CN109544561A - Cell mask method, system and device - Google Patents
Cell mask method, system and device Download PDFInfo
- Publication number
- CN109544561A CN109544561A CN201811318835.6A CN201811318835A CN109544561A CN 109544561 A CN109544561 A CN 109544561A CN 201811318835 A CN201811318835 A CN 201811318835A CN 109544561 A CN109544561 A CN 109544561A
- Authority
- CN
- China
- Prior art keywords
- frame
- cell
- pathological section
- image
- digital pathological
- 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
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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of cell mask method, system and devices.Wherein, this method comprises: display digital pathological section image, wherein digital pathological section image includes multiple cells；Sink block choosing instruction, wherein frame choosing instruction includes one or more cells to be marked on digital pathological section image in frame favored area for determining a frame favored area on digital pathological section image；It is selected and is instructed according to frame, generate the shape contour of each cell in frame favored area；Receive the markup information to the cell addition in any one shape contour.The present invention realizes dynamic and shows data markers as a result, to enhance the technical effect that user waits the user experience in data output process.
Description
Technical field
The present invention relates to digitlization pathology fields, in particular to a kind of cell mask method, system and device.
Background technique
When analyzing pathological section, it is often necessary to be labeled to the cell of lesion region on pathological section.It is existing
Have technology when being labeled to cell, there are mainly two types of modes: the first is to draw the frames choosing figures such as rectangle using drawing tool
Shape frame selects some region on pathological section to be labeled to realize to the cell in the region；Second is manually with a point phase
Mode even describes cell outline to realize the mark to cell.
Analysis is it is found that the first notation methods can not accurately be labeled cell；Second of notation methods is manually retouched
Cell outline is drawn, heavy workload takes time and effort.
Aiming at the problem that the above-mentioned prior art cannot achieve and quickly precisely be marked to cell, not yet propose at present effective
Solution.
Summary of the invention
The embodiment of the invention provides a kind of cell mask method, system and devices, can not at least to solve the prior art
The technical issues of realization quickly precisely marks cell.
According to an aspect of an embodiment of the present invention, a kind of cell mask method is provided, comprising: show that digital pathology is cut
Picture, wherein digital pathological section image includes multiple cells；Sink block choosing instruction, wherein frame choosing instruction is in number
A frame favored area is determined on pathological section image, in frame favored area comprising one to be marked on digital pathological section image or
Multiple cells；It is selected and is instructed according to frame, generate the shape contour of each cell in frame favored area；It receives to any one shape contour
The markup information of interior cell addition.
Further, it is selected and is instructed according to frame, generate the shape contour of each cell, comprising: it is corresponding to obtain frame choosing instruction
Frame constituency area image；Based on deep learning neural network model, corresponding probability matrix is generated according to frame constituency area image,
In, the value of each element in probability matrix is general in cell compartment for characterizing each pixel in the area image of frame constituency
Rate；According to probability matrix, the shape contour of each cell is generated.
Further, according to probability matrix, the shape contour of each cell is generated, comprising: normalizing is carried out to probability matrix
Change processing；Based on the noise in form closed operation removal display foreground；Image is subjected to gradient transformation, obtains corresponding gradient
Figure；Based on watershed algorithm, the shape contour of each cell is determined according to gradient map.
Further, it is being based on deep learning neural network model, corresponding probability square is generated according to frame constituency area image
Before battle array, method further include: frame constituency area image is converted into corresponding grayscale image；Grayscale image is carried out at histogram equalization
Reason obtains the image of contrast enhancing；Noise reduction process is carried out to the image that contrast enhances using gaussian filtering, after obtaining noise reduction
Image；The image after noise reduction is standardized by such as minor function, obtains the defeated of deep learning neural network model
Enter image:
Wherein, x*Grey scale pixel value after indicating standardization, x indicate the grey scale pixel value before standardization, μ table
Show that the average gray of grayscale image, σ indicate the gray standard deviation of grayscale image.
Further, instruction is being selected according to frame, generated in frame favored area after the shape contour of each cell, comprising: will
Digital pathological section image is shown on interface centered on the cell in frame favored area.
Further, digital pathological section image is being shown it on interface centered on the cell in frame favored area
Afterwards, comprising: receive scaling instruction, wherein scaling instruction is for zooming in or out the digital pathological section image shown on interface；
It is instructed according to scaling, adjusts the digital pathological section image shown on interface.
According to another aspect of an embodiment of the present invention, a kind of cell labeling system is additionally provided, comprising: headend equipment is used
In display digital pathological section image, and sink block choosing instructs, wherein digital pathological section image includes multiple cells, frame choosing
Instruction includes on digital pathological section image in frame favored area for determining a frame favored area on digital pathological section image
One or more cells to be marked；Background server is communicated with headend equipment, is instructed for being selected according to frame, and frame constituency is generated
The shape contour of each cell in domain；Wherein, headend equipment is also used to receive to the cell addition in any one shape contour
Markup information.
According to another aspect of an embodiment of the present invention, a kind of cell labeling system is additionally provided, comprising: display screen is used for
Show digital pathological section image, wherein digital pathological section image includes multiple cells；Input equipment is selected for sink block
Instruction, wherein frame choosing instruction includes number in frame favored area for determining a frame favored area on digital pathological section image
One or more cells to be marked on pathological section image；Processor is connect with display screen, is instructed for being selected according to frame, raw
At the shape contour of each cell in frame favored area；Wherein, input equipment is also used to receive in any one shape contour
The markup information of cell addition.
According to another aspect of an embodiment of the present invention, a kind of cell annotation equipment is additionally provided, comprising: display module is used
In display digital pathological section image, wherein digital pathological section image includes multiple cells；First receiving module, for connecing
Receive frame choosing instruction, wherein frame choosing instruction on digital pathological section image for determining a frame favored area, packet in frame favored area
One or more cells to be marked on image containing digital pathological section；Generation module is instructed for being selected according to frame, generates frame choosing
The shape contour of each cell in region；Second receiving module, for receiving to the cell addition in any one shape contour
Markup information.
According to another aspect of an embodiment of the present invention, a kind of cell mask method is additionally provided, comprising: the first boundary of display
Face, wherein the first interface is for showing that digital pathological section image, and reception determine one on digital pathological section image
The frame of frame favored area selects instruction, wherein digital pathological section image includes multiple cells, is cut in frame favored area comprising digital pathology
One or more cells to be marked on picture；Show second contact surface, wherein second contact surface is used to select according to frame and instruct, raw
At the shape contour of each cell in frame favored area, and receives and the mark of the cell addition in any one shape contour is believed
Breath.
According to another aspect of an embodiment of the present invention, a kind of storage medium is additionally provided, storage medium includes the journey of storage
Sequence, wherein program executes the cell mask method of above-mentioned any one.
According to another aspect of an embodiment of the present invention, a kind of processor is additionally provided, processor is used to run program,
In, program executes the cell mask method of above-mentioned any one when running.
In embodiments of the present invention, in the interface for showing digital pathological section image, sink block selects digital pathological section
The frame of any one frame favored area selects instruction on image, according to the frame constituency area image that frame selects instruction box to select, automatically generates the frame
The shape contour of each cell in favored area, and then the cell in each shape contour is labeled, reach and has been selected by frame
The mode purpose that automatically generates cell outline, and quickly cell is precisely marked, thus realize enhancing user experience,
The technical effect of working efficiency is improved, and then solves the prior art and cannot achieve the technology quickly precisely marked to cell
Problem.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is a kind of cell mask method flow chart provided according to embodiments of the present invention；
Fig. 2 is a kind of cell image schematic diagram provided according to embodiments of the present invention；
Fig. 3 is a kind of deep learning neural network model based on U-Net structure provided according to embodiments of the present invention
Export result schematic diagram；
Fig. 4 is a kind of operational flowchart of the optional cell mark provided according to embodiments of the present invention；
Fig. 5 is a kind of cell labeling system schematic diagram provided according to embodiments of the present invention；
Fig. 6 is another the cell labeling system schematic diagram provided according to embodiments of the present invention；
Fig. 7 is a kind of cell annotation equipment schematic diagram according to an embodiment of the present invention；And
Fig. 8 is another the cell mask method flow chart provided according to embodiments of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The model that the present invention protects all should belong in member's every other embodiment obtained without making creative work
It encloses.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
According to embodiments of the present invention, a kind of cell mask method embodiment is provided, it should be noted that in the stream of attached drawing
The step of journey illustrates can execute in a computer system such as a set of computer executable instructions, although also, flowing
Logical order is shown in journey figure, but in some cases, it can be to be different from shown or described by sequence execution herein
The step of.
Fig. 1 is a kind of cell mask method flow chart provided according to embodiments of the present invention, as shown in Figure 1, this method packet
Include following steps:
S101 shows digital pathological section image, wherein digital pathological section image includes multiple cells；
S102, sink block choosing instruction, wherein frame choosing instruction on digital pathological section image for determining a frame constituency
Domain, frame favored area is interior to include one or more cells to be marked on digital pathological section image；
S103 is selected according to frame and is instructed, and generates the shape contour of each cell in frame favored area；
S104 receives the markup information to the cell addition in any one shape contour.
It should be noted that the shape of above-mentioned frame favored area include but is not limited to it is following any one: round, ellipse,
Rectangle, square, triangle, polygon etc..
Scheme disclosed in S101 to S104 through the above steps, it can be seen that the present invention only needs to select instruction (example by frame
Such as, the rectangle for clicking drawing tool is drawn, and draws rectangular area on digital pathological section image), frame selects digital pathology to cut
A region on picture, then can automatically generate the shape contour of the region inner cell, to realize that cell outline is automatic
The purpose of mark.
Specifically, above-mentioned steps S103 can specifically include following steps: obtaining frame choosing and instructs corresponding frame favored area figure
Picture；Based on deep learning neural network model, corresponding probability matrix is generated according to frame constituency area image, wherein probability matrix
In each element value for characterizing probability of each pixel in cell compartment in the area image of frame constituency；According to probability square
Battle array, generates the shape contour of each cell.
It should be noted that above-mentioned deep learning neural network model refers to based on any one following deep learning network
The model that structured training obtains, the input data of the model are the image of digital pathological section, and the output data of the model is general
Rate matrix.
It further, can be first to frame constituency in order to enable deep learning neural network model can obtain optimal effectiveness
Area image is pre-processed.Thus, as an alternative embodiment, it is being based on deep learning neural network model, according to
Before frame constituency area image generates corresponding probability matrix, the above method can also include: that frame constituency area image is converted to phase
The grayscale image answered；Histogram equalization processing is carried out to grayscale image, obtains the image of contrast enhancing；Using gaussian filtering to right
Image than degree enhancing carries out noise reduction process, the image after obtaining noise reduction；The image after noise reduction is marked by such as minor function
Quasi-ization processing, obtains the input picture of deep learning neural network model:
Wherein, x*Grey scale pixel value after indicating standardization, x indicate the grey scale pixel value before standardization, μ table
Show that the average gray of grayscale image, σ indicate the gray standard deviation of grayscale image.
Specifically, in the above-described embodiment, frame constituency area image (including the cell image of cell to be marked) is turned
Grayscale image is turned to remove unnecessary colouring information, histogram equalization then is carried out to improve contrast, by force to grayscale image
Change image detail feature.Since neural network is more sensitive to noise, the embodiment of the present invention using gaussian filtering come to image into
Row noise reduction is weighted and averaged image by Gauss noise reduction, and variation of image grayscale can be made more smooth.Then using mark
Standardization function is standardized the image after noise reduction, so that frame favored area gray level image gray value mean value is 0, standard deviation
It is 1, and obeys standardized normal distribution, to accelerates the convergence rate and precision of prediction of model.
After pre-processing to image, pretreated image can be input to trained deep learning nerve net
Network model carries out semantic segmentation, wherein cell compartment is as prospect, acellular region as background.Optionally, the present invention is implemented
Deep learning neural network model in example uses U-Net structure or similar network structure, and U-Net structure is full convolutional Neural net
One modification of network, it use coder-decoder structure, encoder structurally and functionally with common convolutional neural networks one
It causes, is made of a series of convolutional layers and pond layer, for gradually reducing the dimension of output eigenmatrix, to structurally extract
The feature of input picture；Decoder is then to be restored to the dimension for exporting eigenmatrix gradually originally by transposition convolution operation
The dimension of input picture.There is short connection between corresponding encoder and decoder, for mitigating position caused by pondization operates
Characteristic information loss.The last layer output eigenmatrix of U-Net structure is calculated by Sigmoid function, and available one
Probability matrix of the dimension as input picture, the value of each of which element is in the range of [0,1].In simple terms, the probability
Matrix indicates probability of each pixel in cell compartment.
Fig. 2 shows a kind of cell image schematic diagram of offer according to embodiments of the present invention, Fig. 3 is shown according to the present invention
A kind of output result schematic diagram for deep learning neural network model based on U-Net structure that embodiment provides.Cell compartment
The value of interior point can relatively 1 (white area in Fig. 3), the value of the point in acellular region can be relatively 0 (black in Fig. 3
Color region).
It should be noted that the result obtained through U-Net semantic segmentation is only capable of intuitively providing the general profile of cell, it is
The profile of accurate positionin each cell and the profile of quantization cell are the coordinate of series of points, it is also necessary to post-processing algorithm pair
It is advanced optimized.As an alternative embodiment, generating the shape contour tool of each cell according to probability matrix
Body may include: that probability matrix is normalized；Based on the noise in form closed operation removal display foreground；By image
Gradient transformation is carried out, corresponding gradient map is obtained；Based on watershed algorithm, the shaped wheel of each cell is determined according to gradient map
It is wide.
Specifically, normalized can be done to the segmented image of generation by normalizing formula as follows, makes each pixel
Grey value profile between [0,1]:
Wherein, x' indicates that the grey scale pixel value after normalized, x indicate the grey scale pixel value X table before normalized
Show probability matrix.
Closed operation is first to carry out expansive working to image A using structural element B to carry out etching operation again, be can be expressed as:
WhereinWithRespectively indicate expansion and etching operation.In simple terms, expansive working can be such that the prospect in image expands
, and etching operation is then to corrode prospect by background to make prospect shrink.Noise meeting when using closed operation, in prospect
It is first inflated operation to eliminate, before subsequent etching operation is then by the size before the range shorter of prospect to expansion to reach elimination
Scape noise does not change the effect of picture structure again.
Then, grayscale image being converted into gradient map, gradient transformation is the key that a step in many edge detection algorithms, because
Boundary is often positioned in the most violent place of gray-value variation.Image can be regarded as a two-dimensional discrete function, seek image gradient
It is then to this two-dimensional discrete function derivation, the gradient replacement of point (x, y) is then are as follows:
Wherein, G (x, y) indicates the gradient value of point (x, y), and gx and gy respectively indicate point (x, y) in the single order in the direction x and y
Derivative value is obtained in two-dimensional matrix by the difference of calculating adjacent element.
Finally, determining the contour line of cell in turn using watershed algorithm on the basis of gradient map.It is calculated in watershed
In method, grayscale image is conceptualized as the surface of geology, and the high place of gradient is mountain peak, and the low place of gradient is mountain valley.The original of algorithm
Reason is to the water of each the lowest point (local minimum) addition different colours, until the water in adjacent mountain valley starts to merge, in water
Combined place is placed obstacles, and is flooded all mountain peaks (global maximum) until water is added to, and set obstacle constitutes every
The profile in a mountain valley.Over-segmentation in order to prevent, the prospect (cell compartment) of image is marked in we, that is, discharge water Shi Congbiao
The region of note starts to discharge water.Label is obtained by carrying out erosion operation to grayscale image, is said above, erosion operation can reduce prospect
Range, the foreground area through reducing are that comparison determination is to be located at difference into the cell.Thus method, we have finally been partitioned into figure
Cell in piece, and the contour line of each cell has been determined.The contour line of cell is the collection of the pixel on contour line
It is combined into.
In addition, it should also be noted that, in an alternative embodiment, being instructed being selected according to frame, generation frame favored area
After the shape contour of interior each cell, the above method can also include: by digital pathological section image in frame favored area
It is shown on interface centered on cell.
After extracting cellular informatics, cell is clicked in zone list, the page will will be cut according to selected cell
Piece carries out adaptation interface processing, and whole slice is shown centered on the cell marked, and is sliced and is carried out centered on the cell
It moves and adapts to screen size.
Optionally, based on the above embodiment, by digital pathological section image centered on the cell in frame favored area
After showing on interface, the above method can also include: to receive scaling instruction, wherein scaling instruction is for zooming in or out boundary
The digital pathological section image shown on face；It is instructed according to scaling, adjusts the digital pathological section image shown on interface.
Fig. 4 is a kind of operational flowchart of the optional cell mark provided according to embodiments of the present invention, as shown in figure 4,
Include the following steps:
S401, click start to mark.
S402 selects digital pathological section to be marked, comes out slice document retrieval be transmitted to front end from the background, by front end exhibition
Show slice picture, and jumps to the page of mark cell.
Optionally, the embodiment of the present invention can provide the diagosis selection page, which can provide but be not limited to following function
Can: (1) it clicks slice file and enters slice browsing pages；(2) folder name is clicked, can show all slices under this document folder
File or subfile；(3) folder name is double-clicked, may browse through all slice files under this document.In selection digital pathological section
When, digital pathological section can be selected by any one tool that diagosis selects the page to provide.
S403 selects drawing tool to draw cell.It optionally, can also be to slice progress scaling appropriate, a left side
The operation such as move right or pull.Further, on the left of the page, file description can also be carried out to the slice file.
S404 draws frame favored area.
Specifically, when user is using mark cell function, system automatically extracts the cellular informatics in slice, mark
Point centered on the coordinate corresponding to the initial point position information is believed with coordinate corresponding to initial point position information and final position
The corresponding coordinate of breath draws rectangle, determines position in the form of coordinate from the background, and identify which picture the image of rectangular area is
The cell image of plain grade can also retighten if the pixel size of cell is not suitable with computer screen according to standard from the background
The pixel size of cell image is shown with adapting to screen.
S405 carries out profile processing to the cell in frame favored area from the background.
After the rectangular region image of user annotation is transferred to backstage, backstage can carry out noise reduction to image first and locate in advance
Reason, processed picture are input into trained deep neural network model as input data and carry out semantic segmentation.It should
Deep neural network model can generate corresponding initial cell segmented image (probability matrix), post-processing algorithm according to input picture
It is advanced optimized.Post-processing algorithm does normalized to the segmented image of generation first, makes the ash of each pixel
Angle value is seated between [0,1]；Then using the noise in form closed operation mitigation prospect；Finally calculate the segmented image
Shade of gray figure determines the contour line of cell, the contour line of cell in turn on the basis of gradient map using watershed algorithm
It is that the collection of the pixel on the contour line is combined into.
After backstage is split cell, the cell outline of formation is attached with putting and putting, and with each thin
Born of the same parents have a fixing profile and a fixed ID to store in a manner of character string in lists, complete backstage storage operation.In page
Left side of face has delete button, when we click selection region button, and chooses marked cell, then clicks delete button,
Cell will be deleted, and corresponding backstage carries out data delete operation.
S406 detects whether to add markup information to cell；If so, executing S407；Conversely, executing S408.
Optionally, after cellulation profile, right mouse button can be clicked to cell and add remark information, convenient for users to
The overview of cell can be clearly browsed, and is marked according to the characteristic of cell.It is of different sizes due to slice, and slice
The number of middle cell is different, proportionally can be amplified and be reduced by the pair of rollers slice of mouse with click location button,
It is more conducive to user's identification and mark cell.Optionally, delete operation can also be carried out to the cell marked.
S407 adds markup information.
When adding markup information for some cell compartment, it can select to carry out the cell remarks with right button, or
Click directly on left side choose the information blank space shown after cell to fill in, rear end reception content and according to the ID of cell to corresponding
Cell content saved.
S408 completes mark.
Optionally, the embodiment of the present invention can provide slice browsing pages, and the function which provides includes but is not limited to:
(1) current browse webpage is exited, selects the page into diagosis；(2) enter and be sliced macroscopical display pattern page；(3) according to user
To the moving operation of slice, corresponding region is shown；(4) amplify the slice on browsing pages；(5) cutting on browsing pages is reduced
Piece；(6) selected marker region；(7) marked region is created；(8) some marked region is deleted.
Further, the embodiment of the present invention can provide the marked region page, and the function which provides includes but unlimited
In: (1) click storage zone information；(2) previous slice is switched to；(3) latter slice is switched to；(4) operating surface is packed up
Plate；(5) opening operation panel；(6) input slice whole description information.
Under a kind of application scenarios, cell mask method provided in an embodiment of the present invention can be applied to shown in fig. 5 thin
Born of the same parents' labeling system, as shown in figure 5, the system includes: headend equipment 501 and background server 502.
Wherein, headend equipment 501, for showing digital pathological section image, and sink block choosing instruction, wherein number disease
Managing sectioning image includes multiple cells, and frame choosing instruction on digital pathological section image for determining a frame favored area, frame choosing
Include one or more cells to be marked on digital pathological section image in region.
Specifically, above-mentioned headend equipment 501 can be any one client device, including works as and be not limited to mobile phone, plate
Computer, laptop, computer etc..
Background server 502 is communicated with headend equipment 501, is instructed for being selected according to frame, is generated each thin in frame favored area
The shape contour of born of the same parents.
It should be noted that the communication mode of headend equipment 501 and background server 502 can be wired mode, it can also
Wirelessly, it can be local area network communication, be also possible to communication Internet-based.
Optionally, headend equipment 501 is also used to receive the markup information to the cell addition in any one shape contour.
Under another application scenarios, cell mask method provided in an embodiment of the present invention can be applied to shown in fig. 6
Cell labeling system, as shown in fig. 6, the system includes: display screen 601, input equipment 602 and processor 603.
Wherein, display screen 601, for showing digital pathological section image, wherein digital pathological section image includes multiple
Cell；
Input equipment 602 is used for sink block choosing instruction, wherein frame choosing instruction is for true on digital pathological section image
A fixed frame favored area, frame favored area is interior to include one or more cells to be marked on digital pathological section image；
Processor 603 is connect with display screen 601, is instructed for being selected according to frame, and the shape of each cell in frame favored area is generated
Shape profile；
Wherein, input equipment is also used to receive the markup information to the cell addition in any one shape contour.
It should be noted that above-mentioned display screen 601, input equipment 602 and processor 603 can be on one device
Component, can also be in the component on multiple devices.For example, the display of computer, mouse and processor or smart phone
Touch screen (can both make display screen, input equipment can also be made) and processor.
According to embodiments of the present invention, a kind of device implementation for realizing cell mask method shown in FIG. 1 is additionally provided
Example, Fig. 7 is a kind of cell annotation equipment schematic diagram according to an embodiment of the present invention, as shown in fig. 7, the device includes: display mould
Block 701, the first receiving module 702, generation module 703 and the second receiving module 704.
Wherein, display module 701, for showing digital pathological section image, wherein digital pathological section image includes more
A cell；
First receiving module 702, for sink block choosing instruction, wherein frame choosing instruction is in digital pathological section image
One frame favored area of upper determination, frame favored area is interior to include one or more cells to be marked on digital pathological section image；
Generation module 703 is instructed for being selected according to frame, generates the shape contour of each cell in frame favored area；
Second receiving module 704, for receiving the markup information to the cell addition in any one shape contour.
Herein it should be noted that above-mentioned display module 701, the first receiving module 702, generation module 703 and second connect
Receive the step S101 to S104 that module 704 corresponds in embodiment of the method, the example that above-mentioned module is realized with corresponding step
It is identical with application scenarios, but it is not limited to above method embodiment disclosure of that.It should be noted that above-mentioned module is as dress
That sets a part of can execute in a computer system such as a set of computer executable instructions.
According to embodiments of the present invention, a kind of cell mask method is additionally provided, as shown in figure 8, including the following steps:
S801 shows the first interface, wherein the first interface is for showing digital pathological section image, and reception in number
Determining that the frame of a frame favored area selects instruction on word pathological section image, wherein digital pathological section image includes multiple cells,
Include one or more cells to be marked on digital pathological section image in frame favored area；
S802 shows second contact surface, wherein second contact surface is used to be selected according to frame and instruct, and generates each thin in frame favored area
The shape contour of born of the same parents, and receive the markup information to the cell addition in any one shape contour.
According to embodiments of the present invention, a kind of storage medium is additionally provided, storage medium includes the program of storage, wherein journey
Sequence executes the optional or preferred cell mask method of any one of above method embodiment.
According to embodiments of the present invention, a kind of processor is additionally provided, processor is for running program, wherein program operation
The optional or preferred cell mask method of any one of Shi Zhihang above method embodiment.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
In the above embodiment of the invention, it all emphasizes particularly on different fields to the description of each embodiment, does not have in some embodiment
The part of detailed description, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed technology contents can pass through others
Mode is realized.Wherein, the apparatus embodiments described above are merely exemplary, such as the division of the unit, Ke Yiwei
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of unit or module
It connects, can be electrical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
On unit.It can some or all of the units may be selected to achieve the purpose of the solution of this embodiment according to the actual needs.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can for personal computer, server or network equipment etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned includes: that USB flash disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic or disk etc. be various to can store program code
Medium.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (10)
1. a kind of cell mask method characterized by comprising
Show digital pathological section image, wherein the digital pathological section image includes multiple cells；
Sink block choosing instruction, wherein the frame choosing instruction on the digital pathological section image for determining a frame constituency
Domain, the frame favored area is interior to include one or more cells to be marked on the digital pathological section image；
It is selected and is instructed according to the frame, generate the shape contour of each cell in the frame favored area；
Receive the markup information to the cell addition in any one shape contour.
2. instructing the method according to claim 1, wherein selecting according to the frame, the shape of each cell is generated
Profile, comprising:
Obtain the frame choosing instruction corresponding frame constituency area image；
Based on deep learning neural network model, corresponding probability matrix is generated according to frame constituency area image, wherein described
The value of each element in probability matrix is for characterizing probability of each pixel in cell compartment in the area image of frame constituency；
According to the probability matrix, the shape contour of each cell is generated.
3. according to the method described in claim 2, it is characterized in that, generating the shape of each cell according to the probability matrix
Profile, comprising:
The probability matrix is normalized；
Based on the noise in form closed operation removal display foreground；
Image is subjected to gradient transformation, obtains corresponding gradient map；
Based on watershed algorithm, the shape contour of each cell is determined according to the gradient map.
4. according to the method described in claim 2, it is characterized in that, being based on deep learning neural network model, according to described
Before frame constituency area image generates corresponding probability matrix, the method also includes:
Frame constituency area image is converted into corresponding grayscale image；
Histogram equalization processing is carried out to the grayscale image, obtains the image of contrast enhancing；
Noise reduction process is carried out to the image that the contrast enhances using gaussian filtering, the image after obtaining noise reduction；
The image after the noise reduction is standardized by such as minor function, obtains the deep learning neural network model
Input picture:
Wherein, x*Grey scale pixel value after indicating standardization, x indicate that the grey scale pixel value before standardization, μ indicate ash
The average gray of figure is spent, σ indicates the gray standard deviation of grayscale image.
5. generating the frame favored area the method according to claim 1, wherein instructing selecting according to the frame
After the shape contour of interior each cell, comprising:
The digital pathological section image is shown on interface centered on the cell in the frame favored area.
6. according to the method described in claim 5, it is characterized in that, by the digital pathological section image with the frame constituency
After being shown on interface centered on cell in domain, comprising:
Receive scaling instruction, wherein the scaling instruction is for zooming in or out the digital pathological section image shown on interface；
It is instructed according to the scaling, adjusts the digital pathological section image shown on interface.
7. a kind of cell labeling system characterized by comprising
Headend equipment, for showing digital pathological section image, and sink block choosing instruction, wherein the digital pathological section figure
As comprising multiple cells, the frame choosing instruction is described for determining a frame favored area on the digital pathological section image
Include one or more cells to be marked on the digital pathological section image in frame favored area
Background server is communicated with the headend equipment, is instructed for being selected according to the frame, is generated each in the frame favored area
The shape contour of cell；
Wherein, the headend equipment is also used to receive the markup information to the cell addition in any one shape contour.
8. a kind of cell labeling system characterized by comprising
Display screen, for showing digital pathological section image, wherein the digital pathological section image includes multiple cells；
Input equipment is used for sink block choosing instruction, wherein the frame choosing instruction is for true on the digital pathological section image
A fixed frame favored area, the frame favored area are interior thin comprising one or more to be marked on the digital pathological section image
Born of the same parents；
Processor is connect with the display screen, is instructed for being selected according to the frame, is generated each cell in the frame favored area
Shape contour；
Wherein, the input equipment is also used to receive the markup information to the cell addition in any one shape contour.
9. a kind of cell annotation equipment characterized by comprising
Display module, for showing digital pathological section image, wherein the digital pathological section image includes multiple cells；
First receiving module, for sink block choosing instruction, wherein the frame choosing instruction is in the digital pathological section image
One frame favored area of upper determination, the frame favored area is interior to include one or more to be marked on the digital pathological section image
Cell；
Generation module instructs for being selected according to the frame, generates the shape contour of each cell in the frame favored area；
Second receiving module, for receiving the markup information to the cell addition in any one shape contour.
10. a kind of cell mask method characterized by comprising
Show the first interface, wherein first interface is for showing digital pathological section image, and reception in the number
Determine that the frame of a frame favored area selects instruction on pathological section image, wherein the digital pathological section image includes multiple thin
Born of the same parents, the frame favored area is interior to include one or more cells to be marked on the digital pathological section image；
Show second contact surface, wherein the second contact surface is used to select according to the frame and instruct, and generates each in the frame favored area
The shape contour of cell, and receive the markup information to the cell addition in any one shape contour.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811318835.6A CN109544561A (en) | 2018-11-07 | 2018-11-07 | Cell mask method, system and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811318835.6A CN109544561A (en) | 2018-11-07 | 2018-11-07 | Cell mask method, system and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109544561A true CN109544561A (en) | 2019-03-29 |
Family
ID=65846220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811318835.6A Pending CN109544561A (en) | 2018-11-07 | 2018-11-07 | Cell mask method, system and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109544561A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110349655A (en) * | 2019-07-15 | 2019-10-18 | 上海商汤智能科技有限公司 | A kind of image adjusting method and device, electronic equipment and storage medium |
| CN111430011A (en) * | 2020-03-31 | 2020-07-17 | 杭州依图医疗技术有限公司 | Cell stain image display method, cell stain image processing method, and storage medium |
| CN111815607A (en) * | 2020-07-10 | 2020-10-23 | 济南大学 | Hematopoietic system-oriented bone marrow cell data set construction method and system |
| CN111986150A (en) * | 2020-07-17 | 2020-11-24 | 万达信息股份有限公司 | Interactive marking refinement method for digital pathological image |
| WO2020233254A1 (en) * | 2019-07-12 | 2020-11-26 | 之江实验室 | Medical data analysis system integrating structured image data |
| CN114663383A (en) * | 2022-03-18 | 2022-06-24 | 清华大学 | Blood cell segmentation and identification method and device, electronic equipment and storage medium |
| US11645753B2 (en) * | 2019-11-27 | 2023-05-09 | Case Western Reserve University | Deep learning-based multi-site, multi-primitive segmentation for nephropathology using renal biopsy whole slide images |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629777A (en) * | 2018-04-19 | 2018-10-09 | 麦克奥迪（厦门）医疗诊断系统有限公司 | A kind of number pathology full slice image lesion region automatic division method |
-
2018
- 2018-11-07 CN CN201811318835.6A patent/CN109544561A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629777A (en) * | 2018-04-19 | 2018-10-09 | 麦克奥迪（厦门）医疗诊断系统有限公司 | A kind of number pathology full slice image lesion region automatic division method |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020233254A1 (en) * | 2019-07-12 | 2020-11-26 | 之江实验室 | Medical data analysis system integrating structured image data |
| CN110349655A (en) * | 2019-07-15 | 2019-10-18 | 上海商汤智能科技有限公司 | A kind of image adjusting method and device, electronic equipment and storage medium |
| US11645753B2 (en) * | 2019-11-27 | 2023-05-09 | Case Western Reserve University | Deep learning-based multi-site, multi-primitive segmentation for nephropathology using renal biopsy whole slide images |
| CN111430011A (en) * | 2020-03-31 | 2020-07-17 | 杭州依图医疗技术有限公司 | Cell stain image display method, cell stain image processing method, and storage medium |
| CN111430011B (en) * | 2020-03-31 | 2023-08-29 | 杭州依图医疗技术有限公司 | Display method, processing method and storage medium for cell staining image |
| CN111815607A (en) * | 2020-07-10 | 2020-10-23 | 济南大学 | Hematopoietic system-oriented bone marrow cell data set construction method and system |
| CN111986150A (en) * | 2020-07-17 | 2020-11-24 | 万达信息股份有限公司 | Interactive marking refinement method for digital pathological image |
| CN111986150B (en) * | 2020-07-17 | 2024-02-09 | 万达信息股份有限公司 | The method comprises the following steps of: digital number pathological image Interactive annotation refining method |
| CN114663383A (en) * | 2022-03-18 | 2022-06-24 | 清华大学 | Blood cell segmentation and identification method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109544561A (en) | Cell mask method, system and device | |
| CN112348783B (en) | Image-based person identification method and device and computer-readable storage medium | |
| CN112232349B (en) | Model training method, image segmentation method and device | |
| CN111709420B (en) | Text detection method, electronic device and computer readable medium | |
| US9483835B2 (en) | Depth value restoration method and system | |
| CN106874909B (en) | A kind of recognition methods of image character and its device | |
| CN112966691B (en) | Multi-scale text detection method and device based on semantic segmentation and electronic equipment | |
| CN109522908A (en) | Image significance detection method based on area label fusion | |
| CN106651774B (en) | License plate super-resolution model reconstruction method and device | |
| CN108830780A (en) | Image processing method and device, electronic equipment, storage medium | |
| CN116258719B (en) | Flotation foam image segmentation method and device based on multi-mode data fusion | |
| CN111507226B (en) | Road image recognition model modeling method, image recognition method and electronic equipment | |
| CN112418216A (en) | Method for detecting characters in complex natural scene image | |
| CN112906794A (en) | Target detection method, device, storage medium and terminal | |
| CN109670516B (en) | Image feature extraction method, device, equipment and readable storage medium | |
| Zhao et al. | Automatic blur region segmentation approach using image matting | |
| CN103632153A (en) | Region-based image saliency map extracting method | |
| CN103578107A (en) | Method for interactive image segmentation | |
| CN106446844B (en) | Posture estimation method and device and computer system | |
| CN116612280A (en) | Vehicle segmentation method, device, computer equipment and computer readable storage medium | |
| CN111429512A (en) | Image processing method and device, storage medium and processor | |
| CN113870196A (en) | Image processing method, device, equipment and medium based on anchor point cutting graph | |
| Meena et al. | Image splicing forgery detection using noise level estimation | |
| CN111612802B (en) | Re-optimization training method based on existing image semantic segmentation model and application | |
| CN113421210A (en) | Surface point cloud reconstruction method based on binocular stereo vision |
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: 20190329 |
|
| RJ01 | Rejection of invention patent application after publication |