CN110246216A - Spine model generation method, spine model generate system and terminal - Google Patents
Spine model generation method, spine model generate system and terminal Download PDFInfo
- Publication number
- CN110246216A CN110246216A CN201910450859.5A CN201910450859A CN110246216A CN 110246216 A CN110246216 A CN 110246216A CN 201910450859 A CN201910450859 A CN 201910450859A CN 110246216 A CN110246216 A CN 110246216A
- Authority
- CN
- China
- Prior art keywords
- image
- spine
- lesion
- model
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 61
- 230000003902 lesion Effects 0.000 claims abstract description 119
- 238000012549 training Methods 0.000 claims abstract description 73
- 230000008439 repair process Effects 0.000 claims abstract description 38
- 230000006870 function Effects 0.000 claims description 31
- 238000004590 computer program Methods 0.000 claims description 22
- 239000011159 matrix material Substances 0.000 claims description 19
- 230000011218 segmentation Effects 0.000 claims description 16
- 238000003860 storage Methods 0.000 claims description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000002591 computed tomography Methods 0.000 claims description 6
- 230000004069 differentiation Effects 0.000 claims description 5
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 238000004519 manufacturing process Methods 0.000 abstract description 10
- 230000000694 effects Effects 0.000 abstract description 3
- 239000003814 drug Substances 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 21
- 238000005070 sampling Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000002513 implantation Methods 0.000 description 5
- 239000000463 material Substances 0.000 description 5
- 238000010146 3D printing Methods 0.000 description 4
- 206010028980 Neoplasm Diseases 0.000 description 4
- 238000003709 image segmentation Methods 0.000 description 4
- 238000002271 resection Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000037081 physical activity Effects 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 208000001613 Gambling Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 210000003041 ligament Anatomy 0.000 description 1
- 206010025482 malaise Diseases 0.000 description 1
- 230000007659 motor function Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000002980 postoperative effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 210000004872 soft tissue Anatomy 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- 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
- 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
- G06T2207/20—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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
The application is suitable for engineering in medicine technical field, provides a kind of spine model generation method, spine model generates system and terminal, and wherein method includes: according to non-lesion spine image, and training generates confrontation network model；Network model is fought by the generation after training, image repair is carried out to lesion site in lesion spine image, obtains targets spine image, wherein lesion site described in the targets spine image is repaired as non-lesion spine form；Three-dimensional reconstruction is carried out to the targets spine image, obtains corresponding targets spine model, promotes the precision of spinal prostheses production, the activity limitation after reducing Using prosthesis patient's body.
Description
Technical field
The application belongs to engineering in medicine technical field more particularly to a kind of spine model generation method, spine model generate
System and terminal.
Background technique
Full backbone resection (TES) refers to complete resection one or more spinal segments, including centrum, interverbebral disc, vertebral arch
Root, vertebral plate, spinous process, ligament and the complete resection of operation together of the attached soft tissue of other backbones.The operation be at present in the world
The minimum modus operandi of generally acknowledged treatment tumor of spine recurrence rate.
But with the completion of operation, the backbone of human body, which is influenced continuity by operation, to be interrupted.To realize spinal stability
Reconstruction, the method currently mainly used be by 3D printing technique manufacturing artificial centrum, and by the Using prosthesis produced suffer from
In person's body, human vertebra function is realized to substitute original backbone position.Therefore the design and production of prosthese are for patients
It is extremely important.
And in the prior art, the preoperative cast of implantation material needed for generally requiring production dozens of 3D printing, those arts
The foundation of preceding model is checked on the determining clinician for needing to be rich in experience and is just able to achieve.And it is limited to the influence of the prior art,
There are biggish errors for the prosthese and ideal prosthetic produced, and cause patient's motor function after being implanted into prosthese limited, postoperative ridge
The problems such as vertebra is inadequate in the stability of axial-rotation, low with the bond strength of upper hypocentrum occurs.
Summary of the invention
In view of this, the embodiment of the present application provides a kind of spine model generation method, spine model generates system and end
End, to solve the preoperative cast of implantation material needed for needing to make dozens of 3D printing in the prior art, and the vertebra produced
There is biggish error in prosthese and ideal prosthetic.
The first aspect of the embodiment of the present application provides a kind of spine model generation method, comprising:
According to non-lesion spine image, training generates confrontation network model；
Network model is fought by the generation after training, image is carried out to lesion site in lesion spine image
It repairs, obtains targets spine image, wherein lesion site described in the targets spine image is repaired as non-lesion ridge
Cylindricality state；
Three-dimensional reconstruction is carried out to the targets spine image, obtains corresponding targets spine model.
The second aspect of the embodiment of the present application provides a kind of spine model generation system, comprising:
Model training module, for according to non-lesion spine image, training to generate confrontation network model；
Image repair module, for fighting network model by the generation after training, to sick in lesion spine image
Become happening part and carry out image repair, obtains targets spine image, wherein lesion site quilt described in the target image
Repairing is non-lesion spine form；
Model building module obtains corresponding targets spine mould for carrying out three-dimensional reconstruction to the targets spine image
Type.
The third aspect of the embodiment of the present application provides a kind of terminal, including memory, processor and is stored in described
In memory and the computer program that can run on the processor, the processor are realized when executing the computer program
The step of method as described in relation to the first aspect.
The fourth aspect of the embodiment of the present application provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer program, and the step of method as described in relation to the first aspect is realized when the computer program is executed by processor
Suddenly.
The 5th aspect of the application provides a kind of computer program product, and the computer program product includes computer
Program is realized when the computer program is executed by one or more processors such as the step of above-mentioned first aspect the method.
Therefore in the embodiment of the present application, by being based on non-lesion spine image, training generates confrontation network model,
Network model is fought using the generation after training, image repair is carried out to lesion site in lesion spine image, by lesion
Happening part reparation is non-lesion spine form, obtains targets spine image, carries out three-dimensional reconstruction to targets spine image, obtains
Final targets spine model, the process pass through using confrontation network is generated, restore to lesion spine image, with as far as possible
A vertebral model ideally is obtained, the reconstruction of as a normal as possible, complete centrum is realized, avoids needing at present
The problem of making the three-dimensional preoperative cast of dozens of implantation material promotes the precision of spinal prostheses production, reduces Using prosthesis and suffers from
Activity limitation after in person's body.
Detailed description of the invention
It in order to more clearly explain the technical solutions in the embodiments of the present application, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only some of the application
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is a kind of flow chart one of spine model generation method provided by the embodiments of the present application；
Fig. 2 is a kind of flowchart 2 of spine model generation method provided by the embodiments of the present application；
Fig. 3 is the backbone mask image in the embodiment of the present application；
Fig. 4 is the targets spine image after the reparation in the embodiment of the present application；
Fig. 5 is to carry out the two of the spine regions that semantic segmentation obtains based on V-Net neural network in the embodiment of the present application
It is worth image；
Fig. 6 is the structure chart that a kind of spine model provided by the embodiments of the present application generates system；
Fig. 7 is a kind of structure chart of terminal provided by the embodiments of the present application.
Specific embodiment
In being described below, for illustration and not for limitation, the tool of such as particular system structure, technology etc is proposed
Body details, so as to provide a thorough understanding of the present application embodiment.However, it will be clear to one skilled in the art that there is no these specific
The application also may be implemented in the other embodiments of details.In other situations, it omits to well-known system, device, electricity
The detailed description of road and method, so as not to obscure the description of the present application with unnecessary details.
It should be appreciated that ought use in this specification and in the appended claims, term " includes " instruction is described special
Sign, entirety, step, operation, the presence of element and/or component, but be not precluded one or more of the other feature, entirety, step,
Operation, the presence or addition of element, component and/or its set.
It is also understood that mesh of the term used in this present specification merely for the sake of description specific embodiment
And be not intended to limit the application.As present specification and it is used in the attached claims, unless on
Other situations are hereafter clearly indicated, otherwise " one " of singular, "one" and "the" are intended to include plural form.
It will be further appreciated that the term "and/or" used in present specification and the appended claims is
Refer to any combination and all possible combinations of one or more of associated item listed, and including these combinations.
As used in this specification and in the appended claims, term " if " can be according to context quilt
Be construed to " when ... " or " once " or " in response to determination " or " in response to detecting ".Similarly, phrase " if it is determined that " or
" if detecting [described condition or event] " can be interpreted to mean according to context " once it is determined that " or " in response to true
It is fixed " or " once detecting [described condition or event] " or " in response to detecting [described condition or event] ".
In the specific implementation, terminal described in the embodiment of the present application is including but not limited to such as with touch sensitive surface
The mobile phone, laptop computer or tablet computer of (for example, touch-screen display and/or touch tablet) etc it is other just
Portable device.It is to be further understood that in certain embodiments, the equipment is not portable communication device, but there is touching
Touch the desktop computer of sensing surface (for example, touch-screen display and/or touch tablet).
In following discussion, the terminal including display and touch sensitive surface is described.It is, however, to be understood that
It is that terminal may include one or more of the other physical user-interface device of such as physical keyboard, mouse and/or control-rod.
Terminal supports various application programs, such as one of the following or multiple: drawing application program, demonstration application journey
Sequence, word-processing application, website create application program, disk imprinting application program, spreadsheet applications, game application
Program, telephony application, videoconference application, email application, instant messaging applications, exercise
Support application program, photo management application program, digital camera application program, digital camera application program, web-browsing application
Program, digital music player application and/or video frequency player application program.
The various application programs that can be executed at the terminal can be used such as touch sensitive surface at least one is public
Physical user-interface device.It can adjust and/or change among applications and/or in corresponding application programs and touch sensitive table
The corresponding information shown in the one or more functions and terminal in face.In this way, the public physical structure of terminal is (for example, touch
Sensing surface) it can support the various application programs with user interface intuitive and transparent for a user.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in the present embodiment, each process
Execution sequence should be determined by its function and internal logic, and the implementation process without coping with the embodiment of the present application constitutes any restriction.
In order to illustrate technical solution described herein, the following is a description of specific embodiments.
It is a kind of flow chart one of spine model generation method provided by the embodiments of the present application referring to Fig. 1, Fig. 1.Such as Fig. 1 institute
Show, a kind of spine model generation method, method includes the following steps:
Step 101, according to non-lesion spine image, training generates confrontation network model.
The non-lesion spine image is based on image obtained from the backbone that lesion does not occur.Non-lesion spine image tool
Body is the bianry image of a spine regions, or can be based on the non-lesion spine image, and auxiliary, which is realized, generates confrontation network mould
The other types image of type training, for example, gray level image.
Specifically, generation confrontation network model includes two network models: generating network model and differentiates network model.
Wherein, the network that network is a production is generated, it receives a random noise, generates image by this noise, defeated
Image G out；Differentiate that network is used to judge whether true a picture is, its input x is a picture, and output D (x) indicates
X is that the probability of true picture means that x is a true picture if it is 1, and exporting is 0, means that and is unlikely to be true
Real picture.During model training, the target for generating network is just to try to generate true picture and go to cheat to differentiate network,
And differentiate that the target of network is just to try to determine and generate picture and true image that network generates, such G and D constitute one
A dynamic " gambling process ", final equalization point, that is, Nash equilibrium point.
Generation confrontation network model is exactly by generating network and differentiation for being applied in deep learning neural network
The continuous game of network generates, training is completed so that generating the distribution of e-learning to data if using picture
Afterwards, image true to nature can be generated from one section of random number by generating network.
Step 102, network model is fought by the generation after training, to lesion site in lesion spine image
Image repair is carried out, targets spine image is obtained.
Wherein, lesion site described in the target image is repaired as non-lesion spine form.
Specifically, it is generally the case that in lesion spine image, lesion site can be influenced to make by sick cells such as tumours
Script natural spine area information in spine image is lost, such as backbone profile is imperfect, spine regions abnormal deformation etc..
Those image-regions need to carry out image repair, to obtain the targets spine image under natural spine state.
In this step, network model is fought by the generation after the training of non-lesion spine image, by lesion spine image
It is repaired, is repaired targets spine image for non-lesion spine form to obtain lesion site, with can be to lesion
Spine image is restored, and the image after recovery is applied to correspond to the three-dimension modeling at backbone position after full backbone resection
In the process, to realize manufacturing for ideal spinal prostheses, the dependence to human factor is reduced, promotes the essence of spinal prostheses production
Accuracy, the malaise symptoms after reducing Using prosthesis patient's body.
When specific implementation, how the realization process of image repair is utilized it is to be understood that dig a hole on piece image
This hole completion is allowed human eye that can not identify the part of completion by other information.This problem is easy to the mankind,
But but seem addition difficult for computer, secondly how this problem first utilizes other either with or without the solution uniquely determined
Information, there are also whether the result for how judging completion true enough.
In the embodiment of the present application, as an optional embodiment, wherein fight net by the generation after training
Network model carries out image repair to lesion site in lesion spine image, obtains targets spine image, comprising:
Based on the lesion spine image and preset two-value mask code matrix, generates and covered in lesion site addition
The backbone mask image of code.
It is repaired based on the backbone mask image and backbone referring to image, network mould is fought by the generation after training
Type generates the targets spine image F '=M ⊙ F+ (1-M) ⊙ G (z) after repairing*；
Wherein, F is the lesion spine image；M is the preset two-value mask code matrix, in the two-value mask code matrix
Element it is corresponding with the pixel of the lesion spine image；M ⊙ F is the backbone mask image；G(z)*It is repaired for the backbone
The multiple symbol generated during model training referring to image for the generation confrontation network model referring to image, the backbone reparation
Setting is closed referring to the image of condition；⊙ is the operation that matrix element is multiplied with pixel point value in image.
The process, firstly, defining several symbols for image repair.M indicates the mask code matrix of a two-value, therein
Numerical value only has 0 or 1；The region for needing to repair in 0 one picture of expression, 1 indicates the part to be retained in same picture.
Then, we define how to repair lesion spine image F by given mask, and specific practice is exactly to allow Metzler matrix
In element be multiplied with the corresponding pixel point value in F, the method that this matrix is multiplied with respective pixel point value is called Harbin-to-Dalian horse
Product is indicated with M ⊙ F, and then on the basis of lesion spine image, obtains being added to mask in lesion site wherein
Backbone mask image, as shown in Figure 3.Next, we are from confrontation network model is generated during model training, generator is raw
At result images in find one meet setting referring to condition picture G (z)*, as enterprising in backbone mask image basis
Backbone when row image repair is repaired referring to image.
We can define targets spine image F '=M ⊙ F+ (1-M) ⊙ G (z) after repairing*。
During before, we fight network model (specially WGAN-GP network according to trained generation is obtained
Model), the result i.e. targets spine image using the backbone mask image M ⊙ F after addition mask as its input, after being repaired
F ', as shown in Figure 4.
Wherein, this meets setting referring to the picture G (z) of condition*Network model is fought in model training mistake for aforementioned generation
In the image generated in journey between image F an as far as possible similar optimal figure of the corresponding pixel points in the distribution of value, i.e.,
Similarity exceeds an optimal figure of threshold value.
Image G (z)*In contain spine regions corresponding with lesion site in lesion spine image, and remove the ridge
Other regions except columnar region；(1-M)⊙G(z)*It has then found out image G (z)*In other regions add mask, retain with
The backbone of the corresponding spine regions part of lesion site is repaired referring to mask image in lesion spine image.M ⊙ F is represented
Backbone mask image and (1-M) ⊙ G (z)*The backbone of representative is repaired to be superimposed referring to mask image pixel, after obtaining the reparation
Targets spine image F '.
It further, is realization to optimal figure G (z)*Determination, it is described based on described as an optional embodiment
Backbone mask image and backbone are repaired referring to image, are fought network model by the generation after training, are generated after repairing
Before the targets spine image, further includes:
Obtain the output image for generating confrontation network model and generating during model training；
According to the output image, according to loss function L=Lcontextual+λLperceptual, obtained from the output image
The backbone is taken to repair referring to image；
Wherein, λ is hyper parameter；
Lcontextual=| | M ⊙ G (z)-M ⊙ F | |1, G (z) is the output image；
Lperceptual=log (1-D (G (z))), D (G (z)) be training after the generation confrontation network model in differentiate
Differentiation end value of the device to the output image；
The backbone reparation is referring to the output that image is when the value of the loss function meets sets requirement
Image.
Specifically, this, which meets sets requirement specifically, may is that the variation trend of the value of the loss function meets sets requirement,
The variation trend of for example, value of the loss function is the trend that approach variation is carried out towards a target value, is specially in court
The variation trend of the trend for carrying out approach variation to 0, the i.e. value of the loss function is downward trend.Or this meets sets requirement
It specifically may is that the value of the loss function is in setting range, for example, close in 0 numberical range.
From formula F '=M ⊙ F+ (1-M) ⊙ G (z), and generation confrontation network model is combined to generate during model training
Output image G (z) the fact that have multiple situation, it will be seen that most importantly to the targets spine image after reparation
Section 2 generating portion, that is, need to realize good image repair loss zone algorithm.Find out the optimum image G in G (z)
(z)*。
In order to realize the purpose, it would be desirable to two important informations: contextual information and perception information, the two information
Acquisition mainly by loss function realize, loss function is smaller, and the result of image repair is better.
To guarantee the identical contextual information of input picture, picture F and pixel corresponding in G (z) are similar as far as possible, because
This needs to make punishment, loss function to the G (z) for generating dissimilar pixel are as follows:
Lcontextual=| | M ⊙ G (z)-M ⊙ F | |1；
The most ideal situation is that all pixels value of F and G (z) be it is identical, loss function value be 0.
In order to obtain the reparation picture of high quality, it would be desirable to allow arbiter that there is the ability for correctly differentiating true picture,
Loss function are as follows:
Lperceptual=log (1-D (G (z)))；
Therefore final loss function are as follows:
L=Lcontextual+λLperceptual；
Wherein, λ is set in advance, is a hyper parameter；D(G(z))∈[0,1].
In view of the loss in the reality and image information controlled range in image procossing, when specific implementation, backbone reparation
Referring to the output image that image is when the value of the loss function is in setting range；The setting range be close to
0 range.
Step 103, three-dimensional reconstruction is carried out to the targets spine image, obtains corresponding targets spine model.
The targets spine image is a bianry image.Finally, the targets spine image after reparation is subjected to three-dimensional reconstruction,
Normal spine model is obtained, which can perform the operation by 3D printing technique for TES.
In the embodiment of the present application, by being based on non-lesion spine image, training generates confrontation network model, after training
Generation fight network model, in lesion spine image lesion site carry out image repair, lesion site is repaired
It is again non-lesion spine form, obtains targets spine image, three-dimensional reconstruction is carried out to targets spine image, obtains final target
Spine model, the process pass through using confrontation network is generated, restore to lesion spine image, to obtain a reason as far as possible
Think the vertebral model under state, realizes the reconstruction of as a normal as possible, complete centrum, avoid needing to make dozens of at present
The problem of three-dimensional preoperative cast of implantation material, promotes the precision of spinal prostheses production, after reducing Using prosthesis patient's body
Activity limitation.
The different embodiments of spine model generation method are additionally provided in the embodiment of the present application.
Referring to fig. 2, Fig. 2 is a kind of flowchart 2 of spine model generation method provided by the embodiments of the present application.Such as Fig. 2 institute
Show, a kind of spine model generation method, method includes the following steps:
Step 201, semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, obtains spine regions
First bianry image.
Firstly, carrying out semantic segmentation to non-lesion spine CT image by neural network, acquisition can clearly distinguish spinal area
The binary map in domain and non-spinal region.Specifically, spine regions gray value is 255 in first bianry image, non-spinal region
Gray value is 0.
Specifically, semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, can be based on V-Net
The spine CT image segmentation that neural network carries out.
V-Net is a typical network structure end to end, aiming at the problem that be three-dimensional image segmentation.Here, we
It is used to carry out two dimensional image segmentation.When carrying out non-lesion spine CT image segmentation using V-Net neural network, by non-disease
The two-dimensional image information for becoming spine CT image is inputted, and depth information is ignored.
Specifically, in V-Net neural network structure here, left side is the process of a down-sampling, for capturing up and down
Literary information, point 5 groups of convolution operations carry out, and similar pooling is carried out once after every group of convolution operation and is operated, is specifically exactly to walk
The convolution kernel for the 2*2 that width is 2 carries out deconvolution, is original 1/2 by image down.By four groups of operations by size be 512*
512 image becomes the image that size is 32*32.Right side in structure is upper sampling process, is come in conjunction with each layer information of down-sampling
Detailed information is restored, using 4 groups of deconvolution, picture is extended to original 2 times by up-sampling every time, then by corresponding characteristic pattern
It is replicated and the result of left side down-sampling is linked, the figure of 512*512*16 size is obtained after upper sampling process, most
Port number is reduced to 1 with the convolution kernel of a 1*1 afterwards.Do not have to change the size of image when making convolution during down-sampling, this
The benefit that sample is done is omitted the work of cutting because the size of respective layer be it is the same, convolution operation can be retained in this way and mentioned
The all information got.First bianry image of the spine regions obtained after segmentation is as shown in Figure 5.
Step 202, it is based on first bianry image, the training generation fights network model.
The bianry image of the natural spine CT image of acquisition is put into generation confrontation network to be trained, with trained life
Image repair is carried out to tumor of spine happening part at confrontation network model, obtains the natural spine bianry image at the position.
Specifically, when obtaining the bianry image of non-lesion spine CT image, it is possible that spine regions phase in image
Than the situation too small in background area, therefore first bianry image can also be advanced optimized, by the first bianry image into
Row is cut, and the image after cutting is amplified, first bianry image after being optimized, and then training generates confrontation network
Model.It specifically can be, the image that initial size is 512*512 cut and scaled, 64*64 is become；By treated
Image is put into WGAN-GP and is trained, and finally obtains training pattern, training pattern meeting output model training result, i.e., aforementioned reality
It applies and generates the image that confrontation network model generates during model training in mode.
Step 201 and step 202 realize that, according to non-lesion spine image, training generates the implementation of confrontation network model
Journey.
Step 203, semantic segmentation is carried out to the CT image of lesion backbone, obtains the second bianry image of spine regions.
In the step, it is accomplished that the processing to the CT image of lesion vertebra, generates the second bianry image of spine regions.
This based on CT image carry out semantic segmentation obtain spine regions bianry image treatment process with reference can be made to step 201 in non-
The CT image of lesion backbone carries out semantic segmentation, obtains the treatment process of the bianry image of spine regions, details are not described herein again.
Step 204, determine that second bianry image is the lesion spine image.
When fighting network model progress image repair by the generation after training, the image of reparation need to be a binary map
Therefore picture in this step, by semantic segmentation and handles obtained the second bianry image of spine regions and is determined as lesion backbone
Image is prepared for subsequent image reparation.
Step 205, network model is fought by the generation after training, to lesion site in lesion spine image
Image repair is carried out, targets spine image is obtained.
The realization process of the step and the realization process of the step 102 in aforementioned embodiments are identical, and details are not described herein again.
Step 206, three-dimensional reconstruction is carried out to the targets spine image by marching cubes algorithm, obtains corresponding institute
State targets spine model.
Here, three-dimensional reconstruction is carried out to the tumor of spine bianry image after reparation using marching cubes algorithm, obtained most
Whole targets spine model, technology are easy to accomplish.
In the embodiment of the present application, by being based on non-lesion spine image, training generates confrontation network model, after training
Generation fight network model, in lesion spine image lesion site carry out image repair, lesion site is repaired
It is again non-lesion spine form, obtains targets spine image, three-dimensional reconstruction is carried out to targets spine image, obtains final target
Spine model, the process pass through using confrontation network is generated, restore to lesion spine image, to obtain a reason as far as possible
Think the vertebral model under state, realizes the reconstruction of as a normal as possible, complete centrum, avoid needing to make dozens of at present
The problem of three-dimensional preoperative cast of implantation material, promotes the precision of spinal prostheses production, after reducing Using prosthesis patient's body
Activity limitation.
It is the structure chart that Fig. 6 is a kind of spine model generation system provided by the embodiments of the present application, in order to just referring to Fig. 6
In explanation, part relevant to the embodiment of the present application is illustrated only.
It includes: model training module 301, image repair module 302 and model foundation that the spine model, which generates system 300,
Module 303.
Model training module 301, for according to non-lesion spine image, training to generate confrontation network model；
Image repair module 302, for fighting network model by the generation after training, in lesion spine image
Lesion site carries out image repair, obtains targets spine image, wherein lesion site described in the target image
It is repaired as non-lesion spine form；
Model building module 303 obtains corresponding targets spine for carrying out three-dimensional reconstruction to the targets spine image
Model.
Wherein, the model training module 301 is specifically used for:
Semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, obtains the first two-value of spine regions
Image；
Based on first bianry image, the training generation fights network model.
Described image repair module is specifically used for:
Based on the lesion spine image and preset two-value mask code matrix, generates and covered in lesion site addition
The backbone mask image of code；
It is repaired based on the backbone mask image and backbone referring to image, network mould is fought by the generation after training
Type generates the targets spine image F '=M ⊙ F+ (1-M) ⊙ G (z) after repairing*；
Wherein, F is the lesion spine image；M is the preset two-value mask code matrix, in the two-value mask code matrix
Element it is corresponding with the pixel of the lesion spine image；M ⊙ F is the backbone mask image；G(z)*It is repaired for the backbone
The multiple symbol generated during model training referring to image for the generation confrontation network model referring to image, the backbone reparation
Setting is closed referring to the image of condition；⊙ is the operation that matrix element is multiplied with pixel point value in image.
Described image repair module is also used to:
Obtain the output image for generating confrontation network model and generating during model training；
According to the output image, according to loss function L=Lcontextual+λLperceptual, obtained from the output image
The backbone is taken to repair referring to image；
Wherein, λ is hyper parameter；
Lcontextual=| | M ⊙ G (z)-M ⊙ F | |1, G (z) is the output image；
Lperceptual=log (1-D (G (z))), D (G (z)) be training after the generation confrontation network model in differentiate
Differentiation end value of the device to the output image；
The backbone reparation is referring to the output that image is when the value of the loss function meets sets requirement
Image.
Spine model generates system further include:
Module is obtained, semantic segmentation is carried out for the CT image to lesion backbone, obtains the second binary map of spine regions
Picture；
Determining module, for determining that second bianry image is the lesion spine image.
The model building module 303 is specifically used for:
Three-dimensional reconstruction is carried out to the targets spine image by marching cubes algorithm, obtains the corresponding target ridge
Column model.
Spine model provided by the embodiments of the present application, which generates system, can be realized the implementation of above-mentioned spine model generation method
Each process of example, and identical technical effect can be reached, to avoid repeating, which is not described herein again.
Fig. 7 is a kind of structure chart of terminal provided by the embodiments of the present application.As shown in Fig. 7, the terminal 4 of the embodiment is wrapped
It includes: processor 40, memory 41 and being stored in the computer that can be run in the memory 41 and on the processor 40
Program 42.
Illustratively, the computer program 42 can be divided into one or more module/units, it is one or
Multiple module/units are stored in the memory 41, and are executed by the processor 40, to complete the application.Described one
A or multiple module/units can be the series of computation machine program instruction section that can complete specific function, which is used for
Implementation procedure of the computer program 42 in the terminal 4 is described.For example, the computer program 42 can be divided into
Model training module, image repair module, model building module, acquisition module and determining module, each module concrete function are as follows:
Model training module, for according to non-lesion spine image, training to generate confrontation network model；
Image repair module, for fighting network model by the generation after training, to sick in lesion spine image
Become happening part and carry out image repair, obtains targets spine image, wherein lesion site quilt described in the target image
Repairing is non-lesion spine form；
Model building module obtains corresponding targets spine mould for carrying out three-dimensional reconstruction to the targets spine image
Type.
Wherein, the model training module is specifically used for:
Semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, obtains the first two-value of spine regions
Image；
Based on first bianry image, the training generation fights network model.
Described image repair module is specifically used for:
Based on the lesion spine image and preset two-value mask code matrix, generates and covered in lesion site addition
The backbone mask image of code；
It is repaired based on the backbone mask image and backbone referring to image, network mould is fought by the generation after training
Type generates the targets spine image F '=M ⊙ F+ (1-M) ⊙ G (z) after repairing*；
Wherein, F is the lesion spine image；M is the preset two-value mask code matrix, in the two-value mask code matrix
Element it is corresponding with the pixel of the lesion spine image；M ⊙ F is the backbone mask image；G(z)*It is repaired for the backbone
The multiple symbol generated during model training referring to image for the generation confrontation network model referring to image, the backbone reparation
Setting is closed referring to the image of condition；⊙ is the operation that matrix element is multiplied with pixel point value in image.
Described image repair module is also used to:
Obtain the output image for generating confrontation network model and generating during model training；
According to the output image, according to loss function L=Lcontextual+λLperceptual, obtained from the output image
The backbone is taken to repair referring to image；
Wherein, λ is hyper parameter；
Lcontextual=| | M ⊙ G (z)-M ⊙ F | |1, G (z) is the output image；
Lperceptual=log (1-D (G (z))), D (G (z)) be training after the generation confrontation network model in differentiate
Differentiation end value of the device to the output image；
The backbone reparation is referring to the output that image is when the value of the loss function meets sets requirement
Image.
Module is obtained, semantic segmentation is carried out for the CT image to lesion backbone, obtains the second binary map of spine regions
Picture；
Determining module, for determining that second bianry image is the lesion spine image.
The model building module is specifically used for:
Three-dimensional reconstruction is carried out to the targets spine image by marching cubes algorithm, obtains the corresponding target ridge
Column model.
The terminal 4 can be desktop PC, notebook, palm PC and cloud server etc. and calculate equipment.Institute
Stating terminal 4 may include, but be not limited only to, processor 40, memory 41.It will be understood by those skilled in the art that Fig. 7 is only eventually
The example at end 4, the not restriction of structure paired terminal 4 may include than illustrating more or fewer components, or the certain portions of combination
Part or different components, such as the terminal can also include input-output equipment, network access equipment, bus etc..
Alleged processor 40 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
The memory 41 can be the internal storage unit of the terminal 4, such as the hard disk or memory of terminal 4.It is described
Memory 41 is also possible to the External memory equipment of the terminal 4, such as the plug-in type hard disk being equipped in the terminal 4, intelligence
Storage card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card)
Deng.Further, the memory 41 can also both include the internal storage unit of the terminal 4 or set including external storage
It is standby.The memory 41 is for other programs and data needed for storing the computer program and the terminal.It is described to deposit
Reservoir 41 can be also used for temporarily storing the data that has exported or will export.
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.Each functional unit in embodiment, module can integrate in one processing unit, can also
To be that each unit physically exists alone, can also be integrated in one unit with two or more units, it is above-mentioned integrated
Unit both can take the form of hardware realization, can also realize in the form of software functional units.In addition, each function list
Member, the specific name of module are also only for convenience of distinguishing each other, the protection scope being not intended to limit this application.Above system
The specific work process of middle unit, module, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, is not described in detail or remembers in some embodiment
The part of load may refer to the associated description of other embodiments.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
Scope of the present application.
In embodiment provided herein, it should be understood that disclosed terminal and method can pass through others
Mode is realized.For example, terminal embodiment described above is only schematical, for example, the division of the module or unit,
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be with
In conjunction with or be desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling or direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING of device or unit or
Communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated module/unit be realized in the form of SFU software functional unit and as independent product sale or
In use, can store in a computer readable storage medium.Based on this understanding, the application realizes above-mentioned implementation
All or part of the process in example method, can also instruct relevant hardware to complete, the meter by computer program
Calculation machine program can be stored in a computer readable storage medium, the computer program when being executed by processor, it can be achieved that on
The step of stating each embodiment of the method.Wherein, the computer program includes computer program code, the computer program generation
Code can be source code form, object identification code form, executable file or certain intermediate forms etc..The computer-readable medium
It may include: any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic that can carry the computer program code
Dish, CD, computer storage, read-only memory (ROM, Read-Only Memory), random access memory (RAM,
Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that described
The content that computer-readable medium includes can carry out increasing appropriate according to the requirement made laws in jurisdiction with patent practice
Subtract, such as does not include electric carrier signal and electricity according to legislation and patent practice, computer-readable medium in certain jurisdictions
Believe signal.
Embodiment described above is only to illustrate the technical solution of the application, rather than its limitations；Although referring to aforementioned reality
Example is applied the application is described in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution should all
Comprising within the scope of protection of this application.
Claims (10)
1. a kind of spine model generation method characterized by comprising
According to non-lesion spine image, training generates confrontation network model；
Network model is fought by the generation after training, image is carried out to lesion site in lesion spine image and is repaired
It is multiple, obtain targets spine image, wherein lesion site described in the targets spine image is repaired as non-lesion backbone
Form；
Three-dimensional reconstruction is carried out to the targets spine image, obtains corresponding targets spine model.
2. spine model generation method according to claim 1, which is characterized in that it is described according to non-lesion spine image,
Training generates confrontation network model, comprising:
Semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, obtains the first binary map of spine regions
Picture；
Based on first bianry image, the training generation fights network model.
3. spine model generation method according to claim 1, which is characterized in that the generation by after training
Network model is fought, image repair is carried out to lesion site in lesion spine image, obtains targets spine image, comprising:
Based on the lesion spine image and preset two-value mask code matrix, generate in lesion site addition mask
Backbone mask image；
It is repaired based on the backbone mask image and backbone referring to image, network model is fought by the generation after training,
Generate the targets spine image F '=M ⊙ F+ (1-M) ⊙ G (z) after repairing*；
Wherein, F is the lesion spine image；M is the preset two-value mask code matrix, the member in the two-value mask code matrix
It is plain corresponding with the pixel of the lesion spine image；M ⊙ F is the backbone mask image；G(z)*For backbone reparation ginseng
According to image, the backbone reparation is that described generate sets confrontation meeting of generating during model training of network model referring to image
The fixed image referring to condition；⊙ is the operation that matrix element is multiplied with pixel point value in image.
4. spine model generation method according to claim 3, which is characterized in that described to be based on the backbone mask image
And backbone is repaired referring to image, is fought network model by the generation after training, is generated the targets spine after repairing
Before image, further includes:
Obtain the output image for generating confrontation network model and generating during model training；
According to the output image, according to loss function L=Lcontextual+λLperceptual, institute is obtained from the output image
Backbone is stated to repair referring to image；
Wherein, λ is hyper parameter；
Lcontextual=| | M ⊙ G (z)-M ⊙ F | |1, G (z) is the output image；
Lperceptual=log (1-D (G (z))), D (G (z)) are arbiter in the generation confrontation network model after training to institute
State the differentiation end value of output image；
The backbone reparation is referring to the output image that image is when the value of the loss function meets sets requirement.
5. spine model generation method according to claim 1, which is characterized in that the generation by after training
Network model is fought, image repair is carried out to lesion site in lesion spine image, before obtaining targets spine image, also
Include:
Semantic segmentation is carried out to the CT image of lesion backbone, obtains the second bianry image of spine regions；
Determine that second bianry image is the lesion spine image.
6. spine model generation method according to claim 1, which is characterized in that it is described to the targets spine image into
Row three-dimensional reconstruction obtains corresponding targets spine model, comprising:
Three-dimensional reconstruction is carried out to the targets spine image by marching cubes algorithm, obtains the corresponding targets spine mould
Type.
7. a kind of spine model generates system characterized by comprising
Model training module, for according to non-lesion spine image, training to generate confrontation network model；
Image repair module sends out lesion in lesion spine image for fighting network model by the generation after training
Raw position carries out image repair, obtains targets spine image, wherein lesion site described in the target image is repaired
For non-lesion spine form；
Model building module obtains corresponding targets spine model for carrying out three-dimensional reconstruction to the targets spine image.
8. spine model according to claim 7 generates system, which is characterized in that the model training module is specifically used
In:
Semantic segmentation is carried out to the computer tomography CT image of non-lesion backbone, obtains the first binary map of spine regions
Picture；
Based on first bianry image, the training generation fights network model.
9. a kind of terminal, including memory, processor and storage can be run in the memory and on the processor
Computer program, which is characterized in that the processor is realized when executing the computer program as claim 1 to 6 is any
The step of item the method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In when the computer program is executed by processor the step of any one of such as claim 1 to 6 of realization the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910450859.5A CN110246216B (en) | 2019-05-28 | 2019-05-28 | Spine model generation method, spine model generation system and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910450859.5A CN110246216B (en) | 2019-05-28 | 2019-05-28 | Spine model generation method, spine model generation system and terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110246216A true CN110246216A (en) | 2019-09-17 |
| CN110246216B CN110246216B (en) | 2023-02-28 |
Family
ID=67885234
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910450859.5A Active CN110246216B (en) | 2019-05-28 | 2019-05-28 | Spine model generation method, spine model generation system and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110246216B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127412A (en) * | 2019-12-18 | 2020-05-08 | 广东工业大学 | Pathological image recognition device based on generation countermeasure network |
| CN111274990A (en) * | 2020-02-11 | 2020-06-12 | 广东同天投资管理有限公司 | Computer device and storage medium for classifying spinal morphology |
| CN111353995A (en) * | 2020-03-31 | 2020-06-30 | 成都信息工程大学 | Cervical single cell image data generation method based on generation countermeasure network |
| CN112734740A (en) * | 2021-01-18 | 2021-04-30 | 推想医疗科技股份有限公司 | Method for training target detection model, method for detecting target and device thereof |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105723423A (en) * | 2013-10-30 | 2016-06-29 | 皇家飞利浦有限公司 | Volumetric image data visualization |
| CN107945140A (en) * | 2017-12-20 | 2018-04-20 | 中国科学院深圳先进技术研究院 | A kind of image repair method, device and equipment |
| CN107977971A (en) * | 2017-11-09 | 2018-05-01 | 哈尔滨理工大学 | The method of vertebra positioning based on convolutional neural networks |
| CN108038860A (en) * | 2017-11-30 | 2018-05-15 | 杭州电子科技大学 | Spine segmentation method based on the full convolutional neural networks of 3D |
| CN108460830A (en) * | 2018-05-09 | 2018-08-28 | 厦门美图之家科技有限公司 | Image repair method, device and image processing equipment |
| CN108765340A (en) * | 2018-05-29 | 2018-11-06 | Oppo（重庆）智能科技有限公司 | Fuzzy image processing method, apparatus and terminal device |
| CN108765417A (en) * | 2018-06-15 | 2018-11-06 | 西安邮电大学 | It is a kind of that system and method is generated based on the femur X-ray film of deep learning and digital reconstruction irradiation image |
| CN109242865A (en) * | 2018-09-26 | 2019-01-18 | 上海联影智能医疗科技有限公司 | Medical image auto-partition system, method, apparatus and storage medium based on multichannel chromatogram |
| US20190103190A1 (en) * | 2017-09-29 | 2019-04-04 | K2M, Inc. | Systems and methods for simulating spine and skeletal system pathologies |
| US20190105009A1 (en) * | 2017-10-10 | 2019-04-11 | Holo Surgical Inc. | Automated segmentation of three dimensional bony structure images |
| WO2019079778A1 (en) * | 2017-10-20 | 2019-04-25 | Nuvasive, Inc. | Intervertebral disc modeling |
-
2019
- 2019-05-28 CN CN201910450859.5A patent/CN110246216B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105723423A (en) * | 2013-10-30 | 2016-06-29 | 皇家飞利浦有限公司 | Volumetric image data visualization |
| US20190103190A1 (en) * | 2017-09-29 | 2019-04-04 | K2M, Inc. | Systems and methods for simulating spine and skeletal system pathologies |
| US20190105009A1 (en) * | 2017-10-10 | 2019-04-11 | Holo Surgical Inc. | Automated segmentation of three dimensional bony structure images |
| WO2019079778A1 (en) * | 2017-10-20 | 2019-04-25 | Nuvasive, Inc. | Intervertebral disc modeling |
| CN107977971A (en) * | 2017-11-09 | 2018-05-01 | 哈尔滨理工大学 | The method of vertebra positioning based on convolutional neural networks |
| CN108038860A (en) * | 2017-11-30 | 2018-05-15 | 杭州电子科技大学 | Spine segmentation method based on the full convolutional neural networks of 3D |
| CN107945140A (en) * | 2017-12-20 | 2018-04-20 | 中国科学院深圳先进技术研究院 | A kind of image repair method, device and equipment |
| CN108460830A (en) * | 2018-05-09 | 2018-08-28 | 厦门美图之家科技有限公司 | Image repair method, device and image processing equipment |
| CN108765340A (en) * | 2018-05-29 | 2018-11-06 | Oppo（重庆）智能科技有限公司 | Fuzzy image processing method, apparatus and terminal device |
| CN108765417A (en) * | 2018-06-15 | 2018-11-06 | 西安邮电大学 | It is a kind of that system and method is generated based on the femur X-ray film of deep learning and digital reconstruction irradiation image |
| CN109242865A (en) * | 2018-09-26 | 2019-01-18 | 上海联影智能医疗科技有限公司 | Medical image auto-partition system, method, apparatus and storage medium based on multichannel chromatogram |
Non-Patent Citations (2)
| Title |
|---|
| 刘忠利等: "基于深度学习的脊柱CT图像分割", 《计算机应用与软件》 * |
| 陈锟等: "生成对抗网络在医学图像处理中的应用", 《生命科学仪器》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127412A (en) * | 2019-12-18 | 2020-05-08 | 广东工业大学 | Pathological image recognition device based on generation countermeasure network |
| CN111127412B (en) * | 2019-12-18 | 2021-05-28 | 广东工业大学 | Pathological image recognition device based on generation countermeasure network |
| CN111274990A (en) * | 2020-02-11 | 2020-06-12 | 广东同天投资管理有限公司 | Computer device and storage medium for classifying spinal morphology |
| CN111274990B (en) * | 2020-02-11 | 2024-01-12 | 广东同天科技产业发展有限公司 | Computer device and storage medium for classifying spinal morphology |
| CN111353995A (en) * | 2020-03-31 | 2020-06-30 | 成都信息工程大学 | Cervical single cell image data generation method based on generation countermeasure network |
| CN111353995B (en) * | 2020-03-31 | 2023-03-28 | 成都信息工程大学 | Cervical single cell image data generation method based on generation countermeasure network |
| CN112734740A (en) * | 2021-01-18 | 2021-04-30 | 推想医疗科技股份有限公司 | Method for training target detection model, method for detecting target and device thereof |
| CN112734740B (en) * | 2021-01-18 | 2024-02-23 | 推想医疗科技股份有限公司 | Method for training target detection model, target detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110246216B (en) | 2023-02-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110246216A (en) | Spine model generation method, spine model generate system and terminal | |
| Ismael et al. | An enhanced deep learning approach for brain cancer MRI images classification using residual networks | |
| CN108257134A (en) | Nasopharyngeal Carcinoma Lesions automatic division method and system based on deep learning | |
| CN109966662A (en) | A kind of method and system for verifying radiotherapy dosage | |
| Tomatis et al. | Late rectal bleeding after 3D-CRT for prostate cancer: development of a neural-network-based predictive model | |
| Auricchio et al. | Patient‐specific finite element analysis of carotid artery stenting: a focus on vessel modeling | |
| Veneziano et al. | Surface smoothing, decimation, and their effects on 3D biological specimens | |
| CN110070540A (en) | Image generating method, device, computer equipment and storage medium | |
| CN108765340A (en) | Fuzzy image processing method, apparatus and terminal device | |
| CN113096137B (en) | Adaptive segmentation method and system for OCT (optical coherence tomography) retinal image field | |
| CN109785399B (en) | Synthetic lesion image generation method, device, equipment and readable storage medium | |
| Shiranthika et al. | Supervised optimal chemotherapy regimen based on offline reinforcement learning | |
| Qiu et al. | Learnable ophthalmology sam | |
| Amini et al. | Augmented reality mastectomy surgical planning prototype using the HoloLens template for healthcare technology letters | |
| CN110796064B (en) | Human muscle image establishing method and device, storage medium and electronic equipment | |
| CN114998374A (en) | Image segmentation method, device and equipment based on position prior and storage medium | |
| Jahan et al. | Automated diagnosis of pneumonia from classification of chest x-ray im ages using efficientnet | |
| Jayasinghe et al. | Patient Reported Outcome Measures After Breast Augmentation–Using the BREAST-Q IS | |
| CN115985513B (en) | Data processing method, device and equipment based on multiple groups of chemical cancer typing | |
| CN116612849A (en) | Medical treatment scheme recommendation method, device, equipment and storage medium | |
| CN115760813A (en) | Screw channel generation method, device, equipment, medium and program product | |
| Jerg et al. | Diffuse domain method for needle insertion simulations | |
| Huang et al. | Development of a deep learning-based tool to assist wound classification | |
| Damayanti et al. | Lung cancer classification using convolutional neural network and DenseNet | |
| US20060264713A1 (en) | Disease and therapy dissemination representation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |