CN110120052A - A kind of target area image segmenting system and device - Google Patents
A kind of target area image segmenting system and device Download PDFInfo
- Publication number
- CN110120052A CN110120052A CN201910395463.5A CN201910395463A CN110120052A CN 110120052 A CN110120052 A CN 110120052A CN 201910395463 A CN201910395463 A CN 201910395463A CN 110120052 A CN110120052 A CN 110120052A
- Authority
- CN
- China
- Prior art keywords
- target area
- image
- history
- information
- parted pattern
- 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
Classifications
-
- 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
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- 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
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a kind of target area image segmenting system and devices.The system may include: processor, be configured to: the image of target object is obtained, and, the information for the target area to be split for including in image；At least one target area parted pattern is determined according to the information of target area to be split, and each target area parted pattern at least one target area parted pattern is for dividing different subregions；Image is separately input into each target area parted pattern, obtains the segmented image of each subregion, and determines the target area segmented image of target object according to the segmented image of each sub-regions.The technical solution of the embodiment of the present invention, it can be according to each target area parted pattern of information flexible combination of target area to be split, more quickly and accurately to obtain target area segmented image, solve the problems, such as doctor on the image manual drawing target outline when bring process it is cumbersome and take a long time.
Description
Technical field
The present embodiments relate to technical field of image processing more particularly to a kind of target area image segmenting systems and device.
Background technique
Radiotherapy, chemotherapy, operative treatment etc. are the main means of oncotherapy.Particularly, with above-mentioned hand
The accuracy of the closely related Target delineations of section will directly affect therapeutic effect.Due to every sufferer target area otherness compared with
Greatly, doctor's manual drawing target outline on target area image is needed, but it is excessively cumbersome and take a long time to delineate process.
Summary of the invention
The embodiment of the invention provides a kind of target area image segmenting system and devices, can be according to the target area of target object
Specific auto divide image obtains the target area segmented image of target object.
In a first aspect, may include: the embodiment of the invention provides a kind of target area image segmenting system
Processor is configured to:
The image of target object is obtained, and, the information for the target area to be split for including in image；
At least one target area parted pattern is determined according to the information of target area to be split, at least one target area parted pattern
Each target area parted pattern is for dividing different subregions；
Image is separately input into each target area parted pattern, obtains the segmented image of each subregion, and according to each
The segmented image of subregion determines the target area segmented image of target object.
Optionally, at least one target area parted pattern is determined according to the information of target area to be split, may include:
Corresponding relationship is determined according to the first information in the information of target area to be split, wherein corresponding relationship includes to be split
The corresponding relationship of the second information and target area parted pattern in the information of target area；
At least one target area parted pattern is determined according to the second information and corresponding relationship.
Optionally, on the basis of above system, the processor in the system be can be configured to:
The history image of history object is obtained, and, the history segmented image of each history subregion in history object, and
Using history image and history segmented image as one group of training sample；
Original neural network model is trained based on multiple training samples, to obtain the target area of each history subregion
Parted pattern.
Optionally, obtain history object in each history subregion history segmented image, may include:
According to the user received delineating in history image as a result, and, each history subregion of history object,
The history segmented image of each history subregion is obtained respectively.
Optionally, obtain history object in each history subregion history segmented image, may include:
Based on preset map automatic division method and/or preset artificial intelligence automatic division method, obtains go through respectively
The history segmented image of each history subregion in history object.
Optionally, the information for obtaining target area to be split for including in image is obtained, may include:
At least one of information extracting method, automatic identifying method of default medical image are recorded based on default radiotherapy,
Obtain the information for the target area to be split for including in image.
Optionally, the information of target area to be split may include tumor grade, neoplasm staging, at least one in staging
It is a.
Second aspect, the embodiment of the invention also provides a kind of target area image segmenting device, the devices can include:
Data acquisition module, for obtaining the image of target object, and, the letter for the target area to be split for including in image
Breath；
Model determining module, for determining at least one target area parted pattern, at least one according to the information of target area to be split
Each target area parted pattern in a target area parted pattern is for dividing different subregions；
Image segmentation module obtains point of each subregion for image to be separately input into each target area parted pattern
Image is cut, and determines the target area segmented image of target object according to the segmented image of each sub-regions.
Optionally, model determining module can specifically include:
Corresponding relationship determination unit determines corresponding relationship for the first information in the information according to target area to be split, right
It should be related to the corresponding relationship of the second information and target area parted pattern in the information including target area to be split；
Model determination unit, for determining at least one target area parted pattern according to the second information and corresponding relationship.
Optionally, on the basis of above-mentioned apparatus, which may also include that
Training sample generation module, for obtaining the history image of history object, and, each history in history object
The history segmented image in region, using history image and history segmented image as one group of training sample；
Model obtains module, every to obtain for being trained based on multiple training samples to original neural network model
The target area parted pattern of a history subregion.
The technical solution of the embodiment of the present invention, firstly, target area to be split is divided at least one subregion, each sub-district
Domain is corresponding with the information of respective target area parted pattern and target area to be split；Then, according in the image and image got
The information for the target area to be split for including can determine the associated each sub-regions of the image；In turn, according to subregion and target area point
The relevance of model is cut, can determine the associated each target area parted pattern of the image；Finally, the image is separately input into often
A target area parted pattern, obtains the segmented image of each subregion, to obtain the target area segmented image of target object.Above-mentioned technology
The each subregion Independent modeling of scheme, to improve the accuracy of sub-district regional partition；Furthermore, it is possible to according to the letter of target area to be split
The each target area parted pattern of flexible combination is ceased, more quickly and accurately to obtain target area segmented image, efficiently solves doctor
Teacher on the image manual drawing target outline when bring process it is cumbersome and the problem of take a long time.
Detailed description of the invention
Fig. 1 is the stream for the step of processor in one of embodiment of the present invention one target area image segmenting system is configured
Cheng Tu；
Fig. 2 is the stream for the step of processor in one of embodiment of the present invention two target area image segmenting system is configured
Cheng Tu；
Fig. 3 is the structural block diagram of one of embodiment of the present invention three target area image segmenting device.
Specific embodiment
Invention is further described in detail with reference to the accompanying drawings and examples.It is understood that described herein
Specific embodiment be used only for explaining the present invention rather than limiting the invention.It also should be noted that for the ease of
It describes, only the parts related to the present invention are shown rather than entire infrastructure in attached drawing.
Before introducing the embodiment of the present invention, be first illustrated to the application scenarios of the embodiment of the present invention: the present invention is implemented
The target area to be split that example is related to can be target of prophylactic radiotherapy, target surgical, chemotherapy target area etc., at this with putting in radiotherapy
For treatment target area.Doctor can directly delineate in the image of the patient got and put according to the information of the target of prophylactic radiotherapy of patient
Which region treatment target area, usually tumour invade, and then which region doctor delineates.On this basis, it can attempt doctor
The method for delineating target of prophylactic radiotherapy is combined with existing neural network model, i.e., the target of prophylactic radiotherapy sketched out doctor in the picture
As training sample, existing neural network model is trained, obtains the target area parted pattern for fractionation radiotherapy target area.
However, the specificity of the target of prophylactic radiotherapy of every patient is stronger, the tumor disease that such as every patient suffers from may be different,
The case where target of prophylactic radiotherapy of each patient with identical tumor disease, is also likely to be present difference, therefore, if simply basis
Doctor in the picture delineate result as target area parted pattern obtained by training sample, it is difficult to guarantee target area segmentation it is accurate
Property, subsequent possible needs are largely corrected.
In order to further increase the accuracy of target area segmentation, the target area that each tumor types may relate to can be split
For multiple subregions, specifically, target area can be carried out dissection classification according to the anatomical structure for the concrete position that tumour is related to
To obtain each anatomical categories, and then at least two anatomical categories groups can be combined into a sub-regions with integrative medicine experience,
One anatomical categories can be split as at least two subregions.For example, by taking the radical radiation therapy target area of cervical carcinoma as an example, it can
To be split as CTV1 (Clinical Target Volume 1, be properly termed as clinical target area 1), CTV2 (Clinical Target
Volume 2 is properly termed as clinical target area 2), entire vagina, multiple subregions such as common iliac lymph nodes region.It in turn, can will be every
Sub-regions delineate data as training sample, to carry out stand-alone training to each subregion, obtain corresponding for dividing
The target area parted pattern of subregion, to enhance the robustness and accuracy of each target area parted pattern.
Embodiment one
Fig. 1 is the step of processor in a kind of target area image segmenting system provided in the embodiment of the present invention one is configured
Flow chart.The present embodiment is applicable to the case where target area in segmented image, is particularly suitable for each in segmented image respectively
The case where a target area.This method can be executed by target area image segmenting device provided in an embodiment of the present invention, which can be with
It is realized by the mode of software and/or hardware, which can be configured at target area image segmenting system provided in an embodiment of the present invention
Processor in.
Referring to Fig. 1, target area image segmenting system provided in an embodiment of the present invention may include processor, which can be with
It is configured to following steps:
S110, the image for obtaining target object, and, the information for the target area to be split for including in image.
Wherein, what target object can be subject includes the region of target area to be split, that is, the image got can be with
Be subject include target area to be split region image.On the one hand, the image got can be positron emission
Computed tomography image, magnetic resonance image, CT scan image, ultrasound image etc.；On the other hand, it gets
Image can be target object and be scanned the image obtained in real time afterwards, the image obtained from image data base etc..
Since doctor can directly delineate the target area in image according to the information of the target area to be split of target object, then herein
The information for the target area to be split for including in image, such as tumor grade, neoplasm staging, staging etc. can also be obtained.Tool
Body, staging can be the classification of tumour, such as cervical carcinoma, breast cancer；Tumor grade can be the rank of tumour, such as
In early days, mid-term, advanced stage etc.；Neoplasm staging may include that tumour invades range, lymph node involvement information etc..Optionally,
At least one of information extracting method, automatic identifying method of default medical image can be recorded based on default radiotherapy, obtained
The information for the target area to be split for including in image.The automatic identifying method of the default medical image can be preset electronic calculating
Machine tomoscan image automatic identifying method, default nuclear magnetic resonance image automatic identifying method, default positron emission computer
Fault image automatic identifying method, default ultrasonic image automatic identifying method etc..
S120, at least one target area parted pattern is determined according to the information of target area to be split, mould is divided at least one target area
Each target area parted pattern in type is for dividing different subregions.
Wherein, as previously mentioned, if the target area that each tumor types may relate to is split as at least one son
Region, i.e., the target area to be split that may relate to each tumor types are split as at least one subregion, then each subregion
It is corresponding with respective target area parted pattern.If the information of target area to be split can be associated with target area parted pattern, basis
The information of target area to be split can directly determine at least one target area parted pattern that the image is related to；If each subregion
Can be associated with the information of one or more target areas to be split, then the image institute can be determined according to the information of target area to be split
At least one subregion being related to, and then can determine at least one target area parted pattern involved by the image.It needs
Illustrate, each subregion can be applied in the tumour of different stagings.
Under normal conditions, the range that tumour may invade single anatomical structure incessantly, in other words, the tumors invading of different phase
And range may be different.Therefore, in one embodiment, the target area that different types of tumour may invade can be split as to
Few two sub-regions, each subregion are corresponding with respective target area parted pattern.In use, according to the information of target area to be split
Determine at least two target area parted patterns.
S130, image is separately input into each target area parted pattern, obtains the segmented image of each subregion, and according to
The segmented image of each sub-regions determines the target area segmented image of target object.
Wherein, after determining at least one target area parted pattern according to the information of target area to be split, image can be distinguished
It is input in each target area parted pattern, then each target area parted pattern can export son corresponding with the target area parted pattern
The segmented image in region.In turn, the target area segmented image of target object, example can be determined according to the segmented image of each sub-regions
Such as, it can directly combine the segmented image of each sub-regions, the segmented image of each sub-regions can also be reprocessed,
To obtain target area segmented image.
The technical solution of the embodiment of the present invention, firstly, target area to be split is divided at least one subregion, each sub-district
Domain is corresponding with the information of respective target area parted pattern and target area to be split；Then, according in the image and image got
The information for the target area to be split for including can determine the associated each sub-regions of the image；In turn, according to subregion and target area point
The relevance of model is cut, can determine the associated each target area parted pattern of the image；Finally, the image is separately input into often
A target area parted pattern, obtains the segmented image of each subregion, to obtain the target area segmented image of target object.Above-mentioned technology
The each subregion Independent modeling of scheme, to improve the accuracy of sub-district regional partition；Furthermore, it is possible to according to the letter of target area to be split
The each target area parted pattern of flexible combination is ceased, more quickly and accurately to obtain target area segmented image, efficiently solves doctor
Teacher on the image manual drawing target outline when bring process it is cumbersome and the problem of take a long time.
A kind of optional technical solution determines at least one target area parted pattern according to the information of target area to be split, specifically
It may include: that the first information in the information according to target area to be split determines corresponding relationship, wherein corresponding relationship includes to be split
The corresponding relationship of the second information and target area parted pattern in the information of target area；It is determined at least according to the second information and corresponding relationship
One target area parted pattern.
Wherein, the first information can be staging, i.e., can determine the tumour that the image is related to according to the first information
Classification etc.；And the second information can be tumor grade, neoplasm staging etc.；The second of some tumor types is recorded in corresponding relationship
The corresponding relationship of information and each target area parted pattern, i.e. at least one second information can be at least one target area parted patterns
Corresponding, for example, second information can determine one or more target area parted patterns, multiple second information can determine one
Target area parted pattern etc..So, the staging of the image can be determined according to the first information, and then can be determined and the tumour
Classification associated corresponding relationship；At least one target area parted pattern can be directly determined according to the corresponding relationship and the second information.
Particularly, above-mentioned corresponding relationship can then pass through the sides such as inquiry database, table by forms such as database, tables to present
Formula can determine target area parted pattern.
The specific implementation process of above-mentioned steps in order to better understand, carries out by taking the radical radiation therapy target area of cervical carcinoma as an example
Explanation.Illustratively, which can be split as multiple subregions as shown in Table 1, for ease of description, each sub-district
Domain can be numbered by English alphabet, and each subregion is corresponding with respective target area parted pattern, such as subregion A
Corresponding to target area parted pattern A1；Target of prophylactic radiotherapy, that is, cervical carcinoma corresponding relationship can be as shown in table 2.
On this basis, if the second information of the image got is cervical carcinoma+upper end vagina infringement+lymph node yin
Property, it can determine that with the number of the subregion of second information association be A+B+D+F, and then can determine associated with the image
The number of target area parted pattern is A1, B1, D1 and F1；Image is separately input into target area parted pattern A1, target area parted pattern
B1, target area parted pattern D1 and target area parted pattern F1 respectively obtain the segmented image A2 of subregion A, the segmentation figure of subregion B
It is carried out as the segmented image F2 of B2, the segmented image D2 of subregion D and subregion F, and by the subregion in each segmented image
Combination obtains target area segmented image.
| The number of subregion | The description of subregion |
| A | GTV1 (including primary tumo(u)r, entire uterus and uterine neck) |
| B | GTV2 (including tissue, uterosacral ligament and proximal segment vagina by palace) |
| C | Upper 1/2 vagina |
| D | 2/3 vagina of proximal segment |
| E | Entire vagina |
| F | Ilium is outer, ilium is interior and obturator lymph nodes region |
| G | Common iliac lymph nodes region |
| H | Pelvic cavity and aorta side elongated area |
Table 1
Table 2
Embodiment two
Fig. 2 is the step of processor in a kind of target area image segmenting system provided in the embodiment of the present invention two is configured
Flow chart.The present embodiment is optimized based on above-mentioned each technical solution.In the present embodiment, optionally, above-mentioned processing
Device can be configured to: the history image of history object is obtained, and, the history of each history subregion point in history object
Image is cut, and using history image and history segmented image as one group of training sample；Based on multiple training samples to original nerve
Network model is trained, to obtain the target area parted pattern of each history subregion.Wherein, identical as the various embodiments described above or
Details are not described herein for the explanation of corresponding term.
As shown in Fig. 2, the processor of the present embodiment may be configured to following steps:
S210, the history image for obtaining history object, and, the history segmentation figure of each history subregion in history object
Picture, and using history image and history segmented image as one group of training sample.
Wherein, what history object may be considered history subject includes the region of history target area to be split, and each
A history object can have identical staging, can also not have identical staging.Because of the embodiment of the present invention
In target area parted pattern be it is corresponding with history subregion, since each history subregion can be applied to different stagings
Tumour in, then the staging whether having the same of the history object in training sample is limited without specific herein.Certainly,
If the target area parted pattern of some history subregion in order to obtain is only related to the tumour of some staging, can limit
The staging for determining each history object in training sample is identical.
After obtaining the staging of history object, each history in history object can be determined based on preceding solution
Subregion, and then can determine the history segmented image of each history subregion.By history image and corresponding with the history image
A history subregion history segmented image as one group of training sample, to obtain the target area segmentation of each history subregion
Model.
Optionally, can according to the user received delineating in history image as a result, and, history object it is each
History subregion obtains the history segmented image of each history subregion respectively；Alternatively, being based on the preset map side of segmentation automatically
Method and/or preset artificial intelligence automatic division method obtain the history segmentation of each history subregion in history object respectively
Image.Specifically, doctor can invade range manual drawing target outline on the image according to the tumour of patient, i.e., doctor be not by
Target delineations are carried out according to each history subregion.At this point, delineate result and each history subregion may and it is non-critical
It is corresponding.Therefore, it can be obtained and each history according to the positional relationship for delineating result and each history subregion received
Region is corresponding to be delineated as a result, with the history segmented image of each history subregion of determination.Furthermore it is also possible to based on having instructed
Experienced subregion parted pattern is split history image, obtains the history segmented image of subregion.
S220, original neural network model is trained based on multiple training samples, to obtain each history subregion
Target area parted pattern.
Wherein, using the history image in training sample as input, history segmented image is as output, to original nerve net
Network model is trained, and obtains the corresponding target area parted pattern of the history subregion.It similarly, can also be by going through in training sample
History image carries out original neural network model as output as input, the history segmented image of another history subregion
Training, obtains the target area parted pattern of another history subregion.It moves in circles, until training the history of the staging
The corresponding target area parted pattern of each history subregion of object.Above-mentioned original neural network model can be according to expected
Target area parted pattern determines, for example, original neural network model can be convolutional neural networks model.
S230, the image for obtaining target object, and, the information for the target area to be split for including in image.
S240, at least one target area parted pattern is determined according to the information of target area to be split, mould is divided at least one target area
Each target area parted pattern in type is for dividing different subregions.
S250, image is separately input into each target area parted pattern, obtains the segmented image of each subregion, and according to
The segmented image of each sub-regions determines the target area segmented image of target object.
The technical solution of the embodiment of the present invention, by by history image and a history sub-district corresponding with the history image
The history segmented image in domain is independently trained each history subregion as training sample, to obtain each history sub-district
The corresponding target area parted pattern in domain improves the essence of target area segmentation so that the specific aim of the segmentation of each sub-regions is stronger
Parasexuality.
Embodiment three
Fig. 3 is the structural block diagram for the target area segmented image device that the embodiment of the present invention three provides, which is configured at processing
In device, the processor is configured in target area image segmenting system provided by above-mentioned any embodiment, then the device can be used
The step of processor in execution target area image segmenting system is configured.The device and the target area image of the various embodiments described above point
The system of cutting belongs to the same inventive concept, the detail content of not detailed description in the embodiment of target area image segmenting device, can
With the embodiment with reference to above-mentioned target area image segmenting system.Referring to Fig. 3, the device is specific can include: data acquisition module 310,
Model determining module 320 and image segmentation module 330.
Wherein, data acquisition module 310, for obtaining the image of target object, and, the target to be split for including in image
The information in area；
Model determining module 320, for determining at least one target area parted pattern according to the information of target area to be split, at least
Each target area parted pattern in one target area parted pattern is for dividing different subregions；
Image segmentation module 330 obtains each subregion for image to be separately input into each target area parted pattern
Segmented image determines the target area segmented image of target object according to the segmented image of each sub-regions.
Optionally, model determining module 320, can specifically include:
Corresponding relationship determination unit determines corresponding relationship for the first information in the information according to target area to be split, right
It should be related to the corresponding relationship of the second information and target area parted pattern in the information including target area to be split；
Model determination unit, for determining at least one target area parted pattern according to the second information and corresponding relationship.
Optionally, on the basis of above-mentioned apparatus, which may also include that
Training sample generation module, for obtaining the history image of history object, and, each history in history object
The history segmented image in region, using history image and history segmented image as one group of training sample；
Model obtains module, every to obtain for being trained based on multiple training samples to original neural network model
The target area parted pattern of a history subregion.
Optionally, training sample generation module specifically can be used for:
According to the user received delineating in history image as a result, and, each history subregion of history object,
The history segmented image of each history subregion is obtained respectively.
Optionally, training sample generation module specifically can be used for:
Based on preset map automatic division method and/or preset artificial intelligence automatic division method, obtains go through respectively
The history segmented image of each history subregion in history object.
Optionally, data acquisition module 310 specifically can be used for:
At least one of information extracting method, automatic identifying method of default medical image are recorded based on default radiotherapy,
Obtain the information for the target area to be split for including in image.
Optionally, the information of target area to be split may include tumor grade, neoplasm staging, at least one in staging
It is a.
The embodiment of the present invention three provide target area image segmenting device, by data acquisition module it is available to image with
And the information for the target area to be split in image including；Model determining module can determine the associated each sub-regions of the image, and
According to the relevance of subregion and target area parted pattern, the associated each target area parted pattern of the image can be determined；Image point
The segmented image of the available each subregion of module is cut, and then obtains the target area segmented image of target object.Above-mentioned apparatus
Each subregion can be with Independent modeling, to improve the accuracy of sub-district regional partition；Furthermore, it is possible to according to the information of target area to be split
The each target area parted pattern of flexible combination efficiently solves doctor more quickly and accurately to obtain target area segmented image
Bring process is cumbersome when manual drawing target outline on the image and the problem of taking a long time.
Target area image segmenting device provided by the embodiment of the present invention can execute provided by any embodiment of the invention
The step of processor in target area image segmenting system is configured has the corresponding functional module of execution method and beneficial effect.
It is worth noting that, included each unit and module are only in the embodiment of above-mentioned target area image segmenting device
It is to be divided according to the functional logic, but be not limited to the above division, as long as corresponding functions can be realized；Separately
Outside, the specific name of each functional unit is also only for convenience of distinguishing each other, the protection scope being not intended to restrict the invention.
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 target area image segmenting system characterized by comprising
Processor is configured to:
The image of target object is obtained, and, the information for the target area to be split for including in described image；
At least one target area parted pattern, at least one described target area parted pattern are determined according to the information of the target area to be split
In each target area parted pattern for dividing different subregions；
Described image is separately input into each target area parted pattern, obtains the segmented image of each subregion, and
The target area segmented image of the target object is determined according to the segmented image of each subregion.
2. system according to claim 1, which is characterized in that described to be determined at least according to the information of the target area to be split
One target area parted pattern, comprising:
Corresponding relationship is determined according to the first information in the information of the target area to be split, wherein the corresponding relationship includes institute
State the corresponding relationship of the second information and target area parted pattern in the information of target area to be split；
At least one described target area parted pattern is determined according to second information and the corresponding relationship.
3. system according to claim 1, which is characterized in that the processor is further configured to:
The history image of history object is obtained, and, the history segmented image of each history subregion in the history object, and
Using the history image and the history segmented image as one group of training sample；
Original neural network model is trained based on multiple training samples, to obtain each history subregion
The target area parted pattern.
4. system according to claim 3, which is characterized in that each history subregion goes through in the acquisition history object
History segmented image, comprising:
According to the user received delineating in the history image as a result, and, each history of history object
Region obtains the history segmented image of each history subregion respectively.
5. system according to claim 3, which is characterized in that each history subregion goes through in the acquisition history object
History segmented image, comprising:
Based on preset map automatic division method and/or preset artificial intelligence automatic division method, history pair is obtained respectively
The history segmented image of each history subregion as in.
6. system according to claim 1, which is characterized in that described to obtain the acquisition target to be split for including in described image
The information in area, comprising:
Based at least one of default radiotherapy record information extracting method, the automatic identifying method of default medical image, obtain
The information for the target area to be split for including in described image.
7. system according to claim 1, which is characterized in that the information of the target area to be split includes tumor grade, swells
Tumor by stages, at least one of staging.
8. a kind of target area image segmenting device, which is characterized in that described device is configured in processor, and described device includes:
Data acquisition module, for obtaining the image of target object, and, the letter for the target area to be split for including in described image
Breath；
Model determining module, for determining at least one target area parted pattern according to the information of the target area to be split, it is described extremely
Each target area parted pattern in a few target area parted pattern is for dividing different subregions；
Image segmentation module obtains each son for described image to be separately input into each target area parted pattern
The segmented image in region, and determine according to the segmented image of each subregion the target area segmented image of the target object.
9. device according to claim 8, which is characterized in that the model determining module includes:
Corresponding relationship determination unit determines corresponding relationship for the first information in the information according to the target area to be split,
In, the corresponding relationship includes the corresponding relationship of the second information and target area parted pattern in the information of the target area to be split；
Model determination unit, for determining that mould is divided at least one described target area according to second information and the corresponding relationship
Type.
10. device according to claim 8, which is characterized in that described device further include:
Training sample generation module, for obtaining the history image of history object, and, each history in the history object
The history segmented image in region, and using the history image and the history segmented image as one group of training sample；
Model obtains module, every to obtain for being trained based on multiple training samples to original neural network model
The target area parted pattern of a history subregion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910395463.5A CN110120052B (en) | 2019-05-13 | 2019-05-13 | Target area image segmentation system and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910395463.5A CN110120052B (en) | 2019-05-13 | 2019-05-13 | Target area image segmentation system and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110120052A true CN110120052A (en) | 2019-08-13 |
| CN110120052B CN110120052B (en) | 2021-05-07 |
Family
ID=67522129
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910395463.5A Active CN110120052B (en) | 2019-05-13 | 2019-05-13 | Target area image segmentation system and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110120052B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113633306A (en) * | 2021-08-31 | 2021-11-12 | 上海商汤智能科技有限公司 | Image processing method and related device, electronic equipment and storage medium |
| CN114022491A (en) * | 2021-10-27 | 2022-02-08 | 安徽医科大学 | Small data set esophageal cancer target area image automatic delineation method based on improved spatial pyramid model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120320055A1 (en) * | 2010-03-11 | 2012-12-20 | Koninklijke Philips Electronics N.V. | Probabilistic refinement of model-based segmentation |
| CN104036109A (en) * | 2014-03-14 | 2014-09-10 | 上海大图医疗科技有限公司 | Image based system and method for case retrieving, sketching and treatment planning |
| CN106845138A (en) * | 2017-02-28 | 2017-06-13 | 成都金盘电子科大多媒体技术有限公司 | Method is previewed before a kind of surgery |
| CN106887039A (en) * | 2017-02-28 | 2017-06-23 | 成都金盘电子科大多媒体技术有限公司 | The three-D imaging method and system of a kind of organ and focus based on medical image |
| CN106898044A (en) * | 2017-02-28 | 2017-06-27 | 成都金盘电子科大多媒体技术有限公司 | It is a kind of to be split and operating method and system based on medical image and using the organ of VR technologies |
| CN108447551A (en) * | 2018-02-09 | 2018-08-24 | 北京连心医疗科技有限公司 | A kind of automatic delineation method in target area based on deep learning, equipment and storage medium |
-
2019
- 2019-05-13 CN CN201910395463.5A patent/CN110120052B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120320055A1 (en) * | 2010-03-11 | 2012-12-20 | Koninklijke Philips Electronics N.V. | Probabilistic refinement of model-based segmentation |
| CN104036109A (en) * | 2014-03-14 | 2014-09-10 | 上海大图医疗科技有限公司 | Image based system and method for case retrieving, sketching and treatment planning |
| CN106845138A (en) * | 2017-02-28 | 2017-06-13 | 成都金盘电子科大多媒体技术有限公司 | Method is previewed before a kind of surgery |
| CN106887039A (en) * | 2017-02-28 | 2017-06-23 | 成都金盘电子科大多媒体技术有限公司 | The three-D imaging method and system of a kind of organ and focus based on medical image |
| CN106898044A (en) * | 2017-02-28 | 2017-06-27 | 成都金盘电子科大多媒体技术有限公司 | It is a kind of to be split and operating method and system based on medical image and using the organ of VR technologies |
| CN108447551A (en) * | 2018-02-09 | 2018-08-24 | 北京连心医疗科技有限公司 | A kind of automatic delineation method in target area based on deep learning, equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| ROBERT KADERKA等: ""Geometric and dosimetric evaluation of atlas based auto-segmentation of cardiac structures in breast cancer patients"", 《RADIOTHERAPY AND ONCOLOGY》 * |
| YU JUN等: ""Unsupervised image segmentation via Stacked Denoising Auto-encoder and hierarchical patch indexing"", 《SIGNAL PROCESSING》 * |
| 单书灿等: ""自动勾画软件对鼻咽癌靶区和危及器官勾画结果对比分析"", 《中国医学装备》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113633306A (en) * | 2021-08-31 | 2021-11-12 | 上海商汤智能科技有限公司 | Image processing method and related device, electronic equipment and storage medium |
| CN114022491A (en) * | 2021-10-27 | 2022-02-08 | 安徽医科大学 | Small data set esophageal cancer target area image automatic delineation method based on improved spatial pyramid model |
| CN114022491B (en) * | 2021-10-27 | 2022-05-10 | 安徽医科大学 | Small data set esophageal cancer target area image automatic delineation method based on improved spatial pyramid model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110120052B (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11455732B2 (en) | Knowledge-based automatic image segmentation | |
| McRackan et al. | Comparison of cochlear implant relevant anatomy in children versus adults | |
| WO2021238438A1 (en) | Tumor image processing method and apparatus, electronic device, and storage medium | |
| CN111862021B (en) | Deep learning-based automatic head and neck lymph node and drainage area delineation method | |
| EP3457356B1 (en) | Method and apparatus for fully automated segmentation of joint using patient-specific optimal thresholding and watershed algorithm | |
| WO2009013636A2 (en) | A method for automatic boundary segmentation of object in 2d and/or 3d image | |
| CN110929728B (en) | Image region-of-interest dividing method, image segmentation method and device | |
| Tunçay et al. | Realistic microwave breast models through T1-weighted 3-D MRI data | |
| US20040101184A1 (en) | Automatic contouring of tissues in CT images | |
| Liu et al. | Automatic model-guided segmentation of the human brain ventricular system from CT images | |
| CN111833293A (en) | Method and data processing system for providing lymph node information | |
| US20150302603A1 (en) | Knowledge-based automatic image segmentation | |
| CN113469945B (en) | Sacral nerve modeling method and system based on CT and MRI imaging fusion | |
| CN110120052A (en) | A kind of target area image segmenting system and device | |
| US10621728B2 (en) | Internal organ localization in computed tomography (CT) images | |
| US10987020B2 (en) | Method for tissue classification, computer program product and magnetic resonance imaging system | |
| Gare et al. | Multi-atlas segmentation of the facial nerve from clinical CT for virtual reality simulators | |
| Vasilache et al. | Automated bone segmentation from pelvic CT images | |
| CN114376724B (en) | Method, device and equipment for determining target characteristic points and computer storage medium | |
| CN111127636B (en) | Intelligent complex intra-articular fracture desktop-level three-dimensional diagnosis system | |
| CN113536575A (en) | Organ contour delineation method, medical imaging system and storage medium | |
| Teng et al. | Automatic segmentation of neck CT images | |
| Banik et al. | Delineation of the pelvic girdle in computed tomographic images | |
| CN114998297A (en) | Vertebra numbering method and device, electronic device and storage medium | |
| Björeland | EP-1777: MRT investigation of prostate and lymph nodes movements: implications on planning target volume? |
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 | ||
| CB02 | Change of applicant information |
Address after: 201807 Shanghai City, north of the city of Jiading District Road No. 2258 Applicant after: Shanghai Lianying Medical Technology Co., Ltd Address before: 201807 Shanghai City, north of the city of Jiading District Road No. 2258 Applicant before: SHANGHAI UNITED IMAGING HEALTHCARE Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |