CN116309924A - Model training method, image display method and device - Google Patents
Model training method, image display method and device Download PDFInfo
- Publication number
- CN116309924A CN116309924A CN202310598313.0A CN202310598313A CN116309924A CN 116309924 A CN116309924 A CN 116309924A CN 202310598313 A CN202310598313 A CN 202310598313A CN 116309924 A CN116309924 A CN 116309924A
- Authority
- CN
- China
- Prior art keywords
- images
- image
- generation model
- image generation
- trained
- 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
- 238000000034 method Methods 0.000 title claims abstract description 105
- 238000012549 training Methods 0.000 title claims abstract description 57
- 238000005457 optimization Methods 0.000 claims abstract description 12
- 238000002591 computed tomography Methods 0.000 claims description 32
- 239000011159 matrix material Substances 0.000 claims description 28
- 238000005070 sampling Methods 0.000 claims description 24
- 238000003860 storage Methods 0.000 claims description 18
- 238000005520 cutting process Methods 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 15
- 230000004927 fusion Effects 0.000 claims description 10
- 230000005855 radiation Effects 0.000 abstract description 8
- 230000008569 process Effects 0.000 description 20
- 230000006870 function Effects 0.000 description 15
- 238000010586 diagram Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 10
- 230000006872 improvement Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 239000003814 drug Substances 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 210000004185 liver Anatomy 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000010979 ruby Substances 0.000 description 1
- 229910001750 ruby Inorganic materials 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2211/00—Image generation
- G06T2211/40—Computed tomography
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The specification discloses a model training method, an image display method and an image display device. In the embodiment of the present disclosure, during the training of the image generation model, a first number of first CT images are used as input of the image generation model, and the image generation model is trained with a deviation between a second number of generated images output by the minimized image generation model and a second number of second CT images as an optimization target, so that the trained image generation model can generate a large number of second CT images according to the input small number of first CT images, and then determine CT images with higher resolution according to the generated large number of second CT images, thereby ensuring that a patient does not receive more radiation doses, reducing scanning costs, and simultaneously ensuring that CT images with higher resolution are obtained.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method for model training, a method for image display, and an apparatus for image display.
Background
The electronic computer tomography (Computed Tomography, CT) scanning technology is one of the conventional inspection means of modern medicine, and has the advantages of clear development, high resolution and the like. However, three-dimensional CT scanning is performed layer by layer in the z-axis of space, and if a CT image with high z-axis resolution is desired, a longer time is required for scanning, resulting in a patient receiving more radiation dose and also more scanning costs.
Therefore, how to ensure that the patient does not receive more radiation dose and also ensure that the acquired three-dimensional CT image with high z-axis resolution is a technical problem to be solved.
Disclosure of Invention
The present disclosure provides a method for model training, a method for image display, and a device thereof, so as to partially solve the above-mentioned problems in the prior art.
The technical scheme adopted in the specification is as follows:
the present specification provides a method of model training, comprising:
acquiring a first number of first CT images and a second number of second CT images, wherein the first CT images and the second CT images are CT images obtained by aiming at the same body part of the same patient according to different scanning modes, and the second number is larger than the first number;
inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images as generated images through the image generation model;
the image generation model is trained with a minimum deviation between the second number of generated images and the second number of second CT images as an optimization objective.
Optionally, acquiring the first number of first electronic computer tomography CT images and the second number of second electronic computer tomography CT images specifically includes:
acquiring a second number of second CT images;
and downsampling the second number of second CT images according to a preset sampling rate to obtain the first number of first CT images, wherein the sampling rate corresponds to the first number.
Optionally, the image generation model includes: a feature extraction network and a position coding network;
before inputting the first number of first CT images into the image generation model to be trained, the method further comprises:
for each first CT image, determining the image coordinates of the first CT image under a preset image coordinate system as the corresponding image coordinates of the first CT image;
determining an image position matrix corresponding to the first CT image according to the image coordinates corresponding to the first CT image and the sampling rate;
inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images through the image generation model, wherein the method specifically comprises the following steps of:
Inputting the first CT images of the first quantity and the image position matrixes corresponding to the first CT images into the image generation model so that the image generation model can input the first CT images into the feature extraction network for each first CT image to obtain the image features corresponding to the first CT images, inputting the image position matrixes corresponding to the first CT images into the position coding network to obtain the position codes corresponding to the first CT images, and carrying out feature fusion on the position codes corresponding to the first CT images and the image features corresponding to the first CT images to obtain the fused features corresponding to the first CT images;
and generating the second number of CT images as generated images according to the fused features corresponding to each first CT image through the image generation model.
Optionally, before inputting the first number of first CT images into the image generation model to be trained, the method further comprises:
determining a cutting proportion;
cutting the first CT images with the first quantity according to the determined cutting proportion to obtain first CT images with the first quantity after cutting;
Inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images through the image generation model, wherein the method specifically comprises the following steps of:
inputting the first number of cut first CT images into the image generation model to generate the second number of CT images as generated images through the image generation model.
Optionally, determining the clipping proportion specifically includes:
reading device information of a device executing a training task of the image generation model;
and determining the clipping proportion according to the equipment information.
Optionally, the first CT image includes: a ThickCT image, the second CT image comprising: thinCT images.
The specification provides a method for displaying an image, comprising the following steps:
acquiring a first number of CT images scanned by an electronic computer tomography CT device for scanning a patient;
inputting the first number of CT images into a pre-trained image generation model so that the image generation model generates a second number of CT images according to the first number of CT images, wherein the first number is smaller than the second number, and the image generation model is obtained by training according to the model training method;
Determining CT images to be displayed according to the second number of CT images;
and displaying the CT image to be displayed.
The present specification provides an apparatus for model training, comprising:
the acquisition module is used for acquiring a first number of first CT images and a second number of second CT images, wherein the first CT images and the second CT images are CT images obtained by aiming at the same body part of the same patient according to different scanning modes, and the second number is larger than the first number;
the input module is used for inputting the first number of the CT images into an image generation model to be trained so as to generate the second number of the CT images as generated images through the image generation model;
and the training module is used for training the image generation model by taking the deviation between the minimum second number of generated images and the minimum second number of second CT images as an optimization target.
Optionally, the acquiring module is specifically configured to acquire a second number of second CT images; and downsampling the second number of second CT images according to a preset sampling rate to obtain the first number of first CT images, wherein the sampling rate corresponds to the first number.
Optionally, the image generation model includes: a feature extraction network and a position coding network;
the input module is further used for determining, for each first CT image, an image coordinate of the first CT image under a preset image coordinate system as an image coordinate corresponding to the first CT image before inputting the first number of first CT images into the image generation model to be trained; determining an image position matrix corresponding to the first CT image according to the image coordinates corresponding to the first CT image and the sampling rate;
the input module is specifically configured to input the first number of first CT images and an image position matrix corresponding to each first CT image into the image generation model, so that the image generation model performs feature fusion on the first CT images and the image features corresponding to the first CT images, and input the first CT images into the feature extraction network to obtain image features corresponding to the first CT images, and input the image position matrix corresponding to the first CT images into the position coding network to obtain position codes corresponding to the first CT images, and perform feature fusion on the position codes corresponding to the first CT images and the image features corresponding to the first CT images to obtain fused features corresponding to the first CT images; and generating the second number of CT images as generated images according to the fused features corresponding to each first CT image through the image generation model.
Optionally, the input module is further configured to determine a cropping ratio before inputting the first number of first CT images into an image generation model to be trained; cutting the first CT images with the first quantity according to the determined cutting proportion to obtain first CT images with the first quantity after cutting;
the input module is specifically configured to input the first number of cropped first CT images into the image generation model, so as to generate, through the image generation model, the second number of CT images as generated images.
Optionally, the input module is specifically configured to read device information of a device that performs a training task of the image generation model; and determining the clipping proportion according to the equipment information.
Optionally, the first CT image includes: a ThickCT image, the second CT image comprising: thinCT images.
The present specification provides an apparatus for displaying an image, comprising:
an acquisition module for acquiring a first number of CT images scanned by a patient scanned by an CT apparatus;
the input module is used for inputting the first number of CT images into a pre-trained image generation model so that the image generation model generates a second number of CT images according to the first number of CT images, the first number is smaller than the second number, and the image generation model is trained according to the model training method;
The determining module is used for determining CT images to be displayed according to the second number of CT images;
and the display module is used for displaying the CT image to be displayed.
The present specification provides a computer readable storage medium storing a computer program which when executed by a processor implements the method of model training or the method of image presentation described above.
The present specification provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of model training or the method of image presentation described above when executing the program.
The above-mentioned at least one technical scheme that this specification adopted can reach following beneficial effect:
according to the model training method provided by the specification, as the first quantity of first CT images are used as the input of the image generation model in the training process of the image generation model, and the deviation between the second quantity of generated images output by the image generation model and the second quantity of second CT images is used as an optimization target, the image generation model is trained, so that the trained image generation model can generate a large quantity of second CT images according to the input small quantity of first CT images, and then determine CT images with higher resolution according to the generated large quantity of second CT images, and further, the CT images with higher resolution can be obtained while ensuring that patients do not receive more radiation doses and reducing scanning cost.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification, illustrate and explain the exemplary embodiments of the present specification and their description, are not intended to limit the specification unduly. In the drawings:
FIG. 1 is a flow chart of a method of model training provided in the present specification;
FIG. 2 is a schematic illustration of the relationship between one of the images provided in this specification;
FIG. 3 is a schematic flow chart of model training provided in the present specification;
FIG. 4 is a flow chart of a method of image presentation provided in the present specification;
FIG. 5 is a schematic view of a device structure for model training provided in the present specification;
FIG. 6 is a schematic view of an image display device structure provided in the present specification;
fig. 7 is a schematic structural diagram of the electronic device corresponding to fig. 1 and 4 provided in the present specification.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the present specification more apparent, the technical solutions of the present specification will be clearly and completely described below with reference to specific embodiments of the present specification and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.
The following describes in detail the technical solutions provided by the embodiments of the present specification with reference to the accompanying drawings.
Fig. 1 is a flow chart of a method for model training provided in the present specification, including the following steps:
s101: a first number of first computed tomography CT images and a second number of second computed tomography CT images are acquired, the first CT images and the second CT images being CT images obtained for a same body part of a same patient according to different scanning modes, the second number being greater than the first number.
The execution subject of the model training method in the present specification may be a terminal device such as a desktop computer or a notebook computer, or may be a server, and the model training method in the embodiment of the present specification will be described below by taking the example in which the terminal device is the execution subject.
Today, the technique of computerized tomography (Computed Tomography, CT) scanning is one of the conventional examination means of modern medicine, the CT scanning mentioned above being scanning the patient's body layer by layer through the spatial z-axis to obtain three-dimensional CT images, which can then be used for performing the examination of the modern medicine mentioned above.
However, if a CT image with high z-axis resolution is obtained through the existing CT scanning process, a longer time of scanning is required for the patient's body, and accordingly, the longer the scanning time is, the larger the radiation dose received by the patient will be, and at the same time, the larger the scanning cost will be, so that the existing CT scanning process cannot ensure that the patient does not receive more radiation dose, and the scanning cost is reduced, and at the same time, the CT image with high z-axis resolution is ensured.
The spatial z-axis mentioned above may be set to other axial directions such as the spatial x-axis and the spatial y-axis according to the change of the body direction of the patient, and is not particularly limited in this specification.
In an implementation of the present specification, the first CT image acquired by the terminal device may include: the Thick CT image, i.e., the Thick scan CT image, the second CT image may include: thin CT images, i.e., thin scan CT images.
The first CT image and the second CT image may be CT images obtained by the electronic computed tomography CT apparatus according to different scanning methods for the same body part of the same patient, where the scanning method corresponding to the first CT image is a thick scan CT scanning method and the scanning method corresponding to the second CT image is a thin scan CT scanning method.
The thick scan CT scanning method refers to that when the patient body is scanned layer by layer through the spatial z-axis during the CT scanning process, the number of layers of the scanned spatial z-axis is smaller, so that the number of corresponding obtained first CT images is also smaller, and then the resolution of the image displayed in the z-axis direction in the CT scanning result determined by overlapping the corresponding first CT images of each layer is lower.
In contrast, in the thin scan CT scanning, when the patient body is scanned layer by layer through the spatial z-axis in the CT scanning process, the number of layers of the scanned spatial z-axis is greater, so that the number of corresponding obtained second CT images is also greater, and then the resolution of the image displayed in the z-axis direction in the CT scanning result determined by overlapping the second CT images corresponding to the layers is higher.
Fig. 2 is a schematic diagram of the relationship between one image provided in the present specification.
As shown in fig. 2, the left half of fig. 2 includes 3 first CT images obtained by the thick scan CT scanning method, and the right half includes 11 second CT images obtained by the thin scan CT scanning method.
Of course, the first number of first CT images acquired by the terminal device may also be determined based on the second number of second CT images.
Specifically, the terminal device may first acquire a second number of second CT images, then determine the number of first CT images according to a preset sampling rate, and then downsample the second number of second CT images to obtain a first number of first CT images, where the sampling rate and the first number are in a one-to-one correspondence.
The specific way of sampling from the second number of second CT images may be directly selecting the first number of second CT images from the second number of second CT images as the first CT images, or determining the first number of first CT images from the second number of second CT images by interpolation such as a neighboring interpolation method. The first CT image determined by the interpolation method may be an image in the second number of second CT images, or may be a transition image between images in the second number of second CT images.
For example, the terminal device acquires 3 first CT images and 11 second CT images, where the 3 first CT images may be 3 second CT images randomly extracted from the 11 second CT images, or the 3 first CT images may be 3 transition images generated based on the 11 second CT images by a proximity interpolation method.
It should be noted that, due to the difference in performance parameters between different CT apparatuses in the practical application of the image generation model, the number of the first CT images acquired by the different CT apparatuses may also be different, that is, the resolution of the CT images acquired by the different CT apparatuses in the thick scan mode may also be different.
In the training process of the image generation model, the preset sampling rate may be actually a plurality of sampling rates, so that when the image generation model is trained, the image generation model may be trained by using the first number of first CT images determined according to different sampling rates as sample data, so that no matter what resolution of a plurality of first CT images are input into the finally trained image generation model, the finally trained image generation model can accurately determine CT images with higher resolution.
Of course, the first number of first CT images acquired by the terminal device may be independent from the second number of second CT images, without any dependency.
Continuing with the above example, the 3 first CT images may be obtained by scanning the liver of a patient by means of a thin scan CT scan using an electronic computed tomography CT apparatus, obtaining 11 second CT images, and then continuing to scan the liver of the same patient by means of a thick scan CT scan using an electronic computed tomography CT apparatus.
S102: inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images as generated images through the image generation model.
In practical application of the image generation model, because the computing power performance of different terminal devices is different, in order to enable the finally trained image generation model to be applicable to any terminal device with any computing power performance, during the training process of the image generation model, the first number of first CT images can be cut before being input into the image generation model to be trained, and the image generation model can be trained by using the cut first number of first CT images, so that no matter what size of first number of first CT images are input into the finally trained image generation model, the finally trained image generation model can accurately determine the second number of second CT images corresponding to the input first number of first CT images.
Specifically, the device information for executing the training task of the image generation model may be read first, the clipping ratio corresponding to the first number of first CT images may be determined according to the device information, and then the first number of first CT images may be clipped according to the determined clipping ratio, so as to obtain the first number of clipped first CT images.
Thereafter, the first number of cropped first CT images may be redetermined to a first number of first CT images, and the first number of cropped first CT images may be input into the image generation model to generate the second number of CT images as generated images from the image generation model.
It can be seen from this that, the clipping ratio is determined by the read device information, so that the data that the device can effectively process can be determined according to the actual situation of the device, it can be understood that if the device information determines that the device has higher performance, the size of the image that the device can actually process is also larger, the size of the clipped first CT image obtained by the determined clipping ratio is closer to the original first CT image, and if the device information determines that the device has worse performance, the size of the image that the device can actually process is also smaller, and the size of the clipped first CT image obtained by the determined clipping ratio is smaller than the size of the original first CT image. Among them, the device information mentioned here may refer to parameters of a central processing unit (Central Processing Unit, CPU), a graphics processor (Graphics Processing Unit, GPU), a data processing unit such as a memory, and a storage unit of the device.
In a specific model training process, the second CT image and the image data corresponding to the first CT image at each preset sampling rate may be normalized first, specifically, normalization may be performed by using (0, 1), that is, gray values of the image data are mapped into [0,1] uniformly, and data normalization may also be performed, that is, an image coordinate system of the image data is unified to be an LAS coordinate system.
Then, the image generating model determines, for each first CT image, an image coordinate of the first CT image under a preset image coordinate system as an image coordinate corresponding to the first CT image, and then, may determine, according to the image coordinate corresponding to the first CT image and the sampling rate, an image position matrix corresponding to the first CT image.
Wherein the image coordinates corresponding to the first CT image can be used as%
) To indicate, here
Refers to the starting point coordinates of the first CT image on the x-axis, < >>
Refers to the starting point coordinates of the first CT image on the y-axis.
The image position matrix M corresponding to the first CT image can be used
) Expressed here +.>
Refers to the above mentioned samplingThe sampling rate may be an integer or a decimal representation, for example, s is 3 at this time, assuming that the first number is a and the second number is 3 a.
The terminal device may then input the first number of first CT images and the image location matrix corresponding to each first CT image into the image generation model to obtain a location code corresponding to the first CT image and an image feature corresponding to the first CT image.
Wherein the position code P corresponding to the first CT image comprises B,
Characteristic data of several dimensions H, W, B representing a first number corresponding to a first CT image input into the image generation model, and +.>
The number of channels used to represent the position code P, H is the length of the cross-sectional image of the first CT image, and W is the width of the cross-sectional image of the first CT image. This procedure can be understood as an image position matrix M (++1) with a channel number of 1>
) As input, get +.>
The position of the number of channels codes P, dimension (B, -/-A)>
、H、W）。
The image features corresponding to the first CT image comprise B,
H. Characteristic data of several dimensions W, wherein B represents a first number of first CT images input into the image generation model, +.>
Image features for representing extraction from a first CT imageThe number of characteristic channels of feature F, H is the length of the cross-sectional image of the first CT image, and W is the width of the cross-sectional image of the first CT image.
In the present specification, the image generation model includes a feature extraction network and a position coding network, so when a first number of first CT images and an image position matrix corresponding to each first CT image are input into the image generation model, the image position matrix corresponding to each first CT image is actually input into the position coding network to obtain a position code corresponding to each first CT image, and the first number of first CT images are input into the feature extraction network to obtain an image feature corresponding to each first CT image.
Then, for each first CT image, the image generation model may perform feature fusion on the position code corresponding to the first CT image and the image feature corresponding to the first CT image to obtain a fused feature corresponding to the first CT image, and then, the image generation model generates the second number of CT images as generated images according to the fused feature corresponding to each first CT image.
And generating the second number of CT images according to the fused features corresponding to each first CT image, and when the second number of CT images are used as generated images, the image generating model may specifically determine the fused features corresponding to all the first CT images according to the fused features corresponding to each first CT image. For example, the fused features corresponding to each first CT image may be weighted, the weighted result is determined as the fused features corresponding to all the first CT images, and then the second number of CT images are generated as the generated images according to the fused features corresponding to all the first CT images.
Fig. 3 is a schematic flow chart of model training provided in the present specification.
As shown in fig. 3, once the terminal device acquires the first number of first CT images and the second number of second CT images, device information of a device performing a training task of the image generation model may be first read, a clipping ratio corresponding to the first number of first CT images is determined according to the device information, and the first number of first CT images is clipped according to the clipping ratio, so as to obtain the first number of clipped first CT images, which is "position slicing" in fig. 3.
The terminal device may then redetermine the first number of cropped first CT images to a first number of first CT images and input the first number of cropped first CT images into the image generation model.
Subsequently, for each first CT image, an image coordinate corresponding to the first CT image may be determined, and according to the image coordinate and the sampling rate, an image position matrix corresponding to the first CT image may be determined. Then, the position coding network in the image generation model can determine the position coding corresponding to the first CT image according to the image position matrix, and the feature extraction network in the image generation model can extract the image features corresponding to the first CT image.
Thereafter, the image generation model may perform feature fusion on the position code corresponding to the first CT image and the image feature corresponding to the first CT image to obtain a fused feature corresponding to the first CT image, and generate, according to the fused feature corresponding to each first CT image, the second number of CT images as generated images, that is, the second number of generated images in fig. 3.
In the present specification, in the position encoding network, the image position matrix M of the first CT image may be normalized first to obtain a normalized image position matrix
Wherein->
For representing the function used for normalization.
Then, through standardized image position matrix
An initial position code P of the first CT image is obtained, wherein the first CT image can be obtained byInitial position coding P of CT image:
P =
wherein->
。
It is understood that there is one
Is to find the normalized picture position matrix from the coding matrix a>
And obtaining initial position codes of the first CT image by using a part of the closest matrix data.
Further, the initial position code P of the first CT image is transposed and copied to obtain the channel number of 
Size of +.>
Matrix of->
Matrix->
I.e. the position coding of the obtained first CT image.
Accordingly, the feature extraction network may utilize basic network elements (e.g., a mathematical computation unit such as a convolution layer or a transform that may be back-propagated and optimized) to spatially characterize the cross-sectional image of the first CT image by the number of channels
Is characterized by image features F of the first CT image of (a). In the present specification, the image generation model further includes a super-resolution reconstruction network, so that the image feature F and the first CT image are obtainedPosition coding matrix of CT image>
After that, the image feature F of the first CT image can be assigned to the position-coding matrix corresponding to the first CT image>
Channel splicing is carried out, and the channel splicing is used as the input of the final super-resolution reconstruction network to finally obtain +.>
The super-resolution reconstructed image corresponding to the layer, i.e. the generated image mentioned above.
S103: the image generation model is trained with a minimum deviation between the second number of generated images and the second number of second CT images as an optimization objective.
In order to make the second number of generated images output by the trained image generation model closest to the second number of second CT images, in the model training process, the deviation between the second number of generated images and the second number of second CT images can be minimized as an optimization target, and the image generation model is trained.
Specifically, the loss function corresponding to the model training process may be determined first:
wherein if the feedforward process of the feature extraction network is a function
It is shown that the feed-forward procedure of the above-mentioned super-resolution reconstruction network can be used as a function +.>
Expressed, then the optimization objective of the training of the entire image generation model can be expressed by the following formula:
wherein the method comprises the steps of
，/>
For representing a first CT image,/or->
For representing a second number of second CT images, and (2)>
Representing feature concatenation->
As a loss function.
And then, representing the deviation between the second number of generated images and the second number of second CT images by using the loss value corresponding to the loss function, and training the image generation model by taking the loss value corresponding to the minimized loss function as an optimization target.
After the image generation model is trained, the image generation model may be tested using a test set, and a quantization index peak signal-to-noise ratio (PSNR) and Structural Similarity (SSIM) are calculated from the test results, and a training effect of the image generation model is determined from the quantization index peak signal-to-noise ratio and structural similarity.
Of course, in practical applications, in addition to the image feature of the first CT image and the position code of the first CT image being spliced to obtain the fused feature, the fused feature may be obtained in other manners, for example, by an affine transformation manner, according to the image feature of the first CT image and the position code of the first CT image, to obtain the fused feature corresponding to the first CT image.
The foregoing describes a method for model training of an image generation model, and in the following, how an image generated by the image generation model is presented after the image generation model is trained.
Fig. 4 is a flow chart of a method for displaying an image provided in the present specification.
S401: a first number of CT images are acquired from a patient scanned by an electronic computed tomography CT apparatus.
S402: inputting the first number of CT images into a pre-trained image generation model so that the image generation model generates a second number of CT images according to the first number of CT images, wherein the first number is smaller than the second number, and the image generation model is obtained by training according to the model training method.
S403: and determining CT images to be displayed according to the second number of CT images.
S404: and displaying the CT image to be displayed.
The display platform for displaying the CT image to be displayed may be acted on by the CT apparatus, or may be acted on by the terminal apparatus after the CT apparatus is connected to the terminal apparatus, after the terminal apparatus receives the CT image to be displayed, which is not specifically limited in the present specification.
In practical application, when a patient needs to take CT images, a thick scanning mode can be adopted, a first number of CT images are acquired through a CT device, the first number of CT images are input into a pre-trained image generation model, a second number of CT images with larger numbers are obtained, and when the second number of CT images are used as CT images to be displayed for superposition display, the fact that the patient only needs to receive low radiation dose is achieved, and CT images with higher resolution can be obtained.
In summary, it can be seen from the above method that, in the training process of the image generation model, the first number of first CT images are used as input of the image generation model, and the deviation between the second number of generated images output by the minimized image generation model and the second number of second CT images is used as an optimization target, so that the trained image generation model can generate a large number of second CT images according to the input small number of first CT images, and then determine a CT image with higher resolution (i.e., a CT image with high z-axis resolution) according to the generated large number of second CT images, so that it is possible to ensure that the patient does not receive more radiation dose, reduce scanning cost, and obtain a CT image with higher resolution.
In addition, the image generation model trained by the model training method provided by the specification can receive CT images with multiple resolutions, and is not limited to receiving CT images with one resolution, so that the image generation model can be applicable to multiple scenes and multiple CT devices, and the use cost is greatly reduced.
The foregoing is a method of one or more implementations of the present specification, and based on the same concept, the present specification further provides a corresponding apparatus for model training, as shown in fig. 5, and the present specification further provides a corresponding apparatus for image presentation, as shown in fig. 6.
Fig. 5 is a schematic diagram of a device structure for model training provided in the present specification, including:
an acquisition module 501, configured to acquire a first number of first CT images and a second number of second CT images, where the first CT images and the second CT images are CT images obtained by different scanning modes for a same body part of a same patient, and the second number is greater than the first number;
an input module 502, configured to input the first number of CT images into an image generation model to be trained, so as to generate, by using the image generation model, the second number of CT images as generated images;
A training module 503, configured to train the image generation model with a deviation between the second number of generated images and the second number of second CT images being minimized as an optimization target.
Optionally, the acquiring module 401 is specifically configured to acquire a second number of second CT images; and downsampling the second number of second CT images according to a preset sampling rate to obtain the first number of first CT images, wherein the sampling rate corresponds to the first number.
Optionally, the image generation model includes: a feature extraction network and a position coding network;
the apparatus further comprises:
a position determining module 504, configured to determine, for each first CT image, an image coordinate of the first CT image under a preset image coordinate system as an image coordinate corresponding to the first CT image, before inputting the first number of first CT images into an image generation model to be trained; and determining an image position matrix corresponding to the first CT image according to the image coordinates corresponding to the first CT image and the sampling rate.
The input module 502 is specifically configured to input the first number of first CT images and an image position matrix corresponding to each first CT image into the image generation model, so that the image generation model performs feature fusion on the first CT images and the image features corresponding to the first CT images, input the first CT images into the feature extraction network to obtain image features corresponding to the first CT images, and input the image position matrix corresponding to the first CT images into the position coding network to obtain position codes corresponding to the first CT images, and perform feature fusion on the position codes corresponding to the first CT images and the image features corresponding to the first CT images to obtain fused features corresponding to the first CT images;
And generating the second number of CT images as generated images according to the fused features corresponding to each first CT image through the image generation model.
Optionally, the apparatus further comprises:
a clipping module 505 for determining a clipping ratio; cutting the first CT images with the first quantity according to the determined cutting proportion to obtain first CT images with the first quantity after cutting;
optionally, the input module 402 is specifically configured to input the first number of cropped first CT images into the image generation model, so as to generate, as the generated image, the second number of CT images through the image generation model.
Optionally, the cropping module 505 is specifically configured to read device information of a device that performs the training task of the image generation model; and determining the clipping proportion according to the equipment information.
Optionally, the first CT image includes: a Thick CT image, the second CT image comprising: thin CT images.
Fig. 6 is a schematic diagram of a device structure for displaying images provided in the present specification, including:
an acquisition module 601 for acquiring a first number of CT images scanned by a patient scanned by an electronic computed tomography CT apparatus;
The input module 602 is configured to input the first number of CT images into a pre-trained image generation model, so that the image generation model generates a second number of CT images according to the first number of CT images, where the first number is smaller than the second number, and the image generation model is obtained by training according to the model training method described above;
a determining module 603, configured to determine a CT image to be displayed according to the second number of CT images;
and the display module 604 is used for displaying the CT image to be displayed.
The present specification also provides a computer readable storage medium storing a computer program operable to perform a method of model training as provided in fig. 1 or a method of image presentation as provided in fig. 4, as described above.
The present specification also provides a schematic structural diagram of an electronic device corresponding to fig. 1 or 4 shown in fig. 7. At the hardware level, as shown in fig. 7, the electronic device includes a processor, an internal bus, a network interface, a memory, and a nonvolatile storage, and may of course include hardware required by other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the method of model training described above with respect to fig. 1 or the method of image presentation described above with respect to fig. 4.
Of course, other implementations, such as logic devices or combinations of hardware and software, are not excluded from the present description, that is, the execution subject of the following processing flows is not limited to each logic unit, but may be hardware or logic devices.
In the 90 s of the 20 th century, improvements to one technology could clearly be distinguished as improvements in hardware (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software (improvements to the process flow). However, with the development of technology, many improvements of the current method flows can be regarded as direct improvements of hardware circuit structures. Designers almost always obtain corresponding hardware circuit structures by programming improved method flows into hardware circuits. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device, PLD) (e.g., field programmable gate array (Field Programmable Gate Array, FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A designer programs to "integrate" a digital system onto a PLD without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, such programming is mostly implemented by using "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before the compiling is also written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), but not just one of the hdds, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of the hardware description languages described above.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable logic controllers, and embedded microcontrollers, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in a pure computer readable program code, it is well possible to implement the same functionality by logically programming the method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Such a controller may thus be regarded as a kind of hardware component, and means for performing various functions included therein may also be regarded as structures within the hardware component. Or even means for achieving the various functions may be regarded as either software modules implementing the methods or structures within hardware components.
The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
For convenience of description, the above devices are described as being functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present specification.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present description is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the specification. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises the element.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.
The foregoing is merely exemplary of the present disclosure and is not intended to limit the disclosure. Various modifications and alterations to this specification will become apparent to those skilled in the art. Any modifications, equivalent substitutions, improvements, or the like, which are within the spirit and principles of the present description, are intended to be included within the scope of the claims of the present description.
Claims (16)
1. A method of model training, comprising:
acquiring a first number of first CT images and a second number of second CT images, wherein the first CT images and the second CT images are CT images obtained by aiming at the same body part of the same patient according to different scanning modes, and the second number is larger than the first number;
inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images as generated images through the image generation model;
the image generation model is trained with a minimum deviation between the second number of generated images and the second number of second CT images as an optimization objective.
2. The method of claim 1, wherein acquiring a first number of first electronic computed tomography CT images and a second number of second electronic computed tomography CT images, in particular comprises:
acquiring a second number of second CT images;
and downsampling the second number of second CT images according to a preset sampling rate to obtain the first number of first CT images.
3. The method of claim 2, wherein the image generation model comprises: a feature extraction network and a position coding network;
before inputting the first number of first CT images into the image generation model to be trained, the method further comprises:
for each first CT image, determining the image coordinates of the first CT image under a preset image coordinate system as the corresponding image coordinates of the first CT image;
determining an image position matrix corresponding to the first CT image according to the image coordinates corresponding to the first CT image and the sampling rate;
inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images through the image generation model, wherein the method specifically comprises the following steps of:
inputting the first CT images of the first quantity and the image position matrixes corresponding to the first CT images into the image generation model so that the image generation model can input the first CT images into the feature extraction network for each first CT image to obtain the image features corresponding to the first CT images, inputting the image position matrixes corresponding to the first CT images into the position coding network to obtain the position codes corresponding to the first CT images, and carrying out feature fusion on the position codes corresponding to the first CT images and the image features corresponding to the first CT images to obtain the fused features corresponding to the first CT images;
And generating the second number of CT images as generated images according to the fused features corresponding to each first CT image through the image generation model.
4. A method according to claim 1 or 3, wherein before inputting the first number of first CT images into an image generation model to be trained, the method further comprises:
determining a cutting proportion;
cutting the first CT images with the first quantity according to the determined cutting proportion to obtain first CT images with the first quantity after cutting;
inputting the first number of the first CT images into an image generation model to be trained so as to generate the second number of the CT images through the image generation model, wherein the method specifically comprises the following steps of:
inputting the first number of cut first CT images into the image generation model to generate the second number of CT images as generated images through the image generation model.
5. The method of claim 4, wherein determining the clipping ratio comprises:
reading device information of a device executing a training task of the image generation model;
And determining the clipping proportion according to the equipment information.
6. The method of claim 1, wherein the first CT image comprises: a ThickCT image, the second CT image comprising: thinCT images.
7. A method of image presentation, comprising:
acquiring a first number of CT images scanned by an electronic computer tomography CT device for scanning a patient;
inputting the first number of CT images into a pre-trained image generation model, so that the image generation model generates a second number of CT images according to the first number of CT images, wherein the first number is smaller than the second number, and the image generation model is obtained by training according to the method of any one of claims 1-6;
determining CT images to be displayed according to the second number of CT images;
and displaying the CT image to be displayed.
8. An apparatus for model training, comprising:
the acquisition module is used for acquiring a first number of first CT images and a second number of second CT images, wherein the first CT images and the second CT images are CT images obtained by aiming at the same body part of the same patient according to different scanning modes, and the second number is larger than the first number;
The input module is used for inputting the first number of the CT images into an image generation model to be trained so as to generate the second number of the CT images as generated images through the image generation model;
and the training module is used for training the image generation model by taking the deviation between the minimum second number of generated images and the minimum second number of second CT images as an optimization target.
9. The apparatus of claim 8, wherein the acquisition module is operable to acquire a second number of second CT images; and downsampling the second number of second CT images according to a preset sampling rate to obtain the first number of first CT images, wherein the sampling rate corresponds to the first number.
10. The apparatus of claim 9, wherein the image generation model comprises: a feature extraction network and a position coding network;
the input module is further used for determining, for each first CT image, an image coordinate of the first CT image under a preset image coordinate system as an image coordinate corresponding to the first CT image before inputting the first number of first CT images into the image generation model to be trained; determining an image position matrix corresponding to the first CT image according to the image coordinates corresponding to the first CT image and the sampling rate;
The input module is specifically configured to input the first number of first CT images and an image position matrix corresponding to each first CT image into the image generation model, so that the image generation model performs feature fusion on the first CT images and the image features corresponding to the first CT images, and input the first CT images into the feature extraction network to obtain image features corresponding to the first CT images, and input the image position matrix corresponding to the first CT images into the position coding network to obtain position codes corresponding to the first CT images, and perform feature fusion on the position codes corresponding to the first CT images and the image features corresponding to the first CT images to obtain fused features corresponding to the first CT images; and generating the second number of CT images as generated images according to the fused features corresponding to each first CT image through the image generation model.
11. The apparatus of claim 8 or 10, wherein the input module is further configured to determine a cropping ratio prior to inputting the first number of first CT images into an image generation model to be trained; cutting the first CT images with the first quantity according to the determined cutting proportion to obtain first CT images with the first quantity after cutting;
The input module is specifically configured to input the first number of cropped first CT images into the image generation model, so as to generate, through the image generation model, the second number of CT images as generated images.
12. The apparatus of claim 11, wherein the input module is specifically configured to read device information of a device performing a training task of the image generation model; and determining the clipping proportion according to the equipment information.
13. The apparatus of claim 8, wherein the first CT image comprises: a ThickCT image, the second CT image comprising: thinCT images.
14. An apparatus for displaying an image, comprising:
an acquisition module for acquiring a first number of CT images scanned by a patient scanned by an CT apparatus;
the input module is used for inputting the first number of CT images into a pre-trained image generation model so that the image generation model generates a second number of CT images according to the first number of CT images, the first number is smaller than the second number, and the image generation model is obtained by training according to the method of any one of claims 1-6;
The determining module is used for determining CT images to be displayed according to the second number of CT images;
and the display module is used for displaying the CT image to be displayed.
15. A computer readable storage medium, characterized in that the storage medium stores a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-7.
16. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of the preceding claims 1-7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310598313.0A CN116309924B (en) | 2023-05-25 | 2023-05-25 | Model training method, image display method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310598313.0A CN116309924B (en) | 2023-05-25 | 2023-05-25 | Model training method, image display method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116309924A true CN116309924A (en) | 2023-06-23 |
| CN116309924B CN116309924B (en) | 2023-08-18 |
Family
ID=86818998
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310598313.0A Active CN116309924B (en) | 2023-05-25 | 2023-05-25 | Model training method, image display method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116309924B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170003368A1 (en) * | 2015-07-02 | 2017-01-05 | The General Hospital Corporation | System and Method For High Resolution Diffusion Imaging |
| CN112529150A (en) * | 2020-12-01 | 2021-03-19 | 华为技术有限公司 | Model structure, model training method, image enhancement method and device |
| CN112784857A (en) * | 2021-01-29 | 2021-05-11 | 北京三快在线科技有限公司 | Model training and image processing method and device |
| CN114359935A (en) * | 2021-12-30 | 2022-04-15 | 北京三快在线科技有限公司 | Model training and form recognition method and device |
| CN114596203A (en) * | 2022-03-03 | 2022-06-07 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating images and for training image generation models |
| CN115131214A (en) * | 2022-08-31 | 2022-09-30 | 南京邮电大学 | Indoor aged person image super-resolution reconstruction method and system based on self-attention |
| WO2022213846A1 (en) * | 2021-04-07 | 2022-10-13 | 北京三快在线科技有限公司 | Training of recognition model |
| CN115272667A (en) * | 2022-06-24 | 2022-11-01 | 中科星睿科技(北京)有限公司 | Farmland image segmentation model training method and device, electronic equipment and medium |
| CN115546094A (en) * | 2022-01-26 | 2022-12-30 | 北京歌锐科技有限公司 | Model training method, and CT image optimization method and device |
| CN115953494A (en) * | 2023-03-09 | 2023-04-11 | 南京航空航天大学 | Multi-task high-quality CT image reconstruction method based on low dose and super-resolution |
-
2023
- 2023-05-25 CN CN202310598313.0A patent/CN116309924B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170003368A1 (en) * | 2015-07-02 | 2017-01-05 | The General Hospital Corporation | System and Method For High Resolution Diffusion Imaging |
| CN112529150A (en) * | 2020-12-01 | 2021-03-19 | 华为技术有限公司 | Model structure, model training method, image enhancement method and device |
| CN112784857A (en) * | 2021-01-29 | 2021-05-11 | 北京三快在线科技有限公司 | Model training and image processing method and device |
| WO2022213846A1 (en) * | 2021-04-07 | 2022-10-13 | 北京三快在线科技有限公司 | Training of recognition model |
| CN114359935A (en) * | 2021-12-30 | 2022-04-15 | 北京三快在线科技有限公司 | Model training and form recognition method and device |
| CN115546094A (en) * | 2022-01-26 | 2022-12-30 | 北京歌锐科技有限公司 | Model training method, and CT image optimization method and device |
| CN114596203A (en) * | 2022-03-03 | 2022-06-07 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating images and for training image generation models |
| CN115272667A (en) * | 2022-06-24 | 2022-11-01 | 中科星睿科技(北京)有限公司 | Farmland image segmentation model training method and device, electronic equipment and medium |
| CN115131214A (en) * | 2022-08-31 | 2022-09-30 | 南京邮电大学 | Indoor aged person image super-resolution reconstruction method and system based on self-attention |
| CN115953494A (en) * | 2023-03-09 | 2023-04-11 | 南京航空航天大学 | Multi-task high-quality CT image reconstruction method based on low dose and super-resolution |
Non-Patent Citations (4)
| Title |
|---|
| ZHILEI LIU等: "Joint face completion and super-resolution using multi-scale feature relation learning", 《J. VIS. COMMUN. IMAGE R.》 * |
| 李哲远等: "注意力机制在单图像超分辨率中的分析研究", 《集成技术》, vol. 11, no. 5, pages 58 - 79 * |
| 杨敬钰等: "基于位置约束的参考图引导的遥感影像超分辨率网络", 《天津大学学报》, vol. 56, no. 4, pages 372 - 380 * |
| 许宁宁;郑凯;: "基于多损失融合与谱归一化的图像超分辨率方法", 计算机应用研究, no. 08 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116309924B (en) | 2023-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272454B (en) | Coordinate system calibration method and device of augmented reality equipment | |
| CN116977525B (en) | Image rendering method and device, storage medium and electronic equipment | |
| CN117372631B (en) | Training method and application method of multi-view image generation model | |
| CN117635822A (en) | Model training method and device, storage medium and electronic equipment | |
| CN116309924B (en) | Model training method, image display method and device | |
| CN116030247B (en) | Medical image sample generation method and device, storage medium and electronic equipment | |
| CN117911630B (en) | Three-dimensional human modeling method and device, storage medium and electronic equipment | |
| CN117726907B (en) | Training method of modeling model, three-dimensional human modeling method and device | |
| CN117173321B (en) | Method and device for selecting three-dimensional reconstruction texture view | |
| CN110599527A (en) | MRA image data registration method and device | |
| CN117689822B (en) | Three-dimensional model construction method and device, storage medium and electronic equipment | |
| CN116229218B (en) | Model training and image registration method and device | |
| CN117152040B (en) | Point cloud fusion method and device based on depth map | |
| CN117726760B (en) | Training method and device for three-dimensional human body reconstruction model of video | |
| CN116152246B (en) | Image recognition method, device, equipment and storage medium | |
| CN116579414B (en) | Model training method, MRI thin layer data reconstruction method, device and equipment | |
| CN116612244B (en) | Image generation method and device, storage medium and electronic equipment | |
| CN117392485B (en) | Image generation model training method, service execution method, device and medium | |
| CN116309428A (en) | Method and device for determining region of interest, storage medium and electronic equipment | |
| CN117911630A (en) | Three-dimensional human modeling method and device, storage medium and electronic equipment | |
| CN117745956A (en) | Pose guidance-based image generation method, device, medium and equipment | |
| CN117975202A (en) | Model training method, service execution method, device, medium and equipment | |
| CN117788696A (en) | Task execution method and device based on three-dimensional modeling and electronic equipment | |
| CN117893696A (en) | Three-dimensional human body data generation method and device, storage medium and electronic equipment | |
| CN117808976A (en) | Three-dimensional model construction method and device, storage medium and electronic equipment |
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 |