a standard data set, and training a neural network model by using the standard data set;
s2, acquiring an image of the target stent;
and S3, loading the trained neural network model, and identifying the image of the target stent to evaluate the score value of the target stent corresponding to each preset specification.
Namely, the stent specification identification method provided by the embodiment of the invention can automatically complete the detection of the specification of the stent without manual intervention, thereby solving the problems of high false detection rate and low efficiency when the specification of the stent is manually checked.
The above steps will be described in detail below.
Generally, when a large number of pictures are collected for a neural network model, the picture formats and names may be inconsistent, and in addition, for training of the neural network model, certain requirements are imposed on the image size, the original image size of the stent is related to the resolution of the shooting device, and when the resolution of the shooting device is higher, the shot pictures may be too large to be used for training of the neural network model.
Therefore, in step S1, the method for acquiring images of stents with different preset specifications includes: and judging the size of the stent in the acquired image according to the target characteristics of the stent, if the size is the preset size, reserving the stent, and if the size is not the preset size, discarding the stent.
As mentioned above, the number of stent nodes may be varied to achieve different specifications, and thus, in one embodiment, the target feature may be set to the number of stent nodes S, M, L for training the neural network model according to the number of stent nodes when the standard data set is created. For example, S may indicate the specification of the stent with the number of the node bars being 4, M may indicate the specification of the stent with the number of the node bars being 8, L may indicate the specification of the stent with the number of the node bars being 12, and S, M, L may specifically correspond to the number of the node bars being adjusted according to the number of the node bars of the target stent. When a large number of images are acquired for creating the standard data set, images of stents other than (S, M, L) may be included in the acquired images, and therefore, after the images are acquired, the acquired images should be first screened to discard images of stents other than the preset specification (S, M, L).
In another embodiment, the bending angle of the connecting piece between two adjacent rods can be obtained, and the specification of the bracket can be judged according to the bending angle, and the smaller the bending angle of the connecting piece is, the larger the specification of the bracket is.
In addition, the target feature may also be a combination of the number of the nodal rods and the bending angle of the connecting piece. And when the method is used for the network model training, the actually adopted characteristics can not be limited to the number of the section rods and the bending angles among the section rods.
After acquiring images of stents of different preset specifications, the method for making a standard data set may comprise: and adjusting the sizes of the images of the supports with different preset specifications to a selected size, a selected name and a selected format for unification, and classifying the images according to the specifications.
Specifically, whether the names and formats of the images are uniform is judged, if not, the names and formats of the images are modified, so that the names and formats of all the images in the standard data set are the same, and the formats of all the images are consistent; and determining whether the size of the image meets the criteria, and if not, resizing each image to a selected size, for example, if the original picture size is 2448 × 2048, it may be resized to 224 × 224 for training of the neural network model. The adjustment of format/name, the adjustment of size and the classification of pictures are not in sequence.
When training a neural network model, generally, the standard data set is created to include a training data set and a test data set, so in this embodiment, the training data set and the test data set are both composed of images of scaffolds with three specifications (S, M, L), and names and formats of the images in the training data set and the test data set are uniform.
In the present embodiment, the stent specification identification method is explained by three preset specifications (S, M, L). In some other embodiments, to further improve the testing accuracy, the specification may be further subdivided, for example, into: XS, S, M, L and XL, the number of the nodes corresponding to each specification is gradually increased, which is not illustrated, but it should be understood that even if the kind of the preset specification is changed, the sum of the finally evaluated score values of all the preset specifications should be kept constant, and may be 1, for example. In other embodiments, the sum of all fractional values may also take 100.
In addition, when the neural network model is trained, certain requirements are required for the number of images, and the number of the images is generally required to be at least 2000-3000. Therefore, in this embodiment, preferably, before the training of the neural network model by using the standard data set, the method for training the neural network model further includes: and judging whether the data volume of the standard data set is enough, namely whether the data volume is larger than a set value, and if not, performing data amplification on the standard data set. The method for performing data amplification comprises one or more of brightness amplification, rotation amplification and flip amplification.
In this embodiment, preferably, the neural network model is a neural network model that uses a Global Pooling layer (GAP) instead of a fully connected layer. In a common convolutional neural network, such as a DenseNet201 model, a convolutional layer before a fully-connected layer is responsible for extracting features of an image, after the features are obtained, a traditional method is to connect the fully-connected layer and then perform activation classification, and the idea of a global pooling layer is to replace the fully-connected layer with the global pooling layer (i.e., dimension reduction is performed in a global pooling layer manner), and the more important point is to keep spatial information and semantic information extracted by each convolutional layer and the pooling layer, so that the effect is obviously improved in practical application. In addition, the global pooling layer removes the limitation on the size of the number of input pictures and can minimize overfitting by reducing the total number of parameters in the model.
When the trained neural network model is used for training the image of the target stent, the image size of the target stent needs to be consistent with the image size selected when the standard data set is manufactured, so that the specification can be accurately identified. When the image of the target stent is obtained, similarly, due to the influence of the resolution of the shooting device, it may not be guaranteed that the size of the shot image is exactly the same as the size of the image (i.e., the set size) selected for the neural network model training. Therefore, in this embodiment, after the step S2 is completed and before the step S3, the stent specification identification method may further include: adjusting the image of the target stent to the selected size to enable the trained neural network model identification specification for the target stent.
In step S3, after the image to be detected is input, the trained neural network model is loaded to identify the specifications of the stent in the input image, so as to evaluate the score values of the target stent corresponding to each of the preset specifications, for example, if the sum of the score values is 1, if the score values obtained by identification are [0.97, 0.02, 0.01] corresponding to the three specifications [ L, M, S ], the probability that the target stent is in the L specification is the greatest.
Preferably, the stent specification identification method further includes: setting a credible score threshold, after the score value of the target support corresponding to each preset specification is obtained through evaluation, judging whether the score value exceeds the credible score threshold, if so, acquiring and outputting the label corresponding to the preset specification. After the confidence score threshold is set, when the prediction score is lower than the threshold, the specification of the stent can not be recognized well by the network result. For example, taking the total of the score values as 1, the confidence score threshold value can be set to be 0.75, if the score values corresponding to the three specifications [ L, M, S ] are identified to be [0.97, 0.02, 0.01], and it is obvious that the score value of the L specification is 0.97 exceeding the set confidence score threshold value of 0.75, so the label of the L specification is finally output. The credibility threshold value is preset artificially, different products are obtained, and the credibility threshold value may be different and can be obtained through calibration specifically. In addition, the corresponding label of each specification may be set in advance, for example, the label of L specification may be set to 1, the label of M specification may be set to 2, and the label of S specification may be set to 3.
Based on the same idea, an embodiment of the present invention further provides a stent specification identification device, where the stent specification identification device includes:
an image acquisition module 100 for acquiring images of the stent, including an image of the stent used to make the standard dataset and an image of the target stent;
a standard data set creating module 200, configured to obtain images of stents with different preset specifications from the images of stents obtained by the image obtaining module to create a standard data set;
a model training and saving module 300, configured to train a neural network model using the standard data set;
a stent specification identification module 400, configured to load the trained neural network model to identify the image of the target stent with the image adjusted to the set size, so as to evaluate a score value of the target stent corresponding to each preset specification.
The standard data set creating module 200 is a method for creating the standard data set by acquiring images of stents with different preset specifications from the images of the stents acquired by the image acquiring module, the method for acquiring images of stents with different preset specifications by the standard data set creating module, and the neural network model trained by the model training and storing module 300, which have already been described in the stent specification identification method part in this embodiment, and are not described herein again.
In addition, preferably, the stent specification identifying device further includes: a trustworthiness score threshold setting module 500, the trustworthiness score threshold setting module 500 for setting a trustworthiness score threshold. Correspondingly, the stent specification identification module 400 is further configured to, after obtaining score values corresponding to the preset specifications of the target stent through evaluation, determine that the score values exceed a trusted score threshold, and if yes, obtain and output tags corresponding to the preset specifications.
Similarly, when the neural network model is trained, certain requirements are required for the number of images, and the number of the images is generally required to be at least 2000-3000. Therefore, in this embodiment, preferably, the support specification identification device further includes: a data amplification module 600, wherein the data amplification module 600 is configured to determine whether the data volume of the standard data set is sufficient, and if not, perform data amplification on the standard data set. The method for performing data amplification comprises one or more of brightness amplification, rotation amplification and flip amplification.
In addition, in this embodiment, the stent specification identifying device may further include: an image resizing module 700, configured to resize the image of the target stent to the selected size, so that when the image size of the target stent is not consistent with the selected size, the image size of the target stent is resized.
For convenience of description, the above support specification recognition apparatus is described by dividing functions into various modules and describing the modules respectively. Of course, the functionality of the various modules may be implemented in the same one or more software and/or hardware implementations of the invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solution of the present invention may be embodied in the form of a computer program, which may be stored in a readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., or a part thereof that contributes to the prior art. Therefore, the embodiment of the present invention further provides a readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the stent specification identification method according to the embodiment or some parts of the embodiment of the present invention is provided.
In summary, the readable storage medium, the stent specification identification method and the stent specification identification device provided by the invention detect the specification of the stent in a full-automatic method without manual intervention, and solve the problems of high false detection rate and low efficiency when the specification of the stent is manually checked.
The above description is only for the purpose of describing the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention, and any variations and modifications made by those skilled in the art based on the above disclosure are within the scope of the appended claims.
Claims (16)
1. A stent specification identification method is characterized by comprising the following steps:
acquiring images of stents with different preset specifications to manufacture a standard data set, and training a neural network model by using the standard data set;
and acquiring an image of a target stent, loading the trained neural network model, and identifying the image of the target stent so as to evaluate the score value of the target stent corresponding to each preset specification.
2. The stent specification identification method according to claim 1, further comprising:
setting a trustworthiness score threshold; and the number of the first and second groups,
after the score values of the target bracket corresponding to the preset specifications are obtained through evaluation, whether the score values exceed a credible score threshold value is judged;
if yes, obtaining and outputting the label corresponding to the preset specification.
3. The stent size identification method according to claim 1 or 2, wherein the sum of the point values of the target stent corresponding to the respective preset sizes is 1.
4. The stent specification identification method according to claim 1, wherein the method of acquiring images of stents of different preset specifications comprises:
and judging the size of the stent in the acquired image according to the target characteristics of the stent, if the size is the preset size, reserving the stent, and if the size is not the preset size, discarding the stent.
5. The stent specification identification method according to claim 4, wherein the target features comprise the number of the nodal rods of the stent and/or the bending angle of the connecting member for connecting two adjacent nodal rods.
6. The stent specification identification method of claim 1, wherein the method of producing a standard data set comprises:
and adjusting the sizes of the images of the supports with different preset specifications to a selected size, a selected name and a selected format, unifying the images and classifying the images according to the specifications.
7. The stent specification identifying method according to claim 6, wherein before identifying the image of the target stent, the stent specification identifying method further comprises:
adjusting the image of the target stent to the selected size.
8. The stent specification identification method of claim 6, wherein prior to training the neural network model with the normative dataset, the method of training a neural network model further comprises:
and judging whether the data volume of the standard data set is larger than a set value or not, and if not, performing data amplification on the standard data set.
9. A stent specification identification device, comprising:
the image acquisition module is used for acquiring images of the stent, including an image of the stent for making a standard data set and an image of a target stent;
the standard data set manufacturing module is used for acquiring images of the stents with different preset specifications from the images of the stents acquired by the image acquisition module so as to manufacture the standard data set;
the model training and storing module is used for training a neural network model by utilizing the standard data set;
and the support specification identification module is used for loading the trained neural network model to identify the image of the target support so as to evaluate the score value of the target support corresponding to each preset specification.
10. The stent specification identifying device of claim 9, further comprising: a trustworthiness score threshold setting module for setting a trustworthiness score threshold;
the support specification identification module is further used for judging that the score value exceeds a credible score threshold value after the score value of the target support corresponding to each preset specification is obtained through evaluation, and if the score value exceeds the credible score threshold value, obtaining and outputting the labels corresponding to the preset specifications.
11. The stent size identifying device according to claim 9 or 10, wherein the sum of the point values of the target stent corresponding to the respective preset sizes is 1.
12. The stent specification identification device of claim 9 wherein the method of the standard dataset production module acquiring images of stents of different pre-set specifications comprises:
and judging the size of the stent in the acquired image according to the target characteristics of the stent, if the size is the preset size, reserving the stent, and if the size is not the preset size, discarding the stent.
13. The stent specification identification device according to claim 12, wherein the target feature comprises the number of the nodes of the stent and/or the bending angle of the connecting member for connecting two adjacent nodes.
14. The stent specification identification device of claim 9 wherein the method of making a standard data set by the standard data set making module comprises:
and adjusting the sizes of the images of the supports with different preset specifications to a selected size, a selected name and a selected format, unifying the images and classifying the images according to the specifications.
15. The stent specification identifying device of claim 14, further comprising: an image resizing module to resize the image of the target stent to the selected size.
16. A readable storage medium, in which a computer program is stored, which, when executed by a processor, implements the stent specification identification method according to any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010430500.4A CN113705591A (en) | 2020-05-20 | 2020-05-20 | Readable storage medium, and support specification identification method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010430500.4A CN113705591A (en) | 2020-05-20 | 2020-05-20 | Readable storage medium, and support specification identification method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113705591A true CN113705591A (en) | 2021-11-26 |
Family
ID=78645589
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010430500.4A Pending CN113705591A (en) | 2020-05-20 | 2020-05-20 | Readable storage medium, and support specification identification method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113705591A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117272839A (en) * | 2023-11-20 | 2023-12-22 | 北京阿迈特医疗器械有限公司 | Support press-holding performance prediction method and device based on neural network |
WO2024120857A1 (en) * | 2022-12-07 | 2024-06-13 | Biotronik Ag | Ai-based stent inspection |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015044984A1 (en) * | 2013-09-27 | 2015-04-02 | テルモ株式会社 | Image diagnostic device and control method therefor |
CN106780495A (en) * | 2017-02-15 | 2017-05-31 | 深圳市中科微光医疗器械技术有限公司 | Cardiovascular implantation support automatic detection and appraisal procedure and system based on OCT |
CN108921206A (en) * | 2018-06-15 | 2018-11-30 | 北京金山云网络技术有限公司 | A kind of image classification method, device, electronic equipment and storage medium |
US20190156205A1 (en) * | 2017-11-20 | 2019-05-23 | Koninklijke Philips N.V. | Training first and second neural network models |
CN109977943A (en) * | 2019-02-14 | 2019-07-05 | 平安科技(深圳)有限公司 | A kind of images steganalysis method, system and storage medium based on YOLO |
CN110110131A (en) * | 2019-05-23 | 2019-08-09 | 北京航空航天大学 | It is a kind of based on the aircraft cable support of deep learning and binocular stereo vision identification and parameter acquiring method |
CN110664524A (en) * | 2018-10-08 | 2020-01-10 | 北京昆仑医云科技有限公司 | Devices, systems, and media for guiding stent implantation in a vessel |
CN110717534A (en) * | 2019-09-30 | 2020-01-21 | 中国科学院大学 | Target classification and positioning method based on network supervision |
-
2020
- 2020-05-20 CN CN202010430500.4A patent/CN113705591A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015044984A1 (en) * | 2013-09-27 | 2015-04-02 | テルモ株式会社 | Image diagnostic device and control method therefor |
CN106780495A (en) * | 2017-02-15 | 2017-05-31 | 深圳市中科微光医疗器械技术有限公司 | Cardiovascular implantation support automatic detection and appraisal procedure and system based on OCT |
US20190156205A1 (en) * | 2017-11-20 | 2019-05-23 | Koninklijke Philips N.V. | Training first and second neural network models |
CN108921206A (en) * | 2018-06-15 | 2018-11-30 | 北京金山云网络技术有限公司 | A kind of image classification method, device, electronic equipment and storage medium |
CN110664524A (en) * | 2018-10-08 | 2020-01-10 | 北京昆仑医云科技有限公司 | Devices, systems, and media for guiding stent implantation in a vessel |
CN109977943A (en) * | 2019-02-14 | 2019-07-05 | 平安科技(深圳)有限公司 | A kind of images steganalysis method, system and storage medium based on YOLO |
CN110110131A (en) * | 2019-05-23 | 2019-08-09 | 北京航空航天大学 | It is a kind of based on the aircraft cable support of deep learning and binocular stereo vision identification and parameter acquiring method |
CN110717534A (en) * | 2019-09-30 | 2020-01-21 | 中国科学院大学 | Target classification and positioning method based on network supervision |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2024120857A1 (en) * | 2022-12-07 | 2024-06-13 | Biotronik Ag | Ai-based stent inspection |
CN117272839A (en) * | 2023-11-20 | 2023-12-22 | 北京阿迈特医疗器械有限公司 | Support press-holding performance prediction method and device based on neural network |
CN117272839B (en) * | 2023-11-20 | 2024-02-06 | 北京阿迈特医疗器械有限公司 | Support press-holding performance prediction method and device based on neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11790645B2 (en) | Diagnosis assistance system and control method thereof | |
EP3785603B1 (en) | Machine learning-based fundus image detection method, apparatus, and system | |
US11501428B2 (en) | Method, apparatus and system for detecting fundus image based on machine learning | |
CN111310841B (en) | Medical image classification method, medical image classification device, medical image classification apparatus, medical image classification computer device, and medical image classification storage medium | |
CN110298844B (en) | X-ray radiography image blood vessel segmentation and identification method and device | |
JP4026158B1 (en) | Interpretation support device, interpretation support method and program | |
WO2018104819A1 (en) | A method and system for evaluating quality of semen sample | |
CN110706233A (en) | Retina fundus image segmentation method and device | |
CN113705591A (en) | Readable storage medium, and support specification identification method and device | |
CN111178420B (en) | Coronary artery segment marking method and system on two-dimensional contrast image | |
CN110619332B (en) | Data processing method, device and equipment based on visual field inspection report | |
CN112884729A (en) | Auxiliary diagnosis method and device for fundus diseases based on bimodal deep learning | |
CN110310280A (en) | Hepatic duct and the image-recognizing method of calculus, system, equipment and storage medium | |
CN112819797A (en) | Diabetic retinopathy analysis method, device, system and storage medium | |
CN115578783B (en) | Device and method for identifying eye diseases based on eye images and related products | |
CN113222913A (en) | Circuit board defect detection positioning method and device and storage medium | |
CN115294063A (en) | Method, device, system, electronic device and medium for detecting defect of electronic component | |
CN112869697A (en) | Judgment method for simultaneously identifying stage and pathological change characteristics of diabetic retinopathy | |
CN114429468A (en) | Bone age measuring method, bone age measuring system, electronic device and computer-readable storage medium | |
CN117352164A (en) | Multi-mode tumor detection and diagnosis platform based on artificial intelligence and processing method thereof | |
EP3907696A1 (en) | Method and system for identifying abnormal images in a set of medical images | |
CN113361482A (en) | Nuclear cataract identification method, device, electronic device and storage medium | |
CN113591601B (en) | Method and device for identifying hyphae in cornea confocal image | |
WO2019171398A1 (en) | A fundus image analysis system | |
CN114782452A (en) | Processing system and device of fluorescein fundus angiographic image |
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 |