CN109949288A - Tumor type determines system, method and storage medium - Google Patents
Tumor type determines system, method and storage medium Download PDFInfo
- Publication number
- CN109949288A CN109949288A CN201910199475.0A CN201910199475A CN109949288A CN 109949288 A CN109949288 A CN 109949288A CN 201910199475 A CN201910199475 A CN 201910199475A CN 109949288 A CN109949288 A CN 109949288A
- Authority
- CN
- China
- Prior art keywords
- analyzed
- medical image
- tumor type
- image data
- groups
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The embodiment of the invention discloses tumor types to determine system, method and storage medium, which includes: the one or more groups of medical image datas to be analyzed for obtaining module and being used to obtain subject;In the staging model that tumor type determining module is used to train the medical image data input to be analyzed, with the tumor type of determination the included lesion of medical image data to be analyzed.The tumor type for solving the prior art determines the lower technical problem of the accuracy of method, has reached and has improved the technical effect that tumor type determines accuracy.
Description
Technical field
The present embodiments relate to technical field of medical image processing more particularly to a kind of tumor type to determine system, side
Method and storage medium.
Background technique
Brain tumor is most commonly seen one of the disease of nervous system, is had because its discovery is late, deterioration is fast, operation risk is high
Very high lethality is difficult medical problem recognized in the world.Current brain tumor is many kinds of, and rank is complicated, and inhomogeneity
Not, the brain tumor of rank is in treatment method and rear aspect has very big difference, this before surgery to carry out brain tumor
Positioning qualitative becomes particularly important.Routine CT (Computed Tomography, CT scan, letter at this stage
Claim CT), MRI (Magnetic Resonance Imaging, magnetic resonance, abbreviation MRI) imageological examination can only be to its position, shape
State and size make tentative diagnosis, are difficult to make etiologic diagnosis.
The main stream approach of staging is first to come out lesion segmentation at present, the subsequent feature according to tumor section image into
Row further classification.Lesion segmentation, which refers to, extracts tumor section from image, and the accuracy of segmentation directly influences
The accuracy of staging, but the existing lesion segmentation approach time is long, accuracy is low, and is typically based on and sets when staging
The selected tumour identification feature of meter person is classified, this is readily incorporated the cognitive Bias of subjectivity, to limit staging
Accuracy.
To sum up, the tumor type of the prior art determines that the accuracy of method is lower.
Summary of the invention
The embodiment of the invention provides a kind of tumor types to determine system, method and storage medium, to solve the prior art
Tumor type determine the lower technical problem of the accuracy of method.
In a first aspect, the embodiment of the invention provides a kind of tumor types to determine system, comprising:
Module is obtained, for obtaining one or more groups of medical image datas to be analyzed of subject;
Tumor type determining module, for the medical image data to be analyzed to be inputted the staging model trained
In, with the tumor type of determination the included lesion of medical image data to be analyzed.
Further, if medical image data to be analyzed is multiple groups, difference organize other medical image datas to be analyzed because
The difference of image processing method and/or testing conditions and it is different.
Further, different testing conditions refer to the different image-forming conditions of same medical imaging device.
Further, the image-forming condition is the pulse train of magnetic resonance imaging.
Further, described image processing method includes at least one of denoising, gray proces and image registration.
Further, at least two groups of the medical image data to be analyzed, the staging model trained include
Multiple data channel, correspondingly, tumor type determining module is specifically used for leading to different groups of other medical image datas to be analyzed
It crosses different data channel and inputs the staging model trained, include disease with the determination medical image data to be analyzed
The tumor type of stove.
Further, the staging model trained is neural network model, correspondingly, the tumor type determines
Module is specifically used for different groups of other medical image datas to be analyzed inputting the nerve trained by different data channel
Network model, to determine that lesion belongs to the probability of every kind of tumor type, and using the corresponding tumor type of maximum probability as lesion
Tumor type output.
It further, further include Model selection module, the Model selection module provides the instruction at corresponding different human body position
Experienced staging model.
Second aspect, the embodiment of the invention also provides a kind of tumor types to determine method, comprising:
Obtain one or more groups of medical image datas to be analyzed of subject;
The medical image data to be analyzed is inputted into the staging model trained, obtains the medicine shadow to be analyzed
As the tumor type of the included lesion of data.
The third aspect, it is described the embodiment of the invention also provides a kind of storage medium comprising computer executable instructions
Computer executable instructions as computer processor when being executed for executing the tumor type side of determination as described in second aspect
Method.
Tumor type provided in an embodiment of the present invention determines the technical solution of method, obtains subject's by obtaining module
One or more groups of medical image datas to be analyzed;The medical image data to be analyzed is inputted by tumor type determining module
In the staging model trained, with the tumor type of determination the included lesion of medical image data to be analyzed.Pass through
The staging model trained analyzes one or more groups of medical image datas to be analyzed, can be quickly and accurately obtained
The tumor type of the included lesion of medical image data to be analyzed can provide strong data branch for clinical tumor type diagnostic
It holds.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing does one and simply introduces, it should be apparent that, drawings in the following description are some embodiments of the invention, for this
For the those of ordinary skill of field, without creative efforts, it can also be obtained according to these attached drawings others
Attached drawing.
Fig. 1 is the structural block diagram that the tumor type that the embodiment of the present invention one provides determines system;
Fig. 2A is the original T1 weighted magnetic resonance images that the embodiment of the present invention one provides;
Fig. 2 B is the pretreated T1 weighted magnetic resonance images that the embodiment of the present invention one provides;
Fig. 2 C is the original T2 weighted magnetic resonance images that the embodiment of the present invention one provides;
Fig. 2 D is the pretreated T2 weighted magnetic resonance images that the embodiment of the present invention one provides;
Fig. 3 is the schematic diagram for the operation interface that the embodiment of the present invention one provides;
Fig. 4 is the structural block diagram for the another medical image analysis system that the embodiment of the present invention one provides.
Fig. 5 is the schematic diagram for the another operation interface that the embodiment of the present invention one provides;
Fig. 6 is the flow chart that tumor type provided by Embodiment 2 of the present invention determines method;
Fig. 7 is the flow chart that the staging model that the embodiment of the present invention three provides determines method;
Fig. 8 is the staging model established based on three-dimensional dense connection network algorithm that the embodiment of the present invention three provides
Schematic diagram;
Fig. 9 is the schematic diagram for each Dense Layer that the embodiment of the present invention three provides.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, hereinafter with reference to attached in the embodiment of the present invention
Figure, clearly and completely describes technical solution of the present invention by embodiment, it is clear that described embodiment is the present invention one
Section Example, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not doing
Every other embodiment obtained under the premise of creative work out, shall fall within the protection scope of the present invention.
Embodiment one
Fig. 1 is the structural block diagram for the medical image analysis system that the embodiment of the present invention one provides.The system includes obtaining mould
Block 11 and tumor type determining module 12 obtain one or more groups of medical image numbers to be analyzed that module 11 is used to obtain subject
According to;Tumor type determining module 12 is used to be analysed in the staging model that medical image data input has been trained, with true
The tumor type of fixed the included lesion of medical image data to be analyzed.
Medical image data to be analyzed can be obtained by modes such as network communication and transmission, copies by obtaining module 11.Wherein,
The type of medical image data to be analyzed is that clinical tumor diagnoses common medical image data type, such as MRI, CT, PET-
(Positron Emission Computed Tomography-Computed Tomography, positron emission calculate CT
Machine tomoscan-CT scan, abbreviation PET-CT) etc..If medical image data to be analyzed is multiple groups,
It is different due to the difference of image processing method and/or testing conditions that difference organizes other medical image data to be analyzed.
Wherein, different testing conditions can refer to different detection devices, for example, one group of medical image data to be analyzed is CT
Data, another group of medical image data to be analyzed are MRI data;It can also be the different image-forming conditions of same detection device, with
For MRI machine, one group of medical image data to be analyzed is T1 weighted magnetic resonance images, another group of medical image data to be analyzed
For T2 weighted magnetic resonance images or it is based on magnetic resonance imaging fluid-attenuated inversion recovery sequence (fluid attenuated
Inversion recovery, abbreviation FLAIR) retrieval magnetic resonance image.If difference organizes other medicine shadow to be analyzed
Picture data come from same medical image detection device, and obtaining module 11 can also connect at the image of the medical image detection device
Manage module.
After acquisition module 11 gets medical image data to be analyzed, tumor type determining module 12 is first to acquired
Medical image data to be analyzed carries out image procossing to update the medical image data to be analyzed.Image procossing in the present embodiment
Method is preferably image pre-processing method, such as denoising, gray proces and image registration.By taking head magnetic resonance image as an example, figure
It corrected as pretreatment refers to by resampling, adjustment image direction, bias field, go the sequence of operations such as skull, gray correction, with
Reduce influence of the working condition to picture quality of MRI machine, wherein gray correction can choose grey level histogram matching or
Subtract mean value, the standardization divided by standard deviation.Fig. 2A shows original T1 weighted magnetic resonance images, and Fig. 2 B shows pre- place
T1 weighted magnetic resonance images after reason;Fig. 2 C shows original T2 weighted magnetic resonance images, and Fig. 2 D shows pretreated
T2 weighted magnetic resonance images.It is understood that the difference of image processing method, can be one or more image preprocessing steps
Rapid different, the combination for being also possible to image preprocessing step is different.
It is understood that if two groups of medical image datas to be analyzed of acquired subject because of testing conditions not
It is same and different, it is also necessary to which that different groups of other medical image datas to be analyzed are registrated.
If obtaining module 11 only obtains one group of medical image data to be analyzed, tumor type determining module 12 passes through
The staging model trained analyzes group medical image data to be analyzed, with determination medical image data to be analyzed
The tumor type of included lesion, and tumor type is exported into the analysis results display area to operation interface 14 so that user looks into
It sees, referring to attached drawing 3.
If obtaining module 11 obtains at least two groups medical image data to be analyzed, tumor type determining module 12 is logical
The staging model trained comprising multiple data channel is crossed, medical image data to be analyzed is analyzed, to obtain
The tumor type of the included lesion of medical image data to be analyzed.Preferably, staging model is neural network model, then
The neural network model trained can obtain lesion when analyzing different groups of other medical image datas to be analyzed and belong to often
The probability of kind of tumor type, and using the corresponding tumor type of maximum probability as the output of the tumor type of lesion, and by tumour class
Type exports the analysis results display area to operation interface so that user checks, referring to attached drawing 3.
Wherein, the data channel quantity for the staging model trained is equal to the group of medical image data to be analyzed
Number.
Wherein, neural network model used in tumor type determining module 12 can be based on three-dimensional residual error network algorithm, three
The dense connection network algorithm of three-dimensional for tieing up depth convolutional network algorithm or binding characteristic weighting block is established, and is preferably based on three
It ties up dense connection network algorithm to establish, and the loss function of staging model is Focal Loss or cross entropy.
When due to medical image data acquisition, usually acquired according to human body, such as head, chest, abdomen etc., in order to
The tumor type of multiple location can be distinguished, the present embodiment further includes Model selection module 13, for providing multiple human bodies
The staging model (referring to fig. 4 and Fig. 5) trained.Preferably, it is constructed when each human body is corresponding based on algorithms of different
The staging model trained when, user can first select human body, and then reselection is constructed based on target algorithm
The staging model trained;When each human body only corresponds to a staging model trained, user can be with
Directly select human body corresponding to medical image data to be analyzed.
The technical solution of Medical Imaging System provided in an embodiment of the present invention, one group for obtaining subject by obtaining module
Or plurality of medical image data;The tumour point that medical image data input has been trained is analysed to by tumor type determining module
In class model, to obtain the tumor type of the included lesion of medical image data to be analyzed.Pass through the staging mould trained
Type analyzes one or more groups of medical image datas to be analyzed, can be quickly and accurately obtained medical image data to be analyzed
The tumor type of included lesion can provide strong data for clinical tumor type diagnostic and support.
Embodiment two
Fig. 6 is the flow chart that tumor type provided by Embodiment 2 of the present invention determines method.The technical solution of the present embodiment
The case where suitable for automatically determining tumor type according to one or more groups of medical image datas to be analyzed.This method can be by this hair
Tumor type determining device that bright embodiment provides executes, which can be realized by the way of software and/or hardware, and
Configuration is applied in Medical Imaging System.This method specifically comprises the following steps:
S101, the one or more groups of medical image datas to be analyzed for obtaining subject.
Wherein, the type of medical image data to be analyzed is that clinical tumor diagnoses common medical image data type, than
Such as MRI, CT, PET-CT.If the medical image data to be analyzed of subject is multiple groups, difference organizes other medicine to be analyzed
Image data is different due to the difference of image processing method and/or testing conditions.
Wherein, different testing conditions can be different detection device, for example, one group of medical image data to be analyzed is CT
Data, another group of medical image data to be analyzed are MRI data;It can also be the different image-forming conditions of same detection device, with
For MRI machine, one group of medical image data to be analyzed is the magnetic resonance image of T1 weighting, another group of medical image number to be analyzed
According to the magnetic resonance image weighted for T2 or it is based on magnetic resonance imaging fluid-attenuated inversion recovery sequence (fluid attenuated
Inversion recovery, abbreviation FLAIR) retrieval magnetic resonance image.
If the testing conditions of two groups of medical image datas to be analyzed of subject are different, waited for point using the multiple groups
Before analysing medical image data progress tumor type analysis, need first to match different groups of other medical image datas to be analyzed
Standard, to update the other medical image data to be analyzed of each group, to make different groups of other medical image datas to be analyzed in space
It is matched on position.
Wherein, the image processing method in the present embodiment is preferably image pre-processing method, for example, denoising, gray proces and
Image registration.By taking head magnetic resonance image as an example, image preprocessing is typically referred to through resampling, adjustment image direction, bias
The sequence of operations such as skull, gray correction are gone in field correction, to reduce influence of the working condition to picture quality of MRI machine,
Middle gray correction can choose grey level histogram matching or subtract mean value, the standardization divided by standard deviation.Fig. 2A is shown
Original T1 weighted magnetic resonance images, Fig. 2 B show pretreated T1 weighted magnetic resonance images;Fig. 2 C shows original
T2 weighted magnetic resonance images, Fig. 2 D show pretreated T2 weighted magnetic resonance images.It is understood that image processing method
The difference of method can be one or more image preprocessing steps differences, be also possible to the combination of image preprocessing step
It is different.
S102, it is analysed in the staging model that medical image data input has been trained, with determination medicine to be analyzed
The tumor type of the included lesion of image data.
If only obtaining one group of medical image data to be analyzed, group medical image data input to be analyzed has been instructed
In experienced staging model, to obtain the tumor type of the included lesion of medical image data to be analyzed.
If obtain at least two groups medical image data to be analyzed, having trained comprising multiple data channel is used
Then staging model has been trained different groups of other medical image datas to be analyzed by different data channel inputs
Staging model, to obtain the tumor type of the included lesion of medical image data to be analyzed.Preferably, staging model
For neural network model, then the neural network model trained is analyzed to different groups of other medical image datas to be analyzed
When can obtain the probability that lesion belongs to every kind of tumor type, then the corresponding tumor type of maximum probability is the tumor type of lesion
Output.
Wherein, the data channel quantity for the staging model trained is equal to the group of medical image data to be analyzed
Number.
Wherein, neural network model used in the staging model of the present embodiment can be based on Three dimensional convolution neural network
The dense connection of three-dimensional of algorithm, three-dimensional residual error network algorithm, three dimensional depth convolutional network algorithm or binding characteristic weighting block
What network algorithm was established, it is preferred to use three-dimensional dense connection network algorithm is established, and the loss function of staging model is
Focal Loss or cross entropy.
Tumor type provided in an embodiment of the present invention determines the technical solution of method, one group or more including obtaining subject
Group medical image data to be analyzed;Be analysed in the staging model trained of medical image data input, with obtain to
Analyze the tumor type of the included lesion of medical image data.By the staging model trained to it is one or more groups of to point
Analysis medical image data is analyzed, and the included lesion of medical image data to be analyzed can automatically, be quickly and accurately obtained
Tumor type provides strong data for clinical tumor diagnosis and supports.
Embodiment three
Fig. 7 is the flow chart for the staging model training method that the embodiment of the present invention three provides.The embodiment of the present invention exists
On the basis of above-described embodiment, the step of increasing staging model training method, comprising:
S201, obtain preset quantity subject medical image data to be analyzed, and to the to be analyzed of each subject
Medical image data carries out tumor type mark, to create the training set sample for training staging model.
Illustratively, it includes 336 samples, each sample packet that the present embodiment, which is used to train the training set of staging model,
Two groups of magnetic resonance image of a subject are included, this two groups of magnetic resonance image are respectively T1 weighted magnetic resonance images and T2 weighting magnetic
Resonance image.Tumor type mark is carried out to every group of magnetic resonance image of each subject, for example, with common astrocyte
For tumor, glioblastoma and meningioma.After tumor type mark, this 336 samples include 161 astrocytoma samples
Originally, 104 glioblastoma samples and 71 meningioma samples.
To two groups of original magnetic resonance images (A and Fig. 2 C referring to fig. 2) of each subject carry out matching calibration, resampling,
Bias field correction, the operation such as remove skull, make all image normalizations having a size of 256 × 256 × 256 mm3, voxel size be 1 ×
1×1mm3, direction is the standard picture of standard Descartes's LPI coordinate system, using the standard picture as medical image to be analyzed
Data (B and Fig. 2 D referring to fig. 2) then carry out image registration to two groups of medical image datas to be analyzed in each sample, than
Method is such as registrated using affine transformation, T2 weighted magnetic resonance images are registrated to T1 weighted magnetic resonance images by affine transformation, with
Medical image data to be analyzed is updated, to make T1 weighted magnetic resonance images and T2 weighted magnetic resonance images in spatial position on
Match.
It should be noted that the present embodiment not to the pretreatment of two groups of medical image datas to be analyzed of each sample, match
The sequencing of quasi- and tumor type mark is defined,
S202, the medical image data to be analyzed input staging model after mark is trained, has been instructed with generating
Experienced staging model.
By taking the staging model established based on three-dimensional dense connection network algorithm as an example, which includes two
A data channel, network depth are preferably arranged to 20 layers, and convolution kernel is preferably small size convolution kernel, such as 1 × 1 × 1 or 3 × 3
× 3 convolution kernel to guarantee to obtain enough image informations, and will not ignore lesion.Using Focal Loss as loss function,
Using line rectification function ReLU as activation primitive, using can adaptive regularized learning algorithm rate Adam optimizer to network parameter into
Row optimization.
Specifically, as shown in figure 8, the staging model include input layer, convolutional layer, Block layers of Dense it is (dense
Block layer), pond layer, full articulamentum, output layer and interlayer connection build.In order not to ignore the lesser tumour of size, convolutional layer
Convolution kernel be set as 3 × 3 × 3, step-length is set as 1, the average value pond layer that the convolution kernel of pond layer is 2 × 2 × 2, step-length
To include some Dense Layers in Block layers of 2, Dense.Network depth is preferably 20, and uses 4 Dense Block
Structure, this four Dense Block structures separately include 3,3,6 and 8 Dense Layers, and each Dense Layer
In include batch normalization layer, active coating and convolutional layer.
As shown in figure 9, be preferably 1 × 1 × 1 for the convolution kernel of first convolutional layer of each Dense Layer, step
Long by preferably 1, the convolution kernel of second convolutional layer is preferably 3 × 3 × 3, and step-length is preferably 1, active coating be all using ReLU
Function.In Dense Block, input of the output of each Dense Layer as subsequent all Dense Layer is realized
The recycling of feature strengthens the propagation of feature in a network, has mitigated gradient disappearance problem.Since the present embodiment is to three
Kind tumor type is classified, therefore the output vector of full articulamentum is changed to 1 × 3 by us, while not in view of training sample
The problem of balance, using Focal Loss as loss function, the training pattern in the update of continuous iteration, to obtain final
Model parameter.
The medical image data to be analyzed that all training sets of tumor type will be identified with inputs above-mentioned staging model
It is trained, the staging model trained, wherein two groups of medical image datas to be analyzed of each sample pass through two
A different data channel inputs the staging model.
Illustratively, using the aforementioned staging model trained to the medical image number to be analyzed in forecast set sample
According to being analyzed, wherein forecast set includes 70 samples, specifically includes 33 astrocytoma samples, 22 glioblasts
Tumor sample and 15 meningioma samples, analysis result are as shown in the table:
1 test sample of table analyzes result
Analyze as the result is shown: for astrocytoma, wherein 32 predictions are correct, 1 prediction error is mistaken for colloid mother
Cytoma;For glioblastoma, wherein 20 predictions are correct, 2 prediction errors are mistaken for astrocytoma;It is right
In meningioma, wherein 11 predictions are correct, 4 prediction errors, 3 are mistaken for astrocytoma, and 1 is mistaken for colloid mother carefully
Born of the same parents' tumor.Final total accuracy is (32+20+11)/(33+22+15)=90%.
The embodiment of the present invention instructs staging model using the medical image data to be analyzed in training set
Practice, the staging model trained, so as to pass through the staging model trained to medicine shadow to be analyzed
As data are analyzed.
Example IV
The embodiment of the present invention four also provides a kind of storage medium comprising computer executable instructions, and the computer can be held
Row instruction determines method for executing tumor type when being executed by computer processor, this method comprises:
Obtain one or more groups of medical image datas to be analyzed of subject;
In the staging model that the medical image data input to be analyzed has been trained, to obtain the doctor to be analyzed
Learn the tumor type of the included lesion of image data.
Certainly, a kind of storage medium comprising computer executable instructions, computer provided by the embodiment of the present invention
The method operation that executable instruction is not limited to the described above, can also be performed tumor type provided by any embodiment of the invention
Determine the relevant operation in method.
By the description above with respect to embodiment, it is apparent to those skilled in the art that, the present invention
It can be realized by software and required common hardware, naturally it is also possible to which by hardware realization, but in many cases, the former is more
Good embodiment.Based on this understanding, technical solution of the present invention substantially in other words contributes to the prior art
Part can be embodied in the form of software products, which can store in computer readable storage medium
In, floppy disk, read-only memory (Read-Only Memory, abbreviation ROM), random access memory (Random such as computer
Access Memory, abbreviation RAM), flash memory (FLASH), hard disk or CD etc., including some instructions are used so that a calculating
Machine equipment (can be personal computer, server or the network equipment etc.) executes tumour described in each embodiment of the present invention
Type determines method.
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (10)
1. a kind of tumor type determines system characterized by comprising
Module is obtained, for obtaining one or more groups of medical image datas to be analyzed of subject;
Tumor type determining module, for the medical image data to be analyzed to be inputted in the staging model trained,
With the tumor type of determination the included lesion of medical image data to be analyzed.
2. system according to claim 1, which is characterized in that if the medical image data to be analyzed is multiple groups,
It is different due to the difference of image processing method and/or testing conditions that difference organizes other medical image data to be analyzed.
3. system according to claim 2, which is characterized in that different testing conditions refer to same medical imaging device
Different image-forming conditions.
4. system according to claim 3, which is characterized in that the image-forming condition is the pulse train of magnetic resonance imaging.
5. system according to claim 2, which is characterized in that described image processing method, which includes at least, to be denoised, at gray scale
One of reason and image registration.
6. system according to claim 2, which is characterized in that at least two groups of the medical image data to be analyzed, institute
Stating the staging model trained includes multiple data channel, and the tumor type determining module is specifically used for different groups
Medical image data to be analyzed the staging model trained is inputted by different data channel, it is described wait divide to determine
Analyse the tumor type of the included lesion of medical image data.
7. system according to claim 6, which is characterized in that the staging model trained is neural network mould
Type, the tumor type determining module are specifically used for leading to different groups of other medical image datas to be analyzed by different data
Road inputs the neural network model trained, to determine that lesion belongs to the probability of every kind of tumor type, and maximum probability is corresponding
Tumor type as lesion tumor type export.
8. -7 any system according to claim 1, which is characterized in that it further include Model selection module, the model choosing
It selects module and the staging model trained at corresponding different human body position is provided.
9. a kind of tumor type determines method characterized by comprising
Obtain one or more groups of medical image datas to be analyzed of subject;
The medical image data to be analyzed is inputted into the staging model trained, obtains the medical image number to be analyzed
According to the tumor type of included lesion.
10. a kind of storage medium comprising computer executable instructions, which is characterized in that the computer executable instructions by
Method is determined for executing tumor type as claimed in claim 9 when computer processor executes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910199475.0A CN109949288A (en) | 2019-03-15 | 2019-03-15 | Tumor type determines system, method and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910199475.0A CN109949288A (en) | 2019-03-15 | 2019-03-15 | Tumor type determines system, method and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109949288A true CN109949288A (en) | 2019-06-28 |
Family
ID=67010125
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910199475.0A Pending CN109949288A (en) | 2019-03-15 | 2019-03-15 | Tumor type determines system, method and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109949288A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110600107A (en) * | 2019-08-29 | 2019-12-20 | 上海联影智能医疗科技有限公司 | Method for screening medical images, computer device and readable storage medium |
CN110889836A (en) * | 2019-11-22 | 2020-03-17 | 中国人民解放军第四军医大学 | Image data analysis method and device, terminal equipment and storage medium |
CN112085113A (en) * | 2020-09-14 | 2020-12-15 | 四川大学华西医院 | Severe tumor image recognition system and method |
CN112927799A (en) * | 2021-04-13 | 2021-06-08 | 中国科学院自动化研究所 | Life cycle analysis system fusing multi-example learning and multi-task depth imaging group |
Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106296699A (en) * | 2016-08-16 | 2017-01-04 | 电子科技大学 | Cerebral tumor dividing method based on deep neural network and multi-modal MRI image |
CN106780631A (en) * | 2017-01-11 | 2017-05-31 | 山东大学 | A kind of robot closed loop detection method based on deep learning |
CN106875380A (en) * | 2017-01-12 | 2017-06-20 | 西安电子科技大学 | A kind of heterogeneous image change detection method based on unsupervised deep neural network |
US20170193657A1 (en) * | 2015-12-30 | 2017-07-06 | Case Western Reserve University | Prediction of recurrence of non-small cell lung cancer with tumor infiltrating lymphocyte (til) graphs |
CN107240102A (en) * | 2017-04-20 | 2017-10-10 | 合肥工业大学 | Malignant tumour area of computer aided method of early diagnosis based on deep learning algorithm |
CN107403201A (en) * | 2017-08-11 | 2017-11-28 | 强深智能医疗科技(昆山)有限公司 | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method |
CN107563434A (en) * | 2017-08-30 | 2018-01-09 | 山东大学 | A kind of brain MRI image sorting technique based on Three dimensional convolution neutral net, device |
CN107748900A (en) * | 2017-11-08 | 2018-03-02 | 山东财经大学 | Tumor of breast sorting technique and device based on distinction convolutional neural networks |
CN107767378A (en) * | 2017-11-13 | 2018-03-06 | 浙江中医药大学 | The multi-modal Magnetic Resonance Image Segmentation methods of GBM based on deep neural network |
CN108257134A (en) * | 2017-12-21 | 2018-07-06 | 深圳大学 | Nasopharyngeal Carcinoma Lesions automatic division method and system based on deep learning |
CN108537773A (en) * | 2018-02-11 | 2018-09-14 | 中国科学院苏州生物医学工程技术研究所 | Intelligence auxiliary mirror method for distinguishing is carried out for cancer of pancreas and pancreas inflammatory disease |
CN109214451A (en) * | 2018-08-28 | 2019-01-15 | 北京安德医智科技有限公司 | A kind of classification method and equipment of brain exception |
CN109242839A (en) * | 2018-08-29 | 2019-01-18 | 上海市肺科医院 | A kind of good pernicious classification method of CT images Lung neoplasm based on new neural network model |
CN109242860A (en) * | 2018-08-21 | 2019-01-18 | 电子科技大学 | Based on the brain tumor image partition method that deep learning and weight space are integrated |
CN109447088A (en) * | 2018-10-16 | 2019-03-08 | 杭州依图医疗技术有限公司 | A kind of method and device of breast image identification |
-
2019
- 2019-03-15 CN CN201910199475.0A patent/CN109949288A/en active Pending
Patent Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170193657A1 (en) * | 2015-12-30 | 2017-07-06 | Case Western Reserve University | Prediction of recurrence of non-small cell lung cancer with tumor infiltrating lymphocyte (til) graphs |
CN106296699A (en) * | 2016-08-16 | 2017-01-04 | 电子科技大学 | Cerebral tumor dividing method based on deep neural network and multi-modal MRI image |
CN106780631A (en) * | 2017-01-11 | 2017-05-31 | 山东大学 | A kind of robot closed loop detection method based on deep learning |
CN106875380A (en) * | 2017-01-12 | 2017-06-20 | 西安电子科技大学 | A kind of heterogeneous image change detection method based on unsupervised deep neural network |
CN107240102A (en) * | 2017-04-20 | 2017-10-10 | 合肥工业大学 | Malignant tumour area of computer aided method of early diagnosis based on deep learning algorithm |
CN107403201A (en) * | 2017-08-11 | 2017-11-28 | 强深智能医疗科技(昆山)有限公司 | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method |
CN107563434A (en) * | 2017-08-30 | 2018-01-09 | 山东大学 | A kind of brain MRI image sorting technique based on Three dimensional convolution neutral net, device |
CN107748900A (en) * | 2017-11-08 | 2018-03-02 | 山东财经大学 | Tumor of breast sorting technique and device based on distinction convolutional neural networks |
CN107767378A (en) * | 2017-11-13 | 2018-03-06 | 浙江中医药大学 | The multi-modal Magnetic Resonance Image Segmentation methods of GBM based on deep neural network |
CN108257134A (en) * | 2017-12-21 | 2018-07-06 | 深圳大学 | Nasopharyngeal Carcinoma Lesions automatic division method and system based on deep learning |
CN108537773A (en) * | 2018-02-11 | 2018-09-14 | 中国科学院苏州生物医学工程技术研究所 | Intelligence auxiliary mirror method for distinguishing is carried out for cancer of pancreas and pancreas inflammatory disease |
CN109242860A (en) * | 2018-08-21 | 2019-01-18 | 电子科技大学 | Based on the brain tumor image partition method that deep learning and weight space are integrated |
CN109214451A (en) * | 2018-08-28 | 2019-01-15 | 北京安德医智科技有限公司 | A kind of classification method and equipment of brain exception |
CN109242839A (en) * | 2018-08-29 | 2019-01-18 | 上海市肺科医院 | A kind of good pernicious classification method of CT images Lung neoplasm based on new neural network model |
CN109447088A (en) * | 2018-10-16 | 2019-03-08 | 杭州依图医疗技术有限公司 | A kind of method and device of breast image identification |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110600107A (en) * | 2019-08-29 | 2019-12-20 | 上海联影智能医疗科技有限公司 | Method for screening medical images, computer device and readable storage medium |
CN110600107B (en) * | 2019-08-29 | 2022-07-26 | 上海联影智能医疗科技有限公司 | Method for screening medical images, computer device and readable storage medium |
CN110889836A (en) * | 2019-11-22 | 2020-03-17 | 中国人民解放军第四军医大学 | Image data analysis method and device, terminal equipment and storage medium |
CN112085113A (en) * | 2020-09-14 | 2020-12-15 | 四川大学华西医院 | Severe tumor image recognition system and method |
CN112085113B (en) * | 2020-09-14 | 2021-05-04 | 四川大学华西医院 | Severe tumor image recognition system and method |
CN112927799A (en) * | 2021-04-13 | 2021-06-08 | 中国科学院自动化研究所 | Life cycle analysis system fusing multi-example learning and multi-task depth imaging group |
CN112927799B (en) * | 2021-04-13 | 2023-06-27 | 中国科学院自动化研究所 | Life analysis system integrating multi-example learning and multi-task depth image histology |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10733788B2 (en) | Deep reinforcement learning for recursive segmentation | |
Onofrey et al. | Generalizable multi-site training and testing of deep neural networks using image normalization | |
Palomera-Perez et al. | Parallel multiscale feature extraction and region growing: application in retinal blood vessel detection | |
CN111008984B (en) | Automatic contour line drawing method for normal organ in medical image | |
CN107563434B (en) | Brain MRI image classification method and device based on three-dimensional convolutional neural network | |
CN110391014A (en) | Utilize the medical image acquisition for the sequence prediction for using deep learning | |
CN109949288A (en) | Tumor type determines system, method and storage medium | |
Choi et al. | Convolutional neural network-based MR image analysis for Alzheimer’s disease classification | |
Nigri et al. | Explainable deep CNNs for MRI-based diagnosis of Alzheimer’s disease | |
CN112819076A (en) | Deep migration learning-based medical image classification model training method and device | |
US9773325B2 (en) | Medical imaging data processing apparatus and method | |
EP3703007B1 (en) | Tumor tissue characterization using multi-parametric magnetic resonance imaging | |
Aranguren et al. | Improving the segmentation of magnetic resonance brain images using the LSHADE optimization algorithm | |
CN108629785B (en) | Three-dimensional magnetic resonance pancreas image segmentation method based on self-learning | |
Fantini et al. | Automatic detection of motion artifacts on MRI using Deep CNN | |
CN112885453A (en) | Method and system for identifying pathological changes in subsequent medical images | |
KR20230059799A (en) | A Connected Machine Learning Model Using Collaborative Training for Lesion Detection | |
US20150356733A1 (en) | Medical image processing | |
van Opbroek et al. | Weighting training images by maximizing distribution similarity for supervised segmentation across scanners | |
CN112200802A (en) | Training method of image detection model, related device, equipment and storage medium | |
Archa et al. | Segmentation of brain tumor in MRI images using CNN with edge detection | |
CN114332132A (en) | Image segmentation method and device and computer equipment | |
Khademi et al. | Whole volume brain extraction for multi-centre, multi-disease FLAIR MRI datasets | |
Li et al. | BrainK for structural image processing: creating electrical models of the human head | |
Liu et al. | Automated classification and measurement of fetal ultrasound images with attention feature pyramid network |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190628 |
|
RJ01 | Rejection of invention patent application after publication |