CN116645312A - Image processing method, device, computer equipment and storage medium - Google Patents
Image processing method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN116645312A CN116645312A CN202210151134.8A CN202210151134A CN116645312A CN 116645312 A CN116645312 A CN 116645312A CN 202210151134 A CN202210151134 A CN 202210151134A CN 116645312 A CN116645312 A CN 116645312A
- Authority
- CN
- China
- Prior art keywords
- image
- initial
- contour
- initial sample
- images
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 135
- 238000012549 training Methods 0.000 claims abstract description 99
- 238000006243 chemical reaction Methods 0.000 claims abstract description 96
- 239000000523 sample Substances 0.000 claims description 262
- 238000013507 mapping Methods 0.000 claims description 59
- 238000000034 method Methods 0.000 claims description 43
- 239000013074 reference sample Substances 0.000 claims description 42
- 230000009466 transformation Effects 0.000 claims description 41
- 238000004590 computer program Methods 0.000 claims description 40
- 230000000694 effects Effects 0.000 abstract description 12
- 230000008569 process Effects 0.000 description 20
- 238000010586 diagram Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000002597 diffusion-weighted imaging Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000001308 synthesis method Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The application relates to an image processing method, an image processing device, a computer device and a storage medium. Obtaining a plurality of groups of training samples of an image mode conversion model by obtaining an initial sample set of the image mode conversion model, wherein the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and performing contour alignment processing on the plurality of initial sample images and the initial label images corresponding to each initial sample image; wherein the initial sample image and the initial label image correspond to different modalities; that is, before training an image mode conversion model by adopting initial sample images and initial tag images of different modes, contour alignment processing is performed on the initial sample images and the initial tag images under different modes in advance, so that contour consistency of the initial sample images and the initial tag images can be improved, differences between the initial sample images and the initial tag images can be reduced, and conversion effects of the image mode conversion model can be improved.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to an image processing method, an image processing apparatus, a computer device, and a storage medium.
Background
After scanning an examination object to obtain a medical image of a certain modality, the medical image of the modality may also be converted into a medical image of another modality using medical image synthesis techniques. In this way, a medical image of another modality of the examination subject is obtained, without the examination subject having to be scanned again, which can reduce the electromagnetic radiation to which the examination subject is subjected.
In a conventional medical image synthesis method, an image generation model based on deep learning is generally adopted, and an image of one mode is input into the image generation model to obtain an image of another mode.
However, in the training process of the image generation model, the difference between the sample image and the label image is often ignored, which results in a large difference between the image after the mode conversion and the image before the mode conversion obtained by adopting the image generation model, and the effect of mode conversion image is poor.
Disclosure of Invention
In view of the foregoing, it is desirable to provide an image processing method, apparatus, computer device, and computer-readable storage medium capable of reducing the variability between a sample image and a label image, and thereby improving the conversion effect of a modal converted image.
In a first aspect, the present application provides an image processing method. The method comprises the following steps:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
and performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In one embodiment, performing contour alignment processing on a plurality of initial sample images and an initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model, including:
for each initial sample image, taking the outline of the initial label image corresponding to the initial sample image as a mapping outline, taking the initial sample image as a processing object, and performing outline alignment processing to obtain a reference sample image;
a set of training samples is constructed based on the reference sample image and the initial label image.
In one embodiment, performing contour alignment processing on a plurality of initial sample images and an initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model, including:
For each initial tag image, taking the outline of the initial sample image corresponding to the initial tag image as a mapping outline, taking the initial tag image as a processing object, and performing outline alignment processing to obtain a reference tag image;
a set of training samples is constructed based on the initial sample image and the reference label image.
In one embodiment, performing contour alignment processing on a plurality of initial sample images and an initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model, including:
aiming at each initial sample image and an initial label image corresponding to the initial sample image, taking a preset contour as a mapping contour, taking the initial sample image and the initial label image as processing objects, and performing contour alignment processing to obtain a reference sample image and a reference label image;
a set of training samples is constructed based on the reference sample image and the reference label image.
In one embodiment, performing the contour alignment process includes:
determining a contour difference between the mapped contour and the reference contour; the reference contour is the original contour of the processing object;
and performing contour adjustment on the processing object according to the contour difference.
In one embodiment, determining the contour difference between the mapped contour and the reference contour includes:
registering the reference contour and the mapping contour to obtain deformation field information of mapping the reference contour to the mapping contour.
In one embodiment, the method further comprises: and performing model training on a plurality of groups of training samples to obtain an image mode conversion model.
In one embodiment, the method further comprises: inputting a target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of the contour difference between the target image and the modal transformation image is smaller than a preset value.
In one embodiment, the target image is the same contour as the modal converted image.
In a second aspect, the present application also provides an image processing apparatus. The device comprises:
the acquisition module is used for acquiring an initial sample set of the image modal transformation model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modalities;
and the processing module is used for carrying out contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing said computer program:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
and performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
And performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
The image processing method, the image processing device, the computer equipment and the storage medium, wherein the computer equipment obtains a plurality of groups of training samples of the image mode conversion model by acquiring an initial sample set of the image mode conversion model, wherein the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and then performing contour alignment processing on the plurality of initial sample images and the initial label images corresponding to each initial sample image; wherein the initial sample image and the initial label image correspond to different modalities; that is, before the image mode conversion model training is performed by adopting the initial sample image and the initial label image under different modes, the contour alignment processing is performed on the initial sample image and the initial label image under the different modes in advance, so that the contour consistency of the initial sample image and the initial label image can be improved, and the difference between the initial sample image and the initial label image can be reduced; furthermore, after the image mode conversion model is obtained by training a plurality of groups of training samples subjected to contour alignment treatment and mode conversion is carried out on an input original image, the difference between the obtained mode converted image and the contour of the original image is small, namely, the difference between the mode converted image and the original image can be reduced, and the conversion effect of the image mode conversion model is improved.
Drawings
FIG. 1 is a flow chart of an image processing method in one embodiment;
FIG. 2 is a flow diagram of a contour alignment process in one embodiment;
FIG. 3 is a schematic image of a contour alignment process in one embodiment;
FIG. 4 is a flow chart of a contour alignment process according to another embodiment;
FIG. 5 is a schematic image of another embodiment of a contour alignment process;
FIG. 6 is a flow chart of a contour alignment process in another embodiment;
FIG. 7 is a schematic image of another embodiment of a contour alignment process;
FIG. 8 is a flow diagram of a method of performing contour alignment processing in one embodiment;
FIG. 9 is a schematic diagram of a training process of a CycleGAN network-based image modality conversion model in one embodiment;
FIG. 10 is a block diagram showing the structure of an image processing apparatus in one embodiment;
FIG. 11 is an internal block diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
First, before the technical scheme of the embodiment of the present application is specifically described, a description is first given of a technical background or a technical evolution context on which the embodiment of the present application is based. For the existing medical image synthesis method, under the rapid development of the deep learning technology, a deep learning network is mostly adopted to perform supervised learning to train and obtain an image generation model, training data used by the deep learning network are registered data pairs (namely sample images and label images under two different modes), and voxels of each mode in each data pair correspond to each other one by one. In the process of carrying out data registration on sample images and label images in different modes, rigid registration and deformation registration are usually carried out on complete sample images and complete label images in sequence, and certain registration errors exist due to complex registration algorithm; therefore, for a certain difference still exists between the registered sample image and the label image, especially, a large difference between contours of the registered sample image and the label image (wherein the difference of outer contours may be larger) is caused, so that the consistency of contours of two different mode images is poor, and further, when an image generating model obtained by training the sample image and the label image with the contour difference is used for carrying out mode conversion on an original image to synthesize an image of another mode, a large difference exists between the contour of the image after mode conversion and the contour of the original image, so that the effect of image synthesis is poor.
Therefore, the embodiment of the application provides an image processing method, which enables the contour alignment processing to be carried out on the sample images and the label images of different modes used in the training before the training of the image generation model, so as to keep the contour of the sample images and the label images of different modes consistent, improve the contour consistency of training data pairs, and further improve the image synthesis effect.
The technical scheme related to the embodiment of the application is described below in connection with the scene to which the embodiment of the application is applied.
The image processing method provided by the embodiment of the application can be applied to computer equipment, wherein the computer equipment can be a terminal or a server, the terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, scanning equipment, processing equipment connected with the scanning equipment and the like, and the server can be realized by an independent server or a server cluster formed by a plurality of servers.
In one embodiment, as shown in fig. 1, an image processing method is provided, and the method is applied to the above computer device for illustration, and includes the following steps:
Step 101, an initial sample set of the image modality conversion model is acquired.
The initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and the initial sample images and the initial label images correspond to different modalities.
Optionally, in practical application, the image of any mode may be subjected to mode conversion to generate an image of another mode, and when the image mode conversion model is trained, an initial sample set under different modes used in training may be determined according to the requirement of actual mode conversion, so as to train to obtain different image mode conversion models. For example: the mode conversion from MR image to CT image can be carried out; modality conversion of CT images to MR images; alternatively, for different MR sequence images, for example, a mutual transformation between a T1 sequence image and a T2 sequence image, a mutual transformation between a T1 sequence image and a diffusion Weighted (diffusion Weighted Imaging, DWI) sequence image, a mutual transformation between a DWI sequence image and a magnetically Weighted (SWI) sequence image, etc. is achieved.
Taking the mode conversion from MR image to CT image as an example, training to obtain an image mode conversion model of MR-CT, wherein an initial sample set required by the image mode conversion model of MR-CT comprises a plurality of initial MR sample images and initial CT label images corresponding to each initial MR sample image.
Alternatively, the initial MR sample image may be a plurality of historical MR images acquired by the MR scanning device scanning a plurality of scanning objects, and the initial CT tag image may be a plurality of historical CT images acquired by the plurality of scanning objects by the CT scanning device; the initial MR sample image and the initial CT label image may also be MR images and CT images corresponding to different scan objects obtained from an image database; in this embodiment, the method for obtaining the initial sample set of the image modality conversion model is not limited.
Step 102, performing contour alignment processing on a plurality of initial sample images and initial label images corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
Wherein, an initial sample image and an initial label image corresponding to the initial sample image are a group of initial training samples.
Optionally, before training, contour alignment processing may be performed on each set of initial training samples in the initial sample set in advance, so as to improve contour consistency between images of two different modalities in each set of initial training samples, so as to obtain multiple sets of training samples of the image modality conversion model. The outline of the image may be an outer outline of the image, or may be an outline of a preset tissue in the image, or the like.
Optionally, a preset contour alignment algorithm may be adopted to perform contour alignment processing on each group of initial sample images and initial label images corresponding to the initial sample images, so as to obtain initial sample images and initial label images after contour alignment. Optionally, a non-rigid registration algorithm may be further used to perform contour alignment processing on each set of initial sample images and the initial label image corresponding to the initial sample images, so as to obtain initial sample images and initial label images after contour alignment. In this embodiment, the method and the procedure of the contour alignment process are not particularly limited, as long as the contour between each set of initial sample images and the initial label image corresponding to the initial sample image can be kept consistent.
Optionally, after the computer device obtains the initial sample set of the image mode conversion model, rigid registration can be performed on each group of initial training samples in the initial sample set to obtain a plurality of groups of registered initial training samples; and then, carrying out contour alignment processing on each group of registered initial training samples to obtain a plurality of groups of training samples which are required by the image modal transformation model and are subjected to rigid registration and contour alignment processing.
In the image processing method, the computer equipment obtains a plurality of groups of training samples of the image mode conversion model by acquiring an initial sample set of the image mode conversion model, wherein the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and then performing contour alignment processing on the plurality of initial sample images and the initial label images corresponding to each initial sample image; wherein the initial sample image and the initial label image correspond to different modalities; that is, before the image mode conversion model training is performed by adopting the initial sample image and the initial label image under different modes, the contour alignment processing is performed on the initial sample image and the initial label image under the different modes in advance, so that the contour consistency of the initial sample image and the initial label image can be improved, and the difference between the initial sample image and the initial label image can be reduced; furthermore, after the image mode conversion model is obtained by training a plurality of groups of training samples subjected to contour alignment treatment and mode conversion is carried out on an input original image, the difference between the obtained mode converted image and the contour of the original image is small, namely, the difference between the mode converted image and the original image can be reduced, and the conversion effect of the image mode conversion model is improved.
A variety of different contour alignment processes may be included for the implementation of step 102, as will be described in detail below.
The first way is:
in one embodiment, as shown in fig. 2, the step 102 includes:
in step 201, for each initial sample image, a contour of an initial label image corresponding to the initial sample image is taken as a mapping contour, and the initial sample image is taken as a processing object, and contour alignment processing is performed to obtain a reference sample image.
The reference sample image is an initial sample image with the outline of the initial label image, that is, the outline of the reference sample image is consistent with the outline of the initial label image.
Alternatively, the contour of the initial sample image may be mapped to the contour of the initial label image to obtain an initial sample image having the contour of the initial label image; in the embodiment, as shown in fig. 3, the contour of the initial CT label image is used as a mapping contour, and contour alignment processing is performed on the contour of the initial MR sample image, so that the contour of the initial MR sample image can be mapped to the contour of the initial CT label image to obtain a reference MR sample image with a CT contour, and the reference MR sample image with a CT contour is a reference sample image corresponding to the initial sample image (initial MR sample image).
Step 202, a set of training samples is constructed based on the reference sample image and the initial label image.
In this embodiment, for each initial sample image in the initial sample set, a contour of an initial label image corresponding to the initial sample image is taken as a mapping contour, and the initial sample image is taken as a processing object, and contour alignment processing is performed to obtain a reference sample image; constructing a group of training samples based on the reference sample image and the initial label image; that is, in this embodiment, when the contour alignment process is performed on the initial sample image and the initial tag image, the contour of the initial sample image may be mapped to the contour of the initial tag image, that is, the contour of the initial sample image may be corrected to the contour of the initial tag image, so as to ensure the consistency of the contour of the sample image and the contour of the tag image, reduce the difference between the sample image and the tag image, further improve the transformation effect of the image mode transformation model, improve the consistency between the contour of the image after mode transformation and the contour of the original image, and improve the accuracy of the contour of the image after mode transformation.
The second way is:
In one embodiment, as shown in fig. 4, the step 102 includes:
in step 401, for each initial tag image, a contour of an initial sample image corresponding to the initial tag image is taken as a mapping contour, and the initial tag image is taken as a processing object, and contour alignment processing is performed to obtain a reference tag image.
The reference label image is an initial label image with the outline of the initial sample image, namely the outline of the reference label image is consistent with the outline of the initial sample image.
Alternatively, the contour of the initial label image may be mapped to the contour of the initial sample image to obtain an initial label image having the contour of the initial sample image; in this embodiment, as shown in fig. 5, the contour of the initial MR sample image is used as a mapping contour, and contour alignment processing is performed on the contour of the initial CT label image, so that the contour of the initial CT label image can be mapped to the contour of the initial MR sample image to obtain a reference CT label image with an MR contour, and the reference CT label image with an MR contour is the reference label image corresponding to the initial label image (initial CT label image).
A set of training samples is constructed based on the initial sample image and the reference label image, step 402.
In this embodiment, for each initial tag image in the initial sample set, a contour of the initial sample image corresponding to the initial tag image is taken as a mapping contour, and the initial tag image is taken as a processing object, and contour alignment processing is performed to obtain a reference tag image; constructing a group of training samples based on the initial sample image and the reference label image; that is, in this embodiment, when the contour alignment process is performed on the initial sample image and the initial tag image, the contour of the initial tag image may be mapped to the contour of the initial sample image, that is, the contour of the initial tag image may be corrected, so as to correct the contour of the initial tag image to the contour of the initial sample image, so as to ensure consistency of the contour of the sample image and the contour of the tag image, reduce the difference between the sample image and the tag image, further improve the transformation effect of the image mode transformation model, improve the consistency between the contour of the image after mode transformation and the contour of the original image, and improve the accuracy of the contour of the image after mode transformation.
Third mode:
In one embodiment, as shown in fig. 6, the step 102 includes:
in step 601, for each initial sample image and an initial label image corresponding to the initial sample image, a preset contour is taken as a mapping contour, and the initial sample image and the initial label image are taken as processing objects, and contour alignment processing is performed to obtain a reference sample image and a reference label image.
The reference sample image is an initial sample image with a preset contour, and the reference label image is an initial label image with a preset contour, that is, the contours of the reference sample image and the reference label image are both preset contours.
Alternatively, the contour of the initial sample image may be mapped to a preset contour to obtain an initial label image having the preset contour; and the contour of the initial label image can be mapped to a preset contour to obtain an initial label image with the preset contour; in the above-mentioned mode conversion from MR image to CT image, as shown in fig. 7, in this embodiment, a preset contour is used as a mapping contour, and contour alignment processing is performed on the contour of the initial MR sample image and the contour of the initial CT label image, for example: under the condition that the preset contour is a rectangular contour, mapping the contour of the initial MR sample image to the rectangular contour to obtain a reference MR sample image with the rectangular contour, and mapping the contour of the initial CT label image to the rectangular contour to obtain a reference CT label image with the rectangular contour; the contours of the reference MR sample image after contour alignment processing and the reference CT label image after contour alignment processing are rectangular contours, and the contours of the reference MR sample image and the reference CT label image are consistent. It should be noted that the preset profile may be any profile, for example: circular profile, rectangular profile, oval profile, etc., the shape of the preset profile is not limited in this embodiment.
A set of training samples is constructed based on the reference sample image and the reference label image, step 602.
In this embodiment, for each initial sample image and an initial label image corresponding to the initial sample image, a preset contour is taken as a mapping contour, and the initial sample image and the initial label image are taken as processing objects, and contour alignment processing is performed to obtain a reference sample image and a reference label image; constructing a group of training samples based on the reference sample image and the reference label image; that is, in this embodiment, when the contour alignment process is performed on the initial sample image and the initial tag image, the contour of the initial sample image may be mapped to a preset contour, and the contour of the initial tag image may be mapped to a preset contour, that is, both the contour of the initial sample image and the contour of the initial tag image may be corrected, so that both the contour of the initial sample image and the contour of the initial tag image may be corrected to another uniform contour, so as to ensure the consistency of the contour of the sample image and the contour of the tag image, reduce the difference between the sample image and the tag image, further improve the transformation effect of the image mode transformation model, improve the consistency between the contour of the image after mode transformation and the contour of the original image, and improve the accuracy of the contour of the image after mode transformation.
In an alternative embodiment of the present application, as shown in fig. 8, the above-mentioned process of performing contour alignment processing on a processing object according to a mapped contour may include:
in step 801, a contour difference between the mapped contour and the reference contour is determined.
Wherein the reference contour is an original contour of the processing object.
Optionally, the reference contour and the mapping contour may be registered to obtain deformation field information between the reference contour and the mapping contour, where the deformation field information may be used to characterize a conversion relationship between the reference contour and the mapping contour; alternatively, the reference contour and the mapping contour may be non-rigidly registered to obtain deformation field information of the mapping of the reference contour to the mapping contour.
Step 802, performing contour adjustment on the processing object according to the contour difference.
Optionally, after determining the contour difference between the reference contour and the mapped contour, the computer device may perform contour adjustment on the contour of the processing object according to the contour difference; alternatively, in the case that the contour difference is deformation field information of the reference contour mapped to the mapped contour, the deformation field information may be applied to the processing object to implement contour adjustment of the processing object. In the case of the above-described mode conversion from MR image to CT image, when MR image with CT contour is obtained by mapping MR image to CT image, deformation field information of MR image contour mapped to CT image contour can be obtained by non-rigid registration of MR image contour and CT image contour, and then the deformation field information can be applied to MR image to obtain MR image with CT contour.
In this embodiment, the computer device may improve the feasibility and accuracy of the contour alignment process by determining the contour difference between the mapping contour and the reference contour of the processing object, and performing contour adjustment on the processing object according to the contour difference, so as to adjust the original contour (i.e., the reference contour) of the processing object to the preset contour.
In an optional embodiment of the present application, after performing contour alignment processing on a plurality of initial sample images in an initial sample set and an initial label image corresponding to each initial sample image, and obtaining a plurality of groups of training samples, the computer device may use the plurality of groups of training samples to train an initial modal transformation network, so as to obtain an image modal transformation model.
Optionally, the initial modality conversion network may be an image generation network, which may generate a countercheck network (Generative Adversarial Networks, GAN for short), a loop-consistency generation countercheck network (Cycle-Consistent Generative Adversarial Networks, cycleGAN for short), or the like, or may be an image generation network based on any other type of deep learning network, for example: an image generation network based on a convolutional neural network (Convolutional Neural Networks, abbreviated as CNN network), an image generation network based on a depth residual network (Deep residual network, abbreviated as res net network), or an image generation network based on a CNN network and a res net network, etc.; it should be noted that, in the embodiment of the present application, the network form of the initial modality conversion network is not limited.
Optionally, based on the above image generating networks, an initial sample image may be mapped to an initial label image to obtain a reference sample image, and training the image generating network according to the reference sample image and the initial label image to obtain an image mode conversion model; the initial label image can be mapped to the initial sample image to obtain a reference label image, and the image generation network is trained according to the initial sample image and the reference label image to obtain an image mode conversion model; the initial sample image and the initial label image can be mapped to a preset outline to obtain a reference sample image and a reference label image, and the image generation network is trained according to the reference sample image and the reference label image to obtain an image mode conversion model.
Taking a CycleGAN network as an example, a training process of the image mode conversion model is described below:
as shown in fig. 9, the CycleGAN network is formed by combining two mirror-symmetrical GAN networks, wherein the two GAN networks share two generators and each provide a arbiter, so the CycleGAN network includes two generators and two arbiter, which are respectively: generator G mapping data X to data Y Y And a generator G for mapping data Y to data X X And judging fake X (data Y is processed by generator G X Obtained by processing) a discriminator D of authenticity X And judging fake Y (data X via generator G) Y Obtained by processing) a discriminator D of authenticity Y . In the CycleGAN network, the generator has an encoder-decoder architecture similar to the UNet model, and the architecture combines deep features and shallow features to better recover image details.
Alternatively, the loss function used in the CycleGAN network may include not only the loss function Ladv of the GAN network (as shown in formula (1)), but also a cyclic consistency loss function Lcycle (as shown in formula (2)), where the cyclic consistency loss function can ensure that the data X and fake X are in one-to-one correspondence, i.e., G X (G Y (X))＝G X (fake Y)≈X。
During training, real X and real Y can be input into the CycleGAN network, and real X passes through the generator G Y Get fake Y, real Y passes through generator G X Obtaining fake X; then, the fake Y passes through generator G X Obtaining cycle X, and fake X passing through generator G Y Obtaining cycle Y; further, fake X and real X are input into a discriminator D X And input the fake Y and real Y into a arbiter D Y Calculating a loss value Ladv, and calculating a loss value Lcycle for real X and cycle X, real Y and cycle Y respectively; the model parameters can also be updated by back-propagating gradients of the calculated model parameters using an Adam optimizer. It should be noted that, real X may be an initial sample image or a reference sample image, and real Y may be an initial tag image or a reference tag image, so long as the contours of real X and real Y are identical.
The Adam optimization algorithm is an extension of random gradient descent, and momentum and self-adaptive learning rate are used, so that the convergence speed of the network can be improved. The process of updating the model parameters by the Adam optimizer may include: first, the gradient g of each parameter is obtained according to the back propagation process t And according to gradient g t Estimating the first moment m of each parameter t (as shown in equation (3)) and a second moment v t (as shown in formula (4)), wherein beta 1 And beta 2 Is the exponential decay rate; optionally, to alleviate the inaccuracy of the first and second moment estimation at the initial moment, formulas (5) and (6) may be used to correct the first and second moments; the model parameters θ can then be based on the first and second moments of the gradient t An update is made (as shown in equation (7)), where α is the learning rate.
m t ＝β 1 ·m t-1 +(1-β 1 )·g t (3)
Optionally, before training the initial modal transformation network by using multiple sets of training samples to obtain an image modal transformation model, the computer device may further perform image preprocessing on the multiple sets of training samples to obtain a training sample set and a test sample set, then, may train the initial modal transformation network by using the training sample set to obtain the image modal transformation model, and then, may also test the image modal transformation model by using the test sample set. Optionally, the operations of performing image preprocessing on the multiple sets of training samples may include, but are not limited to, normalization processing, randomly dividing the multiple sets of training samples into a training sample set and a test sample set, performing data enhancement (including, but not limited to, flipping, translation, rotation, deformation, etc.) on the multiple sets of training samples, and so on; it should be noted that, the above-mentioned respective image preprocessing operations may be existing processing procedures, which are not discussed in detail herein, and in addition, the specific operations of image preprocessing in the embodiment of the present application are not limited.
In this embodiment, since the processed multiple sets of training samples have higher consistency of the contours of the sample images and the label images in each set of training samples, the image mode conversion model trained by the multiple sets of training samples after contour alignment processing is used to obtain a mode converted image, and compared with the mode converted image obtained by directly using the model trained by the initial sample set, the mode conversion effect of the model is better because the consistency of the contours of the sample images and the label images in each set of training samples is higher.
In an optional embodiment of the present application, in a case where image mode conversion is required, the computer device may input the obtained target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the outline difference degree between the target image and the modal transformation image is smaller than a preset value; that is, by adopting the image modal transformation model in the embodiment of the application, the contour of the obtained modal transformation image is basically consistent with the contour of the original target image, and the contour difference between the two is smaller. Optionally, the contour of the modal transformation image output by the image modal transformation model is identical to the contour of the original target image, and the obtained modal transformation image has better effect.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides an image processing device for realizing the above-mentioned image processing method. The implementation of the solution provided by the apparatus is similar to the implementation described in the above method, so the specific limitation of one or more embodiments of the image processing apparatus provided below may refer to the limitation of the image processing method hereinabove, and will not be repeated herein.
In one embodiment, as shown in fig. 10, there is provided an image processing apparatus including: an acquisition module 1001 and a processing module 1002, wherein:
an obtaining module 1001, configured to obtain an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modalities;
the processing module 1002 is configured to perform contour alignment processing on a plurality of initial sample images and an initial label image corresponding to each initial sample image, so as to obtain a plurality of groups of training samples of the image mode conversion model.
In one embodiment, the processing module 1002 is specifically configured to, for each initial sample image, perform contour alignment processing with respect to a contour of an initial label image corresponding to the initial sample image as a mapping contour and the initial sample image as a processing object to obtain a reference sample image; a set of training samples is constructed based on the reference sample image and the initial label image.
In one embodiment, the processing module 1002 is specifically configured to, for each initial tag image, perform contour alignment processing with respect to a contour of an initial sample image corresponding to the initial tag image as a mapping contour and the initial tag image as a processing object to obtain a reference tag image; a set of training samples is constructed based on the initial sample image and the reference label image.
In one embodiment, the processing module 1002 is specifically configured to, for each initial sample image and an initial tag image corresponding to the initial sample image, perform contour alignment processing with a preset contour as a mapping contour and the initial sample image and the initial tag image as processing objects to obtain a reference sample image and a reference tag image; a set of training samples is constructed based on the reference sample image and the reference label image.
In one embodiment, the processing module 1002 is specifically configured to determine a contour difference between the mapped contour and the reference contour, and perform contour adjustment on the processing object according to the contour difference; wherein the reference contour is an original contour of the processing object.
In one embodiment, the processing module 1002 is specifically configured to register the reference contour and the mapping contour, so as to obtain deformation field information of mapping the reference contour to the mapping contour.
In one embodiment, the apparatus further comprises a training module; the training module is used for carrying out model training on a plurality of groups of training samples to obtain an image mode conversion model.
In one embodiment, the apparatus further comprises a modality conversion module; the mode conversion module is used for inputting the target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of the contour difference between the target image and the modal transformation image is smaller than a preset value.
In one embodiment, the target image is the same contour as the modal converted image.
The respective modules in the above-described image processing apparatus may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a terminal, and the internal structure thereof may be as shown in fig. 11. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement an image processing method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by those skilled in the art that the structure shown in FIG. 11 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
and performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In one embodiment, the processor when executing the computer program further performs the steps of: for each initial sample image, taking the outline of the initial label image corresponding to the initial sample image as a mapping outline, taking the initial sample image as a processing object, and performing outline alignment processing to obtain a reference sample image; a set of training samples is constructed based on the reference sample image and the initial label image.
In one embodiment, the processor when executing the computer program further performs the steps of: for each initial tag image, taking the outline of the initial sample image corresponding to the initial tag image as a mapping outline, taking the initial tag image as a processing object, and performing outline alignment processing to obtain a reference tag image; a set of training samples is constructed based on the initial sample image and the reference label image.
In one embodiment, the processor when executing the computer program further performs the steps of: aiming at each initial sample image and an initial label image corresponding to the initial sample image, taking a preset contour as a mapping contour, taking the initial sample image and the initial label image as processing objects, and performing contour alignment processing to obtain a reference sample image and a reference label image; a set of training samples is constructed based on the reference sample image and the reference label image.
In one embodiment, the processor when executing the computer program further performs the steps of: determining a contour difference between the mapping contour and the reference contour, and performing contour adjustment on the processing object according to the contour difference; the reference contour is the original contour of the processing object.
In one embodiment, the processor when executing the computer program further performs the steps of: registering the reference contour and the mapping contour to obtain deformation field information of mapping the reference contour to the mapping contour.
In one embodiment, the processor when executing the computer program further performs the steps of: and performing model training on a plurality of groups of training samples to obtain an image mode conversion model.
In one embodiment, the processor when executing the computer program further performs the steps of: inputting a target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of the contour difference between the target image and the modal transformation image is smaller than a preset value.
In one embodiment, the processor when executing the computer program further performs the steps of: the target image is identical to the contour of the modal transformation image.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
and performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In one embodiment, the computer program when executed by the processor further performs the steps of: for each initial sample image, taking the outline of the initial label image corresponding to the initial sample image as a mapping outline, taking the initial sample image as a processing object, and performing outline alignment processing to obtain a reference sample image; a set of training samples is constructed based on the reference sample image and the initial label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: for each initial tag image, taking the outline of the initial sample image corresponding to the initial tag image as a mapping outline, taking the initial tag image as a processing object, and performing outline alignment processing to obtain a reference tag image; a set of training samples is constructed based on the initial sample image and the reference label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: aiming at each initial sample image and an initial label image corresponding to the initial sample image, taking a preset contour as a mapping contour, taking the initial sample image and the initial label image as processing objects, and performing contour alignment processing to obtain a reference sample image and a reference label image; a set of training samples is constructed based on the reference sample image and the reference label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: determining a contour difference between the mapping contour and the reference contour, and performing contour adjustment on the processing object according to the contour difference; the reference contour is the original contour of the processing object.
In one embodiment, the computer program when executed by the processor further performs the steps of: registering the reference contour and the mapping contour to obtain deformation field information of mapping the reference contour to the mapping contour.
In one embodiment, the computer program when executed by the processor further performs the steps of: and performing model training on a plurality of groups of training samples to obtain an image mode conversion model.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting a target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of the contour difference between the target image and the modal transformation image is smaller than a preset value.
In one embodiment, the computer program when executed by the processor further performs the steps of: the target image is identical to the contour of the modal transformation image.
In one embodiment, a computer program product is provided comprising a computer program which, when executed by a processor, performs the steps of:
Acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, wherein the initial sample images and the initial label images correspond to different modes;
and performing contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
In one embodiment, the computer program when executed by the processor further performs the steps of: for each initial sample image, taking the outline of the initial label image corresponding to the initial sample image as a mapping outline, taking the initial sample image as a processing object, and performing outline alignment processing to obtain a reference sample image; a set of training samples is constructed based on the reference sample image and the initial label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: for each initial tag image, taking the outline of the initial sample image corresponding to the initial tag image as a mapping outline, taking the initial tag image as a processing object, and performing outline alignment processing to obtain a reference tag image; a set of training samples is constructed based on the initial sample image and the reference label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: aiming at each initial sample image and an initial label image corresponding to the initial sample image, taking a preset contour as a mapping contour, taking the initial sample image and the initial label image as processing objects, and performing contour alignment processing to obtain a reference sample image and a reference label image; a set of training samples is constructed based on the reference sample image and the reference label image.
In one embodiment, the computer program when executed by the processor further performs the steps of: determining a contour difference between the mapping contour and the reference contour, and performing contour adjustment on the processing object according to the contour difference; the reference contour is the original contour of the processing object.
In one embodiment, the computer program when executed by the processor further performs the steps of: registering the reference contour and the mapping contour to obtain deformation field information of mapping the reference contour to the mapping contour.
In one embodiment, the computer program when executed by the processor further performs the steps of: and performing model training on a plurality of groups of training samples to obtain an image mode conversion model.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting a target image into an image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of the contour difference between the target image and the modal transformation image is smaller than a preset value.
In one embodiment, the computer program when executed by the processor further performs the steps of: the target image is identical to the contour of the modal transformation image.
The user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.
Claims (10)
1. An image processing method, the method comprising:
acquiring an initial sample set of an image modality conversion model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and the initial sample images and the initial label images correspond to different modalities;
and carrying out contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image mode conversion model.
2. The method of claim 1, wherein performing contour alignment processing on the plurality of initial sample images and an initial label image corresponding to each of the initial sample images to obtain a plurality of sets of training samples of the image modality conversion model includes:
for each initial sample image, taking the outline of an initial label image corresponding to the initial sample image as a mapping outline, and taking the initial sample image as a processing object, and performing outline alignment processing to obtain a reference sample image;
a set of training samples is constructed based on the reference sample image and the initial label image.
3. The method of claim 1, wherein performing contour alignment processing on the plurality of initial sample images and an initial label image corresponding to each of the initial sample images to obtain a plurality of sets of training samples of the image modality conversion model includes:
for each initial tag image, taking the outline of an initial sample image corresponding to the initial tag image as a mapping outline, and taking the initial tag image as a processing object, and performing outline alignment processing to obtain a reference tag image;
A set of training samples is constructed based on the initial sample image and the reference label image.
4. The method of claim 1, wherein performing contour alignment processing on the plurality of initial sample images and an initial label image corresponding to each of the initial sample images to obtain a plurality of sets of training samples of the image modality conversion model includes:
for each initial sample image and an initial label image corresponding to the initial sample image, taking a preset contour as a mapping contour, and taking the initial sample image and the initial label image as processing objects, performing contour alignment processing to obtain a reference sample image and a reference label image;
a set of training samples is constructed based on the reference sample image and the reference label image.
5. The method according to any one of claims 2 to 4, wherein the performing contour alignment processing includes:
determining a contour difference between the mapped contour and a reference contour; the reference contour is an original contour of the processing object;
and carrying out contour adjustment on the processing object according to the contour difference.
6. The method of claim 5, wherein the determining a contour difference between the mapped contour and a reference contour comprises:
Registering the reference contour and the mapping contour to obtain deformation field information of the reference contour mapped to the mapping contour.
7. The method according to claim 1, wherein the method further comprises:
inputting a target image into the image mode conversion model to obtain a mode conversion image corresponding to the target image; the degree of contour difference between the target image and the modal transformation image is smaller than a preset value.
8. An image processing apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring an initial sample set of the image modal transformation model; the initial sample set comprises a plurality of initial sample images and initial label images corresponding to each initial sample image, and the initial sample images and the initial label images correspond to different modalities;
and the processing module is used for carrying out contour alignment processing on the plurality of initial sample images and the initial label image corresponding to each initial sample image to obtain a plurality of groups of training samples of the image modal transformation model.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 7 when the computer program is executed.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210151134.8A CN116645312A (en) | 2022-02-14 | 2022-02-14 | Image processing method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210151134.8A CN116645312A (en) | 2022-02-14 | 2022-02-14 | Image processing method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116645312A true CN116645312A (en) | 2023-08-25 |
Family
ID=87619288
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210151134.8A Pending CN116645312A (en) | 2022-02-14 | 2022-02-14 | Image processing method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116645312A (en) |
-
2022
- 2022-02-14 CN CN202210151134.8A patent/CN116645312A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110148192B (en) | Medical image imaging method, device, computer equipment and storage medium | |
| JP7433297B2 (en) | Deep learning-based coregistration | |
| US11120585B2 (en) | Systems and methods for image reconstruction | |
| CN109754447A (en) | Image generating method, device, equipment and storage medium | |
| CN107610193A (en) | Use the image rectification of depth production machine learning model | |
| US8861891B2 (en) | Hierarchical atlas-based segmentation | |
| CN109493417A (en) | Three-dimension object method for reconstructing, device, equipment and storage medium | |
| CN110807821A (en) | Image reconstruction method and system | |
| CN110827369B (en) | Undersampling model generation method, image reconstruction method, apparatus and storage medium | |
| CN110751187A (en) | Training method of abnormal area image generation network and related product | |
| Kläser et al. | Deep boosted regression for MR to CT synthesis | |
| Zhou et al. | Learning stochastic object models from medical imaging measurements by use of advanced ambient generative adversarial networks | |
| CN109961435B (en) | Brain image acquisition method, device, equipment and storage medium | |
| CN116228753B (en) | Tumor prognosis evaluation method, device, computer equipment and storage medium | |
| CN110570425B (en) | Pulmonary nodule analysis method and device based on deep reinforcement learning algorithm | |
| Bazrafkan et al. | Deep neural network assisted iterative reconstruction method for low dose ct | |
| CN116645312A (en) | Image processing method, device, computer equipment and storage medium | |
| CN114494484A (en) | Training method of data recognition model, data recognition method, device and equipment | |
| Patel et al. | Simulating Realistic MRI variations to Improve Deep Learning model and visual explanations using GradCAM | |
| CN115272250B (en) | Method, apparatus, computer device and storage medium for determining focus position | |
| Chen et al. | Reconstruction-aware imaging system ranking by use of a sparsity-driven numerical observer enabled by variational Bayesian inference | |
| WO2023283765A1 (en) | Method and apparatus for training machine learning models, computer device, and storage medium | |
| CN116344004B (en) | Image sample data amplification method and device | |
| US20230099906A1 (en) | Image registration method, computer device, and storage medium | |
| CN115561691B (en) | Radio frequency emission field correction method, device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |