CN114565554A - X-ray image registration method and device based on ultrasonic coronal plane image - Google Patents

X-ray image registration method and device based on ultrasonic coronal plane image Download PDF

Info

Publication number
CN114565554A
CN114565554A CN202210028320.2A CN202210028320A CN114565554A CN 114565554 A CN114565554 A CN 114565554A CN 202210028320 A CN202210028320 A CN 202210028320A CN 114565554 A CN114565554 A CN 114565554A
Authority
CN
China
Prior art keywords
image
network
spine
registration
ray
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
Application number
CN202210028320.2A
Other languages
Chinese (zh)
Inventor
姜娓娓
陈贤挺
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN202210028320.2A priority Critical patent/CN114565554A/en
Publication of CN114565554A publication Critical patent/CN114565554A/en
Pending legal-status Critical Current

Links

Images

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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
    • 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
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone
    • G06T2207/30012—Spine; Backbone

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biophysics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an X-ray image registration method and device based on an ultrasonic coronal plane image, which are used for constructing and training a spine ultrasonic coronal plane image and a spine X-ray registration network, wherein the registration network comprises a first generator network, a first space transformation network, a second generator network, a second space transformation network and a discriminator, the first generator network is used for segmenting a mask according to an input floating image, a floating image segmentation mask, a reference image and a reference image segmentation mask to generate a corresponding deformation field, and the first space transformation network acts the deformation field on the floating image and the segmentation mask thereof to obtain a registration image and a registration image segmentation mask. According to the method and the device, the registration of the X-ray images is realized on the basis of the original spine ultrasonic images and the X-ray images, so that the time spent by the patient can be reduced, and the time of exposure of the patient to radiation can also be reduced.

Description

X-ray image registration method and device based on ultrasonic coronal plane image
Technical Field
The application relates to the field of medicine and the technical field of deep learning, in particular to an X-ray image registration method and device based on an ultrasonic coronal plane image.
Background
Adolescent Idiopathic Scoliosis (AIS) is the most common form of scoliosis, defined as a three-dimensional torsional deformity of the spine and trunk. Scoliosis occurs primarily in puberty and causes lateral curvature, axial rotation, and sagittal plane normal curvature, lordosis and kyphosis. Since rapid growth of adolescents may risk developing tortuosity, adolescents with AIS should be monitored regularly, as well as discovered early and further intervened.
The standard for judging the spinal deformity is to rely on the shooting of a standing X-ray film and judge whether the spinal column is deformed or not through a clear spinal structure presented by the X-ray film. However, repeated X-ray monitoring can lead to a build-up of X-ray exposure, which can increase the risk of cancer development, particularly for juvenile children. Ultrasound is a radiationless imaging method compared to X-rays and is cost effective and real time. Therefore, the use of ultrasound scanning to determine scoliosis has been a focus of research in recent years.
In recent years, deep learning techniques are increasingly used in research and patent inventions. Among them, medical image registration is becoming an active research topic in medical imaging. On the basis of the spine ultrasound image and the X-ray image, the registration of the X-ray image is realized, which not only can reduce the time spent by the patient, but also can reduce the time of the patient exposed to radiation, and is an important research subject for those skilled in the art.
Disclosure of Invention
The method comprises the steps of constructing a registration network model, training a registration network, extracting the characteristics of a spine X-ray image and a spine ultrasonic coronal image by combining corresponding segmentation masks, generating a deformation field, and acting the generated deformation field on the spine X-ray image through a spatial transformation network to obtain a registration image.
In order to achieve the purpose, the technical scheme of the application is as follows:
an X-ray image registration method based on an ultrasonic coronal plane image comprises the following steps:
preprocessing a spine ultrasonic coronal plane image obtained by ultrasonic scanning and a spine X-ray image obtained by standing X-ray shooting to construct a training set;
constructing and training a spine ultrasonic coronal plane image and a spine X-ray registration network, wherein the registration network comprises a first generator network, a first space transformation network, a second generator network, a second space transformation network and a discriminator, and the first generator network is used for generating a corresponding deformation field according to an input floating image, a floating image segmentation mask, a reference image and a reference image segmentation mask
Figure BDA0003465298800000021
The first space transformation network transforms the deformation field
Figure BDA0003465298800000022
Acting on the floating image and its segmentation mask to obtain a registered image and its segmentation mask, and generating the registered image and its segmentation mask into a deformation field by the second generation network
Figure BDA0003465298800000023
Then inputting the image into a second space transformation network to obtain a cyclic verification image and a segmentation mask thereof;
inputting the spine ultrasonic image to be registered, the spine ultrasonic image segmentation mask, the spine X-ray image and the spine X-ray image segmentation mask into the trained spine ultrasonic coronal plane image and the trained spine X-ray registration network, and outputting the registration image and the segmentation mask thereof.
Further, the first generator network and the second generator network are the same in structure, and include: the image encoder and the image decoder fuse the characteristics of the corresponding channel number through the jump connection.
Furthermore, the first spatial transformation network and the second spatial transformation network are the same and comprise a parameter prediction network, a coordinate mapper and a pixel collector.
Further, the discriminator comprises a convolution layer, an activation function layer and a batch normalization layer.
Further, the loss function of the spine ultrasonic coronal plane image and the spine X-ray registration network is as follows:
Figure BDA0003465298800000024
wherein: x, Y, Xseg，YsegRespectively, a floating image, a reference image, a floating image division mask and a reference image division mask, GXRepresenting a first generator network, GYRepresenting a second generator network, DXPresentation discriminator, λ1Is a weight parameter;
Figure BDA0003465298800000025
Figure BDA0003465298800000031
Figure BDA0003465298800000032
Figure BDA0003465298800000033
wherein L isMIMiddle is mutual information, LSSIMFor structural similarity, LsmoothIs a regular term of the deformation field and,
Figure BDA0003465298800000034
in order to be a field of deformation,
Figure BDA0003465298800000035
in order to register the images after the deformation,
Figure BDA0003465298800000036
segmenting a mask for the deformed registered image;
Figure BDA0003465298800000037
the image and its segmentation mask are verified for the loop.
The application also provides an apparatus for registering an X-ray image based on an ultrasound coronal image, comprising a processor and a memory storing computer instructions, wherein the computer instructions, when executed by the processor, implement the steps of the method for registering an X-ray image based on an ultrasound coronal image.
According to the X-ray image registration method and device based on the ultrasonic coronal plane image, the registration of the X-ray image is realized on the basis of the original spine ultrasonic image and the X-ray image by means of a depth learning technology and a related algorithm of image registration, so that not only can the time spent by a patient be reduced, but also the time of exposure of the patient to radiation can be reduced, and the method and device have practical application values.
Drawings
FIG. 1 is a flowchart of an ultrasound coronal image-based X-ray image registration method of the present application;
FIG. 2 is a schematic view of an ultrasonic coronal image of a spine and an X-ray registration network of the present application;
FIG. 3 is a schematic diagram of a network structure generated according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a spatial transformation network according to an embodiment of the present application;
FIG. 5 is a dice curve of the registration network on the test set during training.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In one embodiment, as shown in fig. 1, the present application relates to an ultrasound coronal image-based X-ray image registration method, comprising:
and step S1, preprocessing the spine ultrasonic coronal plane image obtained by ultrasonic scanning and the spine X-ray image obtained by standing X-ray shooting to construct a training set.
The present embodiment is described by taking an ultrasound coronal image and an X-ray image of the spine as an example, and is also applicable to image registration of other parts.
The method collects the spine ultrasonic image data of the scoliosis patient, and collects the spine X-ray image after a certain time of ultrasonic detection. A total of 202 AIS patients (mean age: 16.2. + -. Standard Deviation (SD)3.9 years; BMI: 18.7. + -. 3.0 kg/m) were recruited2). Exclusion criteria were: 1. a patient with a metal implant; 2. patients who have received stent or surgical treatment; 3. BMI index higher than 25.0kg/m2The patient of (1). Prior to performing an ultrasonic examination, it is required to remove all metallic objects of the subject so as not to affect the electromagnetic space equipment. The subject stands on the platform in the desired position, wearing a gown, with the back open to the operator. The operator adjusts the hip plate, chest plate and four supports to maintain the subject in the desired posture. After applying a suitable ultrasound gel to the spinal region, the operator holds the probe against the subject's back to set system parameters, such as depth and frequency of the ultrasound. In this study, the scan covered the region of the spine from the first thoracic vertebra (T1) to the fifth lumbar vertebra (L5). The probe was placed on T1 and L5 to start the scan. Finally, the operator manipulates the probe vertically over the spinal region. During the scan, ultrasound images and spatial data are captured into a computer. Data collection will automatically stop when the probe reaches the highest point of recording. The entire scan typically takes about 30 seconds and provides a coronal image of the spine by Volume Projection Imaging (VPI) methods.
All patients received the EOS imaging system within one month after the 3D ultrasound examination (
Figure BDA0003465298800000041
Imaging, paris, france) of a posterior anterior X-ray image of a total spinal column standing.
And performing preprocessing operation on the acquired spine ultrasonic coronal image and the spine X-ray image, wherein the preprocessing operation comprises size cutting, image scaling, normalization processing and the like on the spine ultrasonic coronal image and the spine X-ray image by using a Photoshop tool. And obtaining a spine ultrasonic image with 256 × 256 pixels and a spine X-ray image training set, and obtaining a corresponding segmentation mask through segmentation.
And step S2, constructing and training a spine ultrasonic coronal plane image and a spine X-ray registration network, wherein the registration network comprises a first generator network, a first space transformation network, a second generator network, a second space transformation network and a discriminator.
Constructing a registration network for spine ultrasound and spine X-ray images, as shown in FIG. 2, the registration network comprising a network for a first generator (G)X) A first spatial transformation network (STN1), a second generator network (G)Y) A second spatial transformation network (STN2) and a discriminator (D)X). The first generator network is used for generating a corresponding deformation field according to the input floating image, the floating image segmentation mask, the reference image and the reference image segmentation mask
Figure BDA0003465298800000042
A first space transformation network for transforming the deformation field
Figure BDA0003465298800000051
And acting on the floating image and the segmentation mask thereof to obtain a registration image and a registration image segmentation mask. The division mask is simply referred to as a mask in the figure and will not be described in detail below. The second generation network generates the registered image and its segmentation mask as a deformation field
Figure BDA0003465298800000052
Then inputting the image into a second space transformation network to obtain a circular verification image
Figure BDA0003465298800000053
And its division mask
Figure BDA0003465298800000054
In the present application, the ultrasound coronal plane image and the X-ray image are registered, and the ultrasound coronal plane image may be used as a floating image and the X-ray image may be used as a reference image for registration. The X-ray image can also be used as a floating image, and the ultrasonic coronal image can be used as a reference image for registration. The X-ray image is taken as a floating image and the ultrasonic coronal plane image is taken as a reference image.
In a specific embodiment, the first generator network is structurally identical to the second generator network, and comprises: image encoder, image decoder and jump connection. As shown in fig. 3, wherein:
an image Encoder, Encoder, includes a convolution module for feature extraction and downsampling. A convolution module in the image encoder performs convolution operation on an input image by adopting four layers of convolution layers, wherein the size of a convolution kernel is (3 multiplied by 3); the convolution step of the convolution kernel is 2, the filling is 1, and the number of output channels from the first layer to the fourth layer is 32, 64, 128 and 256 respectively.
An image Decoder includes a deconvolution module for upsampling and introducing feature information of a corresponding scale into the upsampling. The deconvolution module in the image decoder is composed of four layers of deconvolution layers (convolution kernel is (3 x 3), step size is 2, filling is 1, output channel is 256, 128, 64 and 32) and one layer of convolution layer with convolution kernel is (3 x 3), step size is 1, filling is 2, output channel is 2, and activation function is Relu function.
The image encoder and the image decoder of the embodiment fuse the characteristics of the corresponding channel number through jumping connection, and the reusability of the characteristics is improved.
In a specific embodiment, the first spatial transform network and the second spatial transform network are the same, and as shown in fig. 3, include a parameter prediction network (localization Net), a coordinate mapper (Grid Generator), and a pixel collector (Sampler). Firstly, a group of matrix parameters are obtained through prediction of a parameter prediction network, then a Grid Generator maps pixel point coordinates of a target matrix to pixel point coordinates of an original image, and finally a calculation mode of a value of a target image pixel point is determined through a sampler.
In one particular embodiment, the arbiter network includes a convolution layer, an activation function layer, and a bulk normalization layer.
The discriminator is used as a judge to compare the picture generated in the generator with the real image in the data set, and the generator is restricted by the loss function of the discriminator, so that the deformation field generated by the generator is more and more accurate.
The spine ultrasound image, the spine ultrasound image segmentation mask, the spine X-ray image, and the spine X-ray image segmentation mask obtained in step S1 are input at a size of 256 × 256 pixels, and the constructed registration network is trained. In the training process, the generator continuously generates a deformation field, then generates a registration image through a space transformation network, and continuously performs countermeasure on the registration image and the reference image in the discriminator until the network model is converged, so as to obtain the registration network model after the training is completed. Finally, end-to-end registration can be carried out through the model.
In the training process, taking an X-ray image as a floating image and an ultrasonic coronal plane image as a reference image as examples, the X-ray image and a segmentation mask thereof, and the ultrasonic coronal plane image and the segmentation mask thereof are input into a first generator network to generate a deformation field
Figure BDA0003465298800000061
The first spatial transformation network then transforms the field of deformation
Figure BDA0003465298800000062
And acting on the X-ray image and the segmentation mask thereof to obtain a registration image and the segmentation mask thereof. And inputting the one path of registration image and the segmentation mask thereof into a discriminator, comparing the path of registration image with the ultrasonic coronal plane image and the segmentation mask thereof, and calculating the countermeasure loss. The other path of the registered image and the segmentation mask thereof are input into a second generator network to output a deformation field
Figure BDA0003465298800000063
And then outputting a cyclic verification image and a segmentation mask thereof through a second space transformation network, and calculating cyclic verification loss.
The total loss function of the registration network of this embodiment is:
Figure BDA0003465298800000064
wherein, X, Y, Xseg，YsegRespectively, a floating image, a reference image, a floating image division mask and a reference image division mask, GXRepresenting a first generator network, GYRepresenting a second generator network, DXPresentation discriminator, λ1Is a weight parameter.
The loss function includes a registration loss LregistLoss of cycle consistency LcycleAnd to combat the loss Ladv。
The registration loss Lregist：
Figure BDA0003465298800000065
Wherein L isMIMiddle is mutual information, LSSIMFor structural similarity, LsmoothIs a regular term of the deformation field,
Figure BDA0003465298800000071
in order to be a field of deformation,
Figure BDA0003465298800000072
in order to register the images after the deformation,
Figure BDA0003465298800000073
the mask is segmented for the deformed registered image.
The cycle consistency loss Lcycle：
Figure BDA0003465298800000074
Wherein
Figure BDA0003465298800000075
The image and its segmentation mask are verified for the loop.
The antagonistic loss Ladv：
Figure BDA0003465298800000076
By minimizing-log (D (Y, Y. Y)seg) ) and
Figure BDA0003465298800000077
Figure BDA0003465298800000078
to achieve the best registration effect.
And after calculating the loss function, optimizing the generator and the discriminator in the training by using a gradient descent method to obtain a trained network model. And testing the model in the training process, testing the effect of the registration network by taking the set similarity (dice coefficient), the structural similarity and the Jacobian as indexes, and storing the model with the best effect.
And step S3, inputting the spine ultrasonic image to be registered, the spine ultrasonic image segmentation mask, the spine X-ray image and the spine X-ray image segmentation mask into the trained spine ultrasonic coronal plane image and spine X-ray registration network, and outputting the registered image and the segmentation mask thereof.
After the registration network is trained, inputting the spine ultrasonic image, the spine ultrasonic image segmentation mask, the spine X-ray image and the spine X-ray image segmentation mask into the trained spine ultrasonic coronal plane image and the spine X-ray registration network, and outputting a registration image.
To verify the validity of the network model, 10 sets of test images (not in the training set) were selected for verification. FIG. 5 is a plot of the mean value of the dice coefficients on the validation set during training. The higher the value of the dice is, the better the registration effect is, and as can be seen from fig. 5, the dice coefficient is improved to some extent through the registration network, which indicates that the registration network has a better effect.
In one embodiment, the present application further provides an ultrasound coronal image-based X-ray image registration apparatus, comprising a processor and a memory storing computer instructions, which when executed by the processor, implement the steps of the ultrasound coronal image-based X-ray image registration method.
For specific definition of the apparatus for registering X-ray image based on ultrasound coronal image, reference may be made to the above definition of the method for registering X-ray image based on ultrasound coronal image, and details are not repeated here. The above-mentioned X-ray image registration apparatus based on the ultrasound coronal plane image can be implemented in whole or in part by software, hardware and a combination thereof. The method can be embedded in hardware or independent from a processor in the computer device, and can also be stored in a memory in the computer device in software, so that the processor can call and execute the corresponding operation.
The memory and the processor are electrically connected, directly or indirectly, to enable transmission or interaction of data. For example, the components may be electrically connected to each other via one or more communication buses or signal lines. The memory stores a computer program that can be executed on the processor, and the processor executes the computer program stored in the memory, thereby implementing the network topology layout method in the embodiment of the present invention.
The Memory may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Read-Only Memory (EPROM), an electrically Erasable Read-Only Memory (EEPROM), and the like. The memory is used for storing programs, and the processor executes the programs after receiving the execution instructions.
The processor may be an integrated circuit chip having data processing capabilities. The Processor may be a general-purpose Processor including a Central Processing Unit (CPU), a Network Processor (NP), and the like. The various methods, steps and logic blocks disclosed in embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (6)

1. An X-ray image registration method based on an ultrasonic coronal image is characterized in that the X-ray image registration method based on the ultrasonic coronal image comprises the following steps:
preprocessing a spine ultrasonic coronal plane image obtained by ultrasonic scanning and a spine X-ray image obtained by standing X-ray shooting to construct a training set;
constructing and training a spine ultrasonic coronal plane image and a spine X-ray registration network, wherein the registration network comprises a first generator network, a first space transformation network, a second generator network, a second space transformation network and a discriminator, and the first generator network is used for generating a corresponding deformation field according to an input floating image, a floating image segmentation mask, a reference image and a reference image segmentation mask
Figure FDA0003465298790000011
The first space transformation network transforms the deformation field
Figure FDA0003465298790000012
Acting on the floating image and its segmentation mask to obtain a registered image and its segmentation mask, and generating the registered image and its segmentation mask into a deformation field by the second generation network
Figure FDA0003465298790000013
Then inputting the image into a second space transformation network to obtain a cyclic verification image and a segmentation mask thereof;
inputting the spine ultrasonic image to be registered, the spine ultrasonic image segmentation mask, the spine X-ray image and the spine X-ray image segmentation mask into the trained spine ultrasonic coronal plane image and the trained spine X-ray registration network, and outputting the registration image and the segmentation mask thereof.
2. The method of claim 1, wherein the first generator network is structurally identical to the second generator network, and comprises: the image encoder and the image decoder fuse the characteristics of the corresponding channel number through the jump connection.
3. The method of claim 1, wherein the first and second spatial transformation networks are the same and comprise a parameter prediction network, a coordinate mapper, and a pixel grabber.
4. The method of claim 1, wherein the discriminator comprises a convolution layer, an activation function layer and a batch normalization layer.
5. The method of claim 1, wherein the loss function of the spine ultrasound coronal image and the spine X-ray registration network is:
Figure FDA0003465298790000021
wherein: x, Y, Xseg，YsegRespectively, a floating image, a reference image, a floating image division mask and a reference image division mask, GXRepresenting a first generator network, GYRepresenting a second generator network, DXPresentation discriminator, λ1Is a weight parameter;
Figure FDA0003465298790000022
Figure FDA0003465298790000023
Figure FDA0003465298790000024
wherein L isMIMiddle is mutual information, LSSIMFor structural similarity, LsmoothIs a regular term of the deformation field,
Figure FDA0003465298790000025
in order to be a field of deformation,
Figure FDA0003465298790000026
in order to register the images after the deformation,
Figure FDA0003465298790000027
segmenting a mask for the deformed registered image;
Figure FDA0003465298790000028
the image and its segmentation mask are verified for the loop.
6. An apparatus for ultrasound coronal image based X-ray image registration, comprising a processor and a memory storing computer instructions, wherein the computer instructions, when executed by the processor, implement the steps of the method of any one of claims 1 to 5.
CN202210028320.2A 2022-01-11 2022-01-11 X-ray image registration method and device based on ultrasonic coronal plane image Pending CN114565554A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210028320.2A CN114565554A (en) 2022-01-11 2022-01-11 X-ray image registration method and device based on ultrasonic coronal plane image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210028320.2A CN114565554A (en) 2022-01-11 2022-01-11 X-ray image registration method and device based on ultrasonic coronal plane image

Publications (1)

Publication Number Publication Date
CN114565554A true CN114565554A (en) 2022-05-31

Family

ID=81712507

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210028320.2A Pending CN114565554A (en) 2022-01-11 2022-01-11 X-ray image registration method and device based on ultrasonic coronal plane image

Country Status (1)

Country Link
CN (1) CN114565554A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115908515A (en) * 2022-11-11 2023-04-04 北京百度网讯科技有限公司 Image registration method, and training method and device of image registration model
CN116228796A (en) * 2023-05-05 2023-06-06 北京大学第三医院（北京大学第三临床医学院） CT image pedicle segmentation method, device and storage medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115908515A (en) * 2022-11-11 2023-04-04 北京百度网讯科技有限公司 Image registration method, and training method and device of image registration model
CN115908515B (en) * 2022-11-11 2024-02-13 北京百度网讯科技有限公司 Image registration method, training method and device of image registration model
CN116228796A (en) * 2023-05-05 2023-06-06 北京大学第三医院（北京大学第三临床医学院） CT image pedicle segmentation method, device and storage medium

Similar Documents

Publication Publication Date Title
EP4095797B1 (en) Autonomous segmentation of three-dimensional nervous system structures from medical images
CN110047128A (en) The method and system of X ray CT volume and segmentation mask is rebuild from several X-ray radiogram 3D
EP2186060B1 (en) System and method for adapting a movement model based on an image
CN111507381A (en) Image recognition method and related device and equipment
CN114565554A (en) X-ray image registration method and device based on ultrasonic coronal plane image
CN110310723A (en) Bone image processing method, electronic equipment and storage medium
KR20180044421A (en) Systems and methods for medical imaging of patients with medical implants for use in orthodontic surgery planning
CN113538533B (en) Spine registration method, device and equipment and computer storage medium
EP4191528A1 (en) Methods and systems for image registration
JP5016518B2 (en) Alignment apparatus and program thereof
Bagheri et al. Reliability of three-dimensional spinal modeling of patients with idiopathic scoliosis using EOS system
KR20160001316A (en) Fabrication method of patient customized Implant using 3D imaging technology
KR102204309B1 (en) X-ray Image Display Method Based On Augmented Reality
CN109064472A (en) A kind of approximating method and device of the three-dimensional space model fit Plane of vertebrae
US11704796B2 (en) Estimating bone mineral density from plain radiograph by assessing bone texture with deep learning
CN114145761A (en) Fluorine bone disease medical imaging detection system and use method thereof
CN113243914A (en) Brain image-based assessment method and neuropsychiatric disease assessment system
CN114511507A (en) Non-radiation X-ray image generation method and device based on ultrasonic coronal plane image
CN115844436A (en) CT scanning scheme self-adaptive formulation method based on computer vision
Oulbacha et al. MRI to C‐arm spine registration through Pseudo‐3D CycleGANs with differentiable histograms
US11803970B2 (en) Image judgment device, image judgment method, and storage medium
EP2343687B1 (en) Tomographic image generating apparatus, tomographic image generating method, and program for generating tomographic images
CN117690596B (en) Organ tolerance dose determination device, electronic apparatus, and storage medium
US20190095579A1 (en) Biomechanical model generation for human or animal torsi
CN109414233B (en) Statistical breast model generation and personalization

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