CN106558057B - A kind of medical image cutting method - Google Patents
A kind of medical image cutting method Download PDFInfo
- Publication number
- CN106558057B CN106558057B CN201611039544.4A CN201611039544A CN106558057B CN 106558057 B CN106558057 B CN 106558057B CN 201611039544 A CN201611039544 A CN 201611039544A CN 106558057 B CN106558057 B CN 106558057B
- Authority
- CN
- China
- Prior art keywords
- layer
- area
- repaired
- interest
- volume
- 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.)
- Active
Links
Classifications
-
- 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
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
Landscapes
- Image Processing (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The embodiment of the invention discloses a kind of medical image cutting method, this method is related to technical field of medical image processing.This method includes：Obtain initial segmentation result；Obtain the edit operation to initial segmentation result；Determine area to be repaired；The segmentation result again of the area to be repaired is obtained according to the optimal cut set that energy equation calculates；Segmentation result it will be updated to again in the initial segmentation result.Medical image cutting method provided in an embodiment of the present invention, realize and only regional area to be repaired in initial segmentation result is repaired, improve operation efficiency, while it also avoid the influence of the part to other correct segmentations.
Description
Technical field
The present embodiments relate to technical field of medical image processing, more particularly to a kind of medical image cutting method.
Background technology
Medical image segmentation refers to volume of interest from medical science three dimensional image reconstruction body data (Volume Data)
(Volume Of Interest, VOI) is split from image volumetric data, or will sense from medical science two-dimensional image data
The process that interest region (Region Of Interest, ROI) is split from two-dimensional medical images.Medical image can be
Computed tomography image (Computed Tomography, CT), MRI (Magnetic Resonance, MR) etc..
Medical science segmentation is premise and the basis for being analytical organic, is had great significance in organ surgery.Cure in recent years
It is to sketch the contours closed curve on the basis of segmented good CT images to learn one of edit methods common in image processing field,
Or mark foreground point and background dot and calculate part foreground area and background area, then pass through the segmentations such as walk random and calculate
Method is split again to the volume of interest in CT images, to be modified to initial segmentation result.
But the dividing method is substantially that CT is schemed by introducing interactive information (such as above-mentioned closed curve sketched the contours)
Volume of interest as in is split again.So not only efficiency is low, interested without what is repaired before also easily changing
The profile of volume.
The content of the invention
The embodiment of the present invention provides a kind of medical image cutting method, to realize only to be repaired in initial segmentation result
Regional area repaired, so as to improve operation efficiency, also avoid influenceing other parts correctly split.
In a first aspect, the embodiments of the invention provide a kind of medical image cutting method, this method includes：
Obtain the initial segmentation result of volume of interest in 3 d medical images；
Any figure layer comprising initial segmentation result is obtained to described current as current layer using in 3 d medical images
The edit operation of initial segmentation result in figure layer；
Area to be repaired is determined according to the edit operation, the area to be repaired is included in current layer comprising editor behaviour
Make the area for including neighbouring figure layer initial segmentation result corresponding with the bounding box in the bounding box in region, and neighbouring figure layer
Domain；
Result structure energy based on the initial segmentation result in the area to be repaired and after the edit operation
Equation, the segmentation result again of the area to be repaired is obtained according to the optimal cut set that energy equation calculates；
The segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to body interested
Long-pending segmentation.
Further, based on the initial segmentation result in the area to be repaired and the result after the edit operation
Energy equation is built, the segmentation result again that the optimal cut set calculated according to energy equation obtains the area to be repaired includes：
The area to be repaired in result, the neighbouring figure layer according to the current layer after edit operation
With the initial segmentation result, the area item R (A) in energy function is constructed, according to the current layer and the neighbouring figure layer
In volume of interest half-tone information, tectonic boundary item B (A), wherein energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；
Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
Further, the result according to the current layer after edit operation, described in the neighbouring figure layer
Area to be repaired and the initial segmentation result, the area item R (A) constructed in energy function include：
For the area to be repaired in the current layer, according to the result after edit operation, by the area to be repaired
The region of middle volume of interest carries out the assignment of given threshold, and is 0 to other area assignments in the area to be repaired；
For the area to be repaired in the neighbouring figure layer, by the surrounding border of the area to be repaired and described initial point
The assignment that the intersecting region of the volume of interest split in result carries out given threshold is cut, and to the area to be repaired
In other area assignments be 0.
Further, the half-tone information of the volume of interest in the current layer and the neighbouring figure layer, construction
Border item B (A) includes：
Enter row bound enhancing to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer；
According to enhanced boundary information tectonic boundary item B (A).
Further, the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer is carried out
Border enhancing includes：
Gray scale is carried out according to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer
The statistics of information, and determine gray threshold；
Binaryzation is carried out to the area to be repaired in the current layer and the neighbouring figure layer according to the gray threshold；
Gradient calculation is carried out to binaryzation result, and determines the border of the volume of interest.
Further, methods described also includes：
Area to be repaired in the current layer and the neighbouring figure layer is carried out before splitting again, to described current
It is adaptive down-sampled that figure layer and the neighbouring figure layer treat that interior area to be repaired is carried out；
Area to be repaired in the current layer and the neighbouring figure layer is carried out after splitting again, is recovered described and is worked as
Preceding figure layer and the pixel of the area to be repaired in the neighbouring figure layer.
Further, before the initial segmentation result of volume of interest in obtaining 3 d medical images, in addition to：
Obtain the figure layer sequence for including volume of interest；
The segmentation of volume of interest is carried out to the figure layer sequence, with produce volume of interest in 3 d medical images just
Beginning segmentation result.
Further, the 3 d medical images are MRI or computed tomography image；
The volume of interest includes：Liver, kidney, lung or stomach.
The embodiment of the present invention is by using the result after edit operation to local region to be repaired in initial segmentation result
Split again, and segmentation result will be updated to again in initial segmentation result.So as to realize only to initial segmentation result
In regional area to be repaired repaired, avoid influenceing other parts correctly split, while improve the fortune split again
Calculate efficiency.
Second aspect, the embodiments of the invention provide a kind of medical image cutting method, this method includes：
Any figure layer comprising volume of interest is used as current layer using in 3 d medical images；
The edit operation to volume of interest in the current layer is obtained, and using the result of the edit operation as just
Beginning segmentation result；
Determine include the bounding box in edit operation region in the current layer, and by the current layer and it is described ought
Region corresponding with the bounding box is as region to be split in the neighbouring figure layer of preceding figure layer；
Based on the structure energy equation in region to be split, and treated according to obtaining the optimal cut set that energy equation calculates
The segmentation result of cut zone, to complete the segmentation to volume of interest.
Further, based on the structure energy equation in region to be split, obtained according to the optimal cut set that energy equation calculates
To the segmentation result in the region to be split, included with the segmentation completed to volume of interest：
The area to be split in result and the neighbouring figure layer according to the current layer after edit operation
Domain, the area item R (A) in energy function is constructed, the volume of interest in the current layer and the neighbouring figure layer
Half-tone information, wherein tectonic boundary item B (A), energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；
Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
The embodiment of the present invention is split according to the result after user's edit operation to volume of interest, so as to realize right
Accurate segmentation in the case of initial segmentation result acquisition failure to volume of interest.
Brief description of the drawings
Fig. 1 is a kind of flow chart for medical image cutting method that the embodiment of the present invention one provides；
Fig. 2 a are being shown based on the initial segmentation result of liver in computed tomography image of providing of the embodiment of the present invention one
It is intended to；
Fig. 2 b are a kind of schematic diagrames for edit operation result based on liver that the embodiment of the present invention one provides；
Fig. 2 c are the schematic diagrames for another edit operation result based on liver that the embodiment of the present invention one provides；
Fig. 2 d are a kind of schematic diagrames for bounding box comprising edit operation region that the embodiment of the present invention one provides；
Fig. 3 is the flow chart for another medical image cutting method that the embodiment of the present invention one provides；
Fig. 4 is the signal of bounding box in a kind of each figure layer comprising edit operation region that the embodiment of the present invention one provides
Figure；
Fig. 5 is a kind of flow chart for medical image cutting method that the embodiment of the present invention two provides；
Fig. 6 is a kind of flow chart for medical image cutting method that the embodiment of the present invention three provides；
Fig. 7 is the tectonic model schematic diagram of the area item that the embodiment of the present invention three provides and border item；
Fig. 8 is a kind of flow chart for medical image cutting method that the embodiment of the present invention four provides；
Fig. 9 is the structural representation of a kind of electronic equipment provided in an embodiment of the present invention.
Embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention, rather than limitation of the invention.It also should be noted that in order to just
Part related to the present invention rather than entire infrastructure are illustrate only in description, accompanying drawing.
Embodiment one
Fig. 1 is a kind of flow chart for medical image cutting method that the embodiment of the present invention one provides.The present embodiment is applicable
In the situation that the initial segmentation result to volume of interest is repaired.This method can by a kind of medical image segmentation device Lai
Perform, the device can be configured in CT equipment, be realized by the mode of software and/or hardware.Referring to Fig. 1, the embodiment of the present invention
The medical image cutting method of offer includes：
S110, the initial segmentation result for obtaining volume of interest in 3 d medical images.
Wherein, the 3 d medical images can be MRI or computed tomography image；The body interested
Product can be target organ to be split, can specifically include：Liver, kidney, lung or stomach etc..Volume of interest can also be treated
The non-organ target of segmentation, such as tumour, tubercle, polyp and other focuses.Initial segmentation result can be based on it is interested
The half-tone information of volume is split the to obtain or movable contour model based on volume of interest is split to obtain
, can also be and split to obtain based on dividing method known to other skilled in the art, the present embodiment to this not
Carry out any restrictions.Volume of interest can be split in 3 d medical images according to initial segmentation, in volume of interest
Initial profile corresponding to volume of interest, i.e. initial segmentation result are obtained in each figure layer at place.Fig. 2 a are the embodiment of the present invention
One schematic diagram based on the initial segmentation result of liver in computed tomography image provided.Liver is being counted by initial segmentation
Split in calculation machine faultage image, referring to Fig. 2 a, the liver area 10 split in the figure layer where liver.
Due to partitioning algorithm limitation and patient's diversity, initial segmentation result is sometimes not fully accurate, this when
Manual editing can be carried out by user to initial segmentation result by, which waiting, operates.It can be generally divided into too for initial segmentation inaccuracy
Cut and be included in initial segmentation knot with two kinds of situations of less divided, the region that over-segmentation refers to for a part to be not belonging to volume of interest
Fruit, and less divided refers to a part of original region for belonging to volume of interest not being included in initial segmentation result.For example, ginseng
See Fig. 2 a, the less divided region 20 of corresponding liver is the subregion that liver area 10 is not divided into liver.
S120, any figure layer comprising initial segmentation result as current layer, is obtained to institute using in 3 d medical images
State the edit operation of initial segmentation result in current layer.
Wherein, the edit operation can be realized by dragging the profile of initial segmentation result in current layer to initial
The editor of segmentation result.Current layer can be any figure layer that 3 d medical images include initial segmentation result.If current figure
Initial segmentation result in layer belongs to less divided, then initial profile is outwards dragged to the region of less divided, if in current layer
Initial segmentation result is over-segmentation, then to the inside dragging initial profile of overdivided region, so that the initial segmentation knot by editor
Actual profile of the profile of fruit close to volume of interest.Optionally, the edit operation can also be by free pen to current
The part of less divided or over-segmentation carries out the drafting of curve in initial segmentation result in figure layer, realizes to initial segmentation result
Editor.Referring to Fig. 2 b and Fig. 2 c, continue so that above-mentioned initial segmentation result is to the less divided of liver as an example, user can pass through to
The initial profile of outer dragging liver, so that the less divided region of liver is included in initial profile；User can also pass through freedom
The less divided region curve 30 of corresponding liver is marked by pen.
S130, area to be repaired is determined according to the edit operation, the area to be repaired includes including in current layer
It is corresponding with the bounding box in the bounding box in edit operation region, and neighbouring figure layer to include neighbouring figure layer initial segmentation result
Region.
Wherein, bounding box is the rectangle frame for including edit operation region, for approx replacing obtaining by edit operation
The part for including local and initial segmentation result.As shown in Figure 2 d, continue to owe to divide using above-mentioned initial segmentation result as to liver
Example is segmented into, the edit operation result according to the user of acquisition to the less divided region 20 of liver, is extracted with rectangle frame comprising volume
The regional area of operating area is collected, using the area to be repaired as the initial segmentation result, wherein, the initial segmentation result
Area to be repaired can also include partial segmentation come out liver area 10.Neighbouring figure layer is current in 3 d medical images
Near figure layer, and need the figure layer repaired to initial segmentation result.In an embodiment of the present invention, user is only needed to working as
The initial segmentation result of preceding figure layer carries out edit operation.After the completion of to be edited, automatically according to edit operation result to neighbouring figure layer
Initial segmentation result repaired, to realize the automatic reparation of neighbouring figure layer.The determination of neighbouring figure layer scope is in aft section
It will be described in more detail.
Specifically, can according in current layer with the ratio for the area that volume of interest is corresponded in the neighbouring figure layer,
To determine the scope of the neighbouring figure layer.Wherein, volume of interest is initial segmentation result；Correspondingly the area of volume of interest is
The local area of volume of interest in bounding box.
Typically, if the initial segmentation result belongs to over-segmentation, by the area ratio near the current layer
Figure layer of the value more than or equal to 0.75 is as the neighbouring figure layer；If the initial segmentation result belongs to less divided, described will work as
Figure layer of the area ratio more than or equal to 0.65 near preceding figure layer is as the neighbouring figure layer.
Wherein, initial segmentation result belong to over-segmentation or the judgement of less divided can be by edit operation be initially to divide
Cut in result within the profile of volume of interest or obtained beyond profile；Can also input to obtain by user.If for example, compile
Operation is collected within the profile of initial segmentation result, then is judged as that initial segmentation result belongs to over-segmentation；If edit operation is first
Beyond the profile of beginning segmentation result, then it is judged as that initial segmentation result belongs to less divided.When can also be user's edit operation,
Operation interface sets a selection key, manually selects increase region (corresponding less divided) for user or deletion region is (corresponding
Over-segmentation), the drafting of free pen is carried out afterwards.
S140, the result structure based on the initial segmentation result in the area to be repaired and after the edit operation
Energy equation, the segmentation result again of the area to be repaired is obtained according to the optimal cut set that energy equation calculates.
S150, the segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to sense
The segmentation of volume of interest.
Fig. 3 is the flow chart for another medical image cutting method that the embodiment of the present invention one provides.Referring to Fig. 3, in reality
In the application of border, interact editor in current layer first, i.e., the above-mentioned edit operation based on initial segmentation result, and according to friendship
Mutual edited result is adaptive to should determine that area to be repaired, and wherein area to be repaired is the regional area for including above-mentioned edit operation；For
The amount of calculation split again to area to be repaired is reduced, area to be repaired is carried out adaptive down-sampled；Current layer
Area to be repaired is the area to be repaired of corresponding neighbouring figure layer, referring to Fig. 4, continued with above-mentioned in the same position of neighbouring figure layer
Initial segmentation result be to the less divided of liver exemplified by, by position of the area to be repaired in current layer, it is determined that corresponding adjacent
The area to be repaired of nearly figure layer；Then, according to the local and initial segmentation result and above-mentioned volume included in each figure layer area to be repaired
Figure of the result structure based on energy equation for collecting operation cuts model, and completes the segmentation again to area to be repaired；Finally, to again
The result of secondary segmentation carries out smooth and denoising using image processing techniques, and by the smooth and denoising of process again
The result of segmentation, which is updated to, initially divides in segmentation result, to complete the reparation to initial segmentation result.
The technical scheme of the embodiment of the present invention, by using the result after edit operation to locally being treated in initial segmentation result
The region of reparation is split again, and segmentation result will be updated to again in initial segmentation result.So as to realize only to first
Regional area to be repaired is repaired in beginning segmentation result, avoids influenceing other parts correctly split, while improve again
The operation efficiency of secondary segmentation.
Further, before the initial segmentation result of volume of interest in obtaining 3 d medical images, can also include：
Obtain the figure layer sequence for including volume of interest；
The segmentation of volume of interest is carried out to the figure layer sequence, with produce volume of interest in 3 d medical images just
Beginning segmentation result.
Embodiment two
Fig. 5 is a kind of flow chart for medical image cutting method that the embodiment of the present invention two provides.The present embodiment is upper
State a kind of alternative proposed on the basis of embodiment.The medical image cutting method bag provided referring to Fig. 5, the present embodiment
Include：
S210, the initial segmentation result for obtaining volume of interest in 3 d medical images.
S220, any figure layer comprising initial segmentation result as current layer, is obtained to institute using in 3 d medical images
State the edit operation of initial segmentation result in current layer.
S230, area to be repaired is determined according to the edit operation, the area to be repaired includes including in current layer
It is corresponding with the bounding box in the bounding box in edit operation region, and neighbouring figure layer to include neighbouring figure layer initial segmentation result
Region.
It is described to be repaired in S240, the result according to the current layer after edit operation, the neighbouring figure layer
Multiple region and the initial segmentation result, the area item R (A) in energy function is constructed, according to the current layer and the neighbour
The half-tone information of volume of interest in nearly figure layer, tectonic boundary item B (A), wherein energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result.
Wherein, pixel belongs to the possibility of volume of interest region and background area in area item R (A) expressions area to be repaired
Property size, border item B (A) is punishment when neighbor pixel is divided into volume of interest region and background area, with its gradient category
Property represent.
S250, according to the method for max-flow min-cut minimize energy function, obtain optimal cut set.
S260, the segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to sense
The segmentation of volume of interest.
Technical scheme provided in an embodiment of the present invention, energy function is constructed by area item R (A) and border item B (A), and
Energy function is minimized to obtain optimal cut set according to the method for max-flow min-cut, standard is carried out to area to be repaired so as to realize
Really split again.
It is interested in the current layer and the neighbouring figure layer to strengthen the value of boundary in border item B (A)
The half-tone information of volume, tectonic boundary item B (A) can include：
Enter row bound enhancing to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer；
According to enhanced boundary information tectonic boundary item B (A).
Wherein, in order to highlight the gradient information of figure layer to complete the division to unascertainable region, to the current figure
The volume of interest in area to be repaired in layer and the neighbouring figure layer, which enters row bound enhancing, to be included：
Gray scale is carried out according to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer
The statistics of information, and determine gray threshold；
Binaryzation is carried out to the area to be repaired in the current layer and the neighbouring figure layer according to the gray threshold；
Gradient calculation is carried out to binaryzation result, and determines the border of the volume of interest.
Embodiment three
Fig. 6 is a kind of flow chart for medical image cutting method that the embodiment of the present invention three provides.The present embodiment is upper
State a kind of alternative proposed on the basis of embodiment.The medical image cutting method bag provided referring to Fig. 6, the present embodiment
Include：
S310, the initial segmentation result for obtaining volume of interest in 3 d medical images.
S320, any figure layer comprising initial segmentation result as current layer, is obtained to institute using in 3 d medical images
State the edit operation of initial segmentation result in current layer.
S330, area to be repaired is determined according to the edit operation, the area to be repaired includes including in current layer
It is corresponding with the bounding box in the bounding box in edit operation region, and neighbouring figure layer to include neighbouring figure layer initial segmentation result
Region.
S340, for the area to be repaired in the current layer, according to the result after edit operation, for described neighbouring
Area to be repaired in figure layer, by what is split in the surrounding border of the area to be repaired and the initial segmentation result
The intersecting region of volume of interest is considered area-of-interest, by the surrounding border of the area to be repaired and the initial segmentation
As a result the region that the volume of interest split in does not intersect is considered background area, and remaining is uncertain region.
Specifically, the area item R (A) of neighbouring figure layer is constructed according to following table, wherein, represent region of interest with Rs (A)
Domain；Use RT(A) background area is represented, the assignment rule of pixel P area item can be described as follows：If P belongs to interested
Region, then Rs (AP) carry out the assignment of higher value, such as assignment 100, and to RT(AP) carry out the assignment of smaller value, such as 0；If P
Belong to background area, then Rs (AP) carry out the assignment of smaller value, such as assignment 0, and to RT(AP) carry out higher value assignment, example
Such as 100；If P is not known, Rs (AP) and RT (AP) it is entered as 0.It is understood that A=(A1,A2,…,Ap,..An) table
Show a kind of segmentation result, ApFor the affiliated label of pixel p in the figure layer sequence.
Table 1
| Region | P belongs to area-of-interest | P belongs to background area | P is not known |
| Rs(AP) | Higher value (100) | Smaller value (0) | 0 |
| RT(AP) | Smaller value (0) | Higher value (100) | 0 |
S350, basis are carried out to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer
The statistics of half-tone information, and determine gray threshold.
Specifically, the determination of gray threshold can determine according to the overall gray scale of volume of interest.
S360, according to the gray threshold in the current layer and the neighbouring figure layer area to be repaired carry out two
Value.
S370, gradient calculation is carried out to binaryzation result, and determine the border of the volume of interest, to construct current figure
The border item B (A) of layer and neighbouring figure layer.
Further, also include before gradient calculation is carried out to binaryzation result：Smooth mould is carried out to binaryzation result
Paste, so that the border of the volume of interest links up.
S380, energy function constructed according to above-mentioned zone item R (A) and border item B (A), and according to max-flow min-cut
Method solves energy function, to obtain optimal cut set,
Wherein energy function E (A) is：
E (A)=λ R (A)+B (A).
S390, the segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to sense
The segmentation of volume of interest.
Fig. 7 is the tectonic model schematic diagram of the area item that the embodiment of the present invention three provides and border item.Referring to Fig. 7, wherein,
P, q, r, 0, w and v are the pixel in figure layer, and T and S represent background area label and volume of interest area label respectively, often
Individual pixel and S and T connection are t-links, and for the connection between pixel n_links, there is a weight table on each side
Show, side right value is bigger in t-links, and line is thicker, represents that the pixel belongs to T or S probability is bigger；In n_links
Side right value is bigger, same thicker of line, represents that two pixels of side connection are more similar.It will determine as in the present embodiment interested
The t-links of the region of volume or the pixel of background area carries out larger assignment, and to the picture of uncertain area assignment
Vegetarian refreshments is entered as 0, i.e. t-links disconnects, so as to reduce shadows of the t-links of uncertain region to the accuracy rate of segmentation result
Ring.
Technical scheme provided in an embodiment of the present invention, by being directly obtained according to the result after edit operation in current layer
The region and background area of volume of interest, and according to the surrounding border of area to be repaired with being partitioned into initial segmentation result
The position relationship of the volume of interest come obtains the region and background area of the volume of interest in neighbouring figure layer, so as to avoid
The region and background area of volume of interest are distinguished using half-tone information, and then solves region and the background area of volume of interest
Domain because gray value it is close, and be not easy division the problem of.
To reduce the amount of calculation split again to area to be repaired, methods described also includes：
Area to be repaired in the current layer and the neighbouring figure layer is carried out before splitting again, to described current
It is adaptive down-sampled that figure layer and the neighbouring figure layer treat that interior area to be repaired is carried out；
Area to be repaired in the current layer and the neighbouring figure layer is carried out after splitting again, is recovered described and is worked as
Preceding figure layer and the pixel of the area to be repaired in the neighbouring figure layer.
Example IV
Fig. 8 is a kind of flow chart for medical image cutting method that the embodiment of the present invention four provides.The present embodiment is applied to
Initial segmentation result obtains the situation of failure.Referring to Fig. 8, medical image cutting method provided in an embodiment of the present invention includes：
S410, any figure layer comprising volume of interest is used as current layer using in 3 d medical images.
The edit operation of S420, acquisition to volume of interest in the current layer, the result of the edit operation is made
For initial segmentation result.
S430, determine in the current layer include the bounding box in edit operation region, and by the current layer with
Region corresponding with the bounding box is as region to be split in the neighbouring figure layer of the current layer.
S440, based on the region to be split structure energy equation, institute is obtained according to the optimal cut set that energy equation calculates
The segmentation result in region to be split is stated, to complete the segmentation to volume of interest.
Further, based on the structure energy equation in region to be split, obtained according to the optimal cut set that energy equation calculates
To the segmentation result in the region to be split, included with the segmentation completed to volume of interest：
The area to be split in result and the neighbouring figure layer according to the current layer after edit operation
Domain, the area item R (A) in energy function is constructed, the volume of interest in the current layer and the neighbouring figure layer
Half-tone information, wherein tectonic boundary item B (A), energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；
Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
Typically, area that can be using the volume of interest region edited in the region to be split as volume of interest
Domain, using the non-volume of interest region edited in the region to be split as background area, build energy equation.According to energy
The optimal cut set that equation calculates obtains the segmentation result again in the region to be split.
The technical scheme of the embodiment of the present invention, by being divided according to the result after user's edit operation volume of interest
Cut, so as to realize the accurate segmentation in the case where obtaining failure to initial segmentation result to volume of interest.
Fig. 9 is the structural representation of a kind of electronic equipment provided in an embodiment of the present invention.The embodiment of the present invention additionally provides
A kind of electronic equipment, as shown in figure 9, the equipment includes：
One or more processors 710, in Fig. 9 by taking a processor 710 as an example；
Memory 720；
The electronic equipment can also include：Input unit 730 and output device 740.
Processor 710, memory 720, input unit 730 and output device 740 in the electronic equipment can pass through
Bus or other modes connect, in Fig. 9 exemplified by being connected by bus.
Memory 720 is used as a kind of non-transient computer readable storage medium storing program for executing, can available for storage software program, computer
Configuration processor and module, programmed instruction/module as corresponding to the date storage method in the embodiment of the present application.Processor 710
Software program, instruction and the module being stored in by operation in memory 720, so as to the various function application of execute server
And data processing, that is, realize the medical image cutting method of above method embodiment.
Memory 720 can include storing program area and storage data field, wherein, storing program area can store operation system
Application program required for system, at least one function；Storage data field can store uses created number according to electronic equipment
According to etc..In addition, memory 720 can include high-speed random access memory, non-transitory memory can also be included, such as extremely
Few a disk memory, flush memory device or other non-transitory solid-state memories.In certain embodiments, memory
720 is optional including that can pass through network connection to end relative to the remotely located memory of processor 710, these remote memories
End equipment.The example of above-mentioned network includes but is not limited to internet, intranet, LAN, mobile radio communication and combinations thereof.
Input unit 730 can be used for the numeral or character information for receiving input, and produces and set with the user of electronic equipment
Put and the input of key signals that function control is relevant.Output device 740 may include the display devices such as display screen.
Namely：Above-mentioned electronic equipment includes：
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are by one or more of computing devices so that one or more of processing
When device is realized, proceed as follows：
Obtain the initial segmentation result of volume of interest in 3 d medical images；
Any figure layer comprising initial segmentation result is obtained to described current as current layer using in 3 d medical images
The edit operation of initial segmentation result in figure layer；
Area to be repaired is determined according to the edit operation, the area to be repaired is included in current layer comprising editor behaviour
Make the area for including neighbouring figure layer initial segmentation result corresponding with the bounding box in the bounding box in region, and neighbouring figure layer
Domain；
Result structure energy based on the initial segmentation result in the area to be repaired and after the edit operation
Equation, the segmentation result again of the area to be repaired is obtained according to the optimal cut set that energy equation calculates；
The segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to body interested
Long-pending segmentation.
Further, the segmentation result again of the area to be repaired is obtained in the optimal cut set calculated according to energy equation
Before, in addition to：
The volume of interest according to the initial segmentation result in the current layer with the neighbouring figure layer
Area ratio, determine the scope of the neighbouring figure layer.
Further, the volume of interest according to the initial segmentation result in the current layer with the neighbour
Area ratio in nearly figure layer, determining the scope of the neighbouring figure layer includes：
If the initial segmentation result belongs to over-segmentation, the area ratio near the current layer is more than etc.
In 0.75 figure layer as the neighbouring figure layer；
If the initial segmentation result belongs to less divided, the area ratio near the current layer is more than etc.
In 0.65 figure layer as the neighbouring figure layer.
Further, based on the initial segmentation result in the area to be repaired and the result after the edit operation
Energy equation is built, the segmentation result again that the optimal cut set calculated according to energy equation obtains the area to be repaired includes：
The area to be repaired in result, the neighbouring figure layer according to the current layer after edit operation
With the initial segmentation result, the area item R (A) in energy function is constructed, according to the current layer and the neighbouring figure layer
In volume of interest half-tone information, tectonic boundary item B (A), wherein energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；
Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
Further, the result according to the current layer after edit operation, described in the neighbouring figure layer
Area to be repaired and the initial segmentation result, the area item R (A) constructed in energy function include：
For the area to be repaired in the current layer, according to the result after edit operation, by the area to be repaired
The region of middle volume of interest carries out the assignment of given threshold, and is 0 to other area assignments in the area to be repaired；
For the area to be repaired in the neighbouring figure layer, by the surrounding border of the area to be repaired and described initial point
The assignment that the intersecting region of the volume of interest split in result carries out given threshold is cut, and to the area to be repaired
In other area assignments be 0.
Further, the half-tone information of the volume of interest in the current layer and the neighbouring figure layer, construction
Border item B (A) includes：
Enter row bound enhancing to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer；
According to enhanced boundary information tectonic boundary item B (A).
Further, the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer is carried out
Border enhancing includes：
Gray scale is carried out according to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer
The statistics of information, and determine gray threshold；
Binaryzation is carried out to the area to be repaired in the current layer and the neighbouring figure layer according to the gray threshold；
Gradient calculation is carried out to binaryzation result, and determines the border of the volume of interest.
Further, methods described also includes：
Area to be repaired in the current layer and the neighbouring figure layer is carried out before splitting again, to described current
It is adaptive down-sampled that figure layer and the neighbouring figure layer treat that interior area to be repaired is carried out；
Area to be repaired in the current layer and the neighbouring figure layer is carried out after splitting again, is recovered described and is worked as
Preceding figure layer and the pixel of the area to be repaired in the neighbouring figure layer.
Further, before the initial segmentation result of volume of interest in obtaining 3 d medical images, in addition to：
Obtain the figure layer sequence for including volume of interest；
The segmentation of volume of interest is carried out to the figure layer sequence, with produce volume of interest in 3 d medical images just
Beginning segmentation result.
Further, the 3 d medical images are MRI or computed tomography image.
Further, the volume of interest includes：Liver, kidney, lung or stomach.
Or
Any figure layer comprising volume of interest is used as current layer using in 3 d medical images；
The edit operation to volume of interest in the current layer is obtained, and using the result of the edit operation as just
Beginning segmentation result；
Determine include the bounding box in edit operation region in the current layer, and by the current layer and it is described ought
Region corresponding with the bounding box is as region to be split in the neighbouring figure layer of preceding figure layer；
Based on the structure energy equation in region to be split, and treated according to obtaining the optimal cut set that energy equation calculates
The segmentation result of cut zone, to complete the segmentation to volume of interest.
Further, based on the structure energy equation in region to be split, obtained according to the optimal cut set that energy equation calculates
To the segmentation result in the region to be split, included with the segmentation completed to volume of interest：
The area to be split in result and the neighbouring figure layer according to the current layer after edit operation
Domain, the area item R (A) in energy function is constructed, the volume of interest in the current layer and the neighbouring figure layer
Half-tone information, wherein tectonic boundary item B (A), energy function E (A) are：
E (A)=λ R (A)+B (A)
Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；
Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
Pay attention to, above are only presently preferred embodiments of the present invention and institute's application technology principle.It will be appreciated by those skilled in the art that
The invention is not restricted to specific embodiment described here, can carry out for a person skilled in the art various obvious changes,
Readjust and substitute without departing from protection scope of the present invention.Therefore, although being carried out by above example to the present invention
It is described in further detail, but the present invention is not limited only to above example, without departing from the inventive concept, also
Other more equivalent embodiments can be included, and the scope of the present invention is determined by scope of the appended claims.
Claims (10)
- A kind of 1. medical image cutting method, it is characterised in that including：Obtain the initial segmentation result of volume of interest in 3 d medical images；Any figure layer comprising initial segmentation result is obtained to the current layer as current layer using in 3 d medical images The edit operation of interior initial segmentation result；Area to be repaired is determined according to the edit operation, the area to be repaired includes including edit operation area in current layer The region for including neighbouring figure layer initial segmentation result corresponding with the bounding box in the bounding box in domain, and neighbouring figure layer；Result structure energy equation based on the initial segmentation result in the area to be repaired and after the edit operation, The segmentation result again of the area to be repaired is obtained according to the optimal cut set that energy equation calculates；The segmentation result again of the area to be repaired is updated in the initial segmentation result, completed to volume of interest Segmentation.
- 2. according to the method for claim 1, it is characterised in that based on the initial segmentation knot in the area to be repaired Result structure energy equation after fruit and the edit operation, the optimal cut set calculated according to energy equation obtains described to be repaired The segmentation result again in region includes：Result according to the current layer after edit operation, the area to be repaired in the neighbouring figure layer and institute Initial segmentation result is stated, the area item R (A) in energy function is constructed, according in the current layer and the neighbouring figure layer The half-tone information of volume of interest, wherein tectonic boundary item B (A), energy function E (A) are：E (A)=λ R (A)+B (A)Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
- 3. according to the method for claim 2, it is characterised in that the knot according to the current layer after edit operation Fruit, the area to be repaired in the neighbouring figure layer and the initial segmentation result, construct the area item R in energy function (A) include：For the area to be repaired in the current layer, according to the result after edit operation, will feel in the area to be repaired The region of volume of interest carries out the assignment of given threshold, and is 0 to other area assignments in the area to be repaired；For the area to be repaired in the neighbouring figure layer, by the surrounding border of the area to be repaired and the initial segmentation knot The region that the volume of interest split in fruit intersects carries out the assignment of given threshold, and in the area to be repaired Other area assignments are 0.
- 4. according to the method for claim 2, it is characterised in that according to the sense in the current layer and the neighbouring figure layer The half-tone information of volume of interest, tectonic boundary item B (A) include：Enter row bound enhancing to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer；According to enhanced boundary information tectonic boundary item B (A).
- 5. according to the method for claim 4, it is characterised in that to be repaired in the current layer and the neighbouring figure layer Volume of interest in multiple region, which enters row bound enhancing, to be included：Half-tone information is carried out according to the volume of interest in the area to be repaired in the current layer and the neighbouring figure layer Statistics, and determine gray threshold；Binaryzation is carried out to the area to be repaired in the current layer and the neighbouring figure layer according to the gray threshold；Gradient calculation is carried out to binaryzation result, and determines the border of the volume of interest.
- 6. according to the method for claim 1, it is characterised in that also include：Area to be repaired in the current layer and the neighbouring figure layer is carried out before splitting again, to the current layer Carried out with the area to be repaired in the neighbouring figure layer adaptive down-sampled；Area to be repaired in the current layer and the neighbouring figure layer is carried out after splitting again, recovers the current figure The pixel of area to be repaired in layer and the neighbouring figure layer.
- 7. according to the method for claim 1, it is characterised in that volume of interest is initial in 3 d medical images are obtained Before segmentation result, in addition to：Obtain the figure layer sequence for including volume of interest；The segmentation of volume of interest is carried out to the figure layer sequence, to produce initial point of volume of interest in 3 d medical images Cut result.
- 8. according to the method for claim 1, it is characterised in that the 3 d medical images are MRI or calculating Machine faultage image；The volume of interest includes：Liver, kidney, lung or stomach.
- A kind of 9. medical image cutting method, it is characterised in that including：Any figure layer comprising volume of interest is used as current layer using in 3 d medical images；The edit operation to volume of interest in the current layer is obtained, using the result of the edit operation as initial segmentation As a result；The bounding box for including edit operation region is determined in the current layer, by the current layer and the current layer Neighbouring figure layer in region corresponding with the bounding box as region to be split；Based on the structure energy equation in region to be split, the area to be split is obtained according to the optimal cut set that energy equation calculates The segmentation result in domain, to complete the segmentation to volume of interest.
- 10. according to the method for claim 9, it is characterised in that energy equation is built based on the region to be split, according to The optimal cut set that energy equation calculates obtains the segmentation result in the region to be split, to complete the segmentation bag to volume of interest Include：The region to be split in result and the neighbouring figure layer according to the current layer after edit operation, structure The area item R (A) in energy function is made, the gray scale letter of the volume of interest in the current layer and the neighbouring figure layer Breath, tectonic boundary item B (A), wherein energy function E (A) are：E (A)=λ R (A)+B (A)Wherein, the factors of the λ between area item and border item, A represent a kind of segmentation result；Energy function is solved according to the method for max-flow min-cut, obtains optimal cut set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611039544.4A CN106558057B (en) | 2016-11-21 | 2016-11-21 | A kind of medical image cutting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611039544.4A CN106558057B (en) | 2016-11-21 | 2016-11-21 | A kind of medical image cutting method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106558057A CN106558057A (en) | 2017-04-05 |
| CN106558057B true CN106558057B (en) | 2018-03-20 |
Family
ID=58444418
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611039544.4A Active CN106558057B (en) | 2016-11-21 | 2016-11-21 | A kind of medical image cutting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106558057B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106997589A (en) * | 2017-04-12 | 2017-08-01 | 上海联影医疗科技有限公司 | image processing method, device and equipment |
| CN109584249B (en) * | 2018-11-21 | 2022-11-25 | 大连理工大学 | Three-dimensional volume data segmentation method based on closed form solution |
| CN109801271B (en) * | 2019-01-04 | 2021-11-23 | 上海联影医疗科技股份有限公司 | Method and device for locating calcified cluster, computer equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101639935A (en) * | 2009-09-07 | 2010-02-03 | 南京理工大学 | Digital human serial section image segmentation method based on geometric active contour target tracking |
| CN104899849A (en) * | 2014-01-21 | 2015-09-09 | 武汉联影医疗科技有限公司 | Multi-target interactive image segmentation method and device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8498481B2 (en) * | 2010-05-07 | 2013-07-30 | Microsoft Corporation | Image segmentation using star-convexity constraints |
| US9129192B2 (en) * | 2013-12-16 | 2015-09-08 | Adobe Systems Incorporated | Semantic object proposal generation and validation |
-
2016
- 2016-11-21 CN CN201611039544.4A patent/CN106558057B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101639935A (en) * | 2009-09-07 | 2010-02-03 | 南京理工大学 | Digital human serial section image segmentation method based on geometric active contour target tracking |
| CN104899849A (en) * | 2014-01-21 | 2015-09-09 | 武汉联影医疗科技有限公司 | Multi-target interactive image segmentation method and device |
Non-Patent Citations (3)
| Title |
|---|
| A survey of current methods in medical image segmentation;Dzung L. Pham 等;《Annual Review of Biomedical Engineering》;20001231;第315-337页 * |
| 医学图像分割方法综述;黄文博 等;《长春师范学院学报（自然科学版）》;20130430;第32卷(第2期);第22-25页 * |
| 基于模糊C均值聚类的医学图像分割研究;张翡 等;《计算机工程与应用》;20141231;第50卷(第4期);第144-151页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106558057A (en) | 2017-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liew et al. | Regional interactive image segmentation networks | |
| JP6884853B2 (en) | Image segmentation using neural network method | |
| Wang et al. | CT male pelvic organ segmentation using fully convolutional networks with boundary sensitive representation | |
| JP6932182B2 (en) | Systems and methods for image segmentation with convolutional neural networks | |
| Gu et al. | Automated delineation of lung tumors from CT images using a single click ensemble segmentation approach | |
| CN108010021A (en) | A kind of magic magiscan and method | |
| CN110335277A (en) | Image processing method, device, computer readable storage medium and computer equipment | |
| US8131076B2 (en) | Editing of pre-segmented images using seeds derived from contours | |
| CN108022242A (en) | Use the automatic segmentation of the priori of deep learning | |
| CN108038862B (en) | Interactive medical image intelligent segmentation modeling method | |
| JP2019531783A5 (en) | ||
| JP2020516427A (en) | RECIST assessment of tumor progression | |
| EP2620909B1 (en) | Method, system and computer readable medium for automatic segmentation of a medical image | |
| CN104573309A (en) | Apparatus and method for computer-aided diagnosis | |
| CN105608728A (en) | Semantic medical image to 3D print of anatomic structure | |
| CN105167793A (en) | Image display apparatus, display control apparatus and display control method | |
| CN107622493A (en) | Method and data processing unit for the object in Medical Image Segmentation | |
| CN106558057B (en) | A kind of medical image cutting method | |
| CN110648309B (en) | Method and related equipment for generating anti-network synthesized erythrocyte image based on condition | |
| US9317926B2 (en) | Automatic spinal canal segmentation using cascaded random walks | |
| US8050470B2 (en) | Branch extension method for airway segmentation | |
| CN105389821B (en) | It is a kind of that the medical image cutting method being combined is cut based on cloud model and figure | |
| Tan et al. | An approach for pulmonary vascular extraction from chest CT images | |
| WO2020110774A1 (en) | Image processing device, image processing method, and program | |
| CN106462974A (en) | Optimization of parameters for segmenting an image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 201807 Shanghai City, north of the city of Jiading District Road No. 2258 Patentee after: Shanghai Lianying Medical Technology Co., Ltd Address before: 201807 Shanghai City, north of the city of Jiading District Road No. 2258 Patentee before: SHANGHAI UNITED IMAGING HEALTHCARE Co.,Ltd. |