CN104915942B - A kind of detection method and device of skin of breast line - Google Patents
A kind of detection method and device of skin of breast line Download PDFInfo
- Publication number
- CN104915942B CN104915942B CN201410088614.XA CN201410088614A CN104915942B CN 104915942 B CN104915942 B CN 104915942B CN 201410088614 A CN201410088614 A CN 201410088614A CN 104915942 B CN104915942 B CN 104915942B
- Authority
- CN
- China
- Prior art keywords
- image
- gradient
- skin
- line
- breast
- 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
- 210000000481 Breast Anatomy 0.000 title claims abstract description 214
- 210000003491 Skin Anatomy 0.000 title claims abstract description 179
- 238000001514 detection method Methods 0.000 title claims abstract description 49
- 210000001519 tissues Anatomy 0.000 claims abstract description 22
- 230000002708 enhancing Effects 0.000 claims abstract description 15
- 238000000844 transformation Methods 0.000 claims description 58
- 230000001131 transforming Effects 0.000 claims description 58
- 238000000034 method Methods 0.000 claims description 25
- 238000009825 accumulation Methods 0.000 claims description 18
- 238000006243 chemical reaction Methods 0.000 claims description 16
- 230000000875 corresponding Effects 0.000 claims description 14
- 210000000981 Epithelium Anatomy 0.000 claims description 7
- 239000004744 fabric Substances 0.000 claims description 5
- 238000009499 grossing Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 238000003325 tomography Methods 0.000 description 6
- 210000004907 Glands Anatomy 0.000 description 5
- 210000002264 Mammary Glands, Animal Anatomy 0.000 description 3
- 210000004293 Mammary Glands, Human Anatomy 0.000 description 3
- 210000002445 Nipples Anatomy 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000006011 modification reaction Methods 0.000 description 3
- 238000003708 edge detection Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000000052 comparative effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002601 radiography Methods 0.000 description 1
Abstract
The present invention provides a kind of detection method and device of skin of breast line, and described method includes following steps：(1) the initial three-dimensional sequence image for inputting breast tissue determines the area-of-interest of left and right breast according to breast tissue characteristic point in the picture and feature locations；(2) according to the statistical distribution of skin line feature, enhance the skin of breast line on image in left and right breast area of interest；(3) respectively according to the intermediate tomographic image of left and right breast area of interest, by dynamic programming method, the skin line on tomographic image among left and right breast area of interest is determined；(4) skin line on the intermediate tomographic image based on left and right breast area of interest determines the skin line on the tomographic image in left and right breast area of interest other than the middle layer successively respectively.The technical program can accurately and efficiently realize the full-automatic detection of skin of breast line.
Description
Technical field
The present invention relates to image processing field more particularly to a kind of detection methods and device of skin of breast line.
Background technology
In clinic, breast dynamic contrast enhancing magnetic resonance image generally comprises several Magnetic Resonance Image scan sequences
Row, to meet the different demands of radiologist.Radiography is squeezed into for the ease of breast inner tissue in comparative observation magnetic resonance image
Enhancing situation before and after agent, the pressure fat image sequence of Magnetic Resonance Image scan sequence generally use T1 weightings.T1 adds
The more other types of image sequences of pressure fat image sequence of power have higher noise level, the interference of stronger artifact and lower figure
As grey-scale contrast, and gray scale and the gradient distribution of skin of breast line have in different scanning tomography and same scan tomography
Diversity.In addition, closing on for mammary gland fibr tissue and skin line position also increases skin of breast line and accurately examines in image
The difficulty of survey.
For the detection of skin of breast line in three-dimensional magnetic resonance image, method for semi-automatically detecting that is manual and needing user's auxiliary
More complicated, efficiency is low and there are between larger observer and the difference of observer itself.
In full-automatic context of detection, main research at present concentrates on T1 weightings and does not press in the magnetic resonance image sequence of fat, and
The full-automatic detection method of skin of breast line is also fewer in the magnetic resonance image sequence that T1 weights pressure fat.Due to T1 weightings
Pressure fat image sequence does not press fat image sequence to have higher noise level, the interference of stronger artifact and lower gradation of image
Contrast, therefore, existing method is not suitable for pressing fat image sequence on not pressing fat image sequence.Due to different scanning tomography
And the skin of breast that gray scale and the gradient distribution of skin of breast line have diversity, certain sufferers relatively thin in same scan tomography
There is line body of gland fibr tissue in lower contrast and image to be positioned proximate to skin line after the processing of over-pressed fat, cause
Detection method based on threshold value or the detection method based on gradient are not enough to accurately extract skin of breast line.Based on model
Detection method need a large amount of training sample to go training pattern to obtain accurate testing result, and these precisely training
The acquisition of sample is a more arduous and complicated task.
Invention content
Problems solved by the invention is to provide a kind of detection method and device of skin of breast line, not only in conjunction with skin line spy
Skin line in the statistical distribution enhancing original image of sign eliminates body of gland by the saturation processing of gradient image different stage
The interference that fibr tissue and the outer artifact of skin line detect skin line, it is special herein in connection with the continuity of skin line in adjacent scanning tomography
Point accurately and efficiently realizes the full-automatic detection of skin line.
To solve the above-mentioned problems, the present invention provides a kind of detection method of skin of breast line, include the following steps：
(1) the initial three-dimensional sequence image for inputting breast tissue, according to the characteristic point of the breast tissue in the picture and
Feature locations determine the area-of-interest of left and right breast；
(2) according to the statistical distribution of skin line feature, enhance the breast on image in the left and right breast area of interest
Skin line；
(3) it is determined respectively by dynamic programming method according to the intermediate tomographic image of the left and right breast area of interest
Skin line among the left and right breast area of interest on tomographic image；
(4) skin line on intermediate tomographic image based on the left and right breast area of interest is distinguished described in determining successively
The skin line on tomographic image in left and right breast area of interest other than the middle layer.
A kind of detection method of above-mentioned skin of breast line, it is preferable that the breast in the enhancing left and right breast area of interest
The process of room skin line is：
(1) data set of the statistical distribution for obtaining the skin line feature is determined；
(2) position and the feature for concentrating skin of breast line based on the data, obtain the statistical of the skin line feature
Cloth；
(3) respectively by the feature of each pixel in the left and right breast area of interest and the skin line feature
Statistical distribution in corresponding feature distribution percentage value be multiplied.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the statistical distribution of the skin line feature is united for gray scale
Score cloth.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that determine the left and right breast area of interest middle layer
The process of skin line is as follows on image：
(1) according to ray scanning starting point, ray scanning radius and flying spot number, based on polar coordinates conversion and ray
Tomographic image among the left and right breast area of interest is converted to two-dimensional transformations image by scan method respectively；Wherein, described to penetrate
Line scan start point is central point left and right breast area of interest described in the intermediate tomographic image of the left and right breast
Subpoint on lower boundary；Distance of the ray scanning starting point apart from the left and right papilla of breast determines that the ray is swept
Retouch radius；
(2) the object boundary line in the two-dimensional transformations image, the object boundary are obtained by the method for Dynamic Programming
Line be skin line among the left and right breast area of interest on tomographic image after conversion in two-dimensional transformations image
Position；
(3) judge whether the object boundary line has fluctuation, if with fluctuation, pass through second level gradient saturation
The fluctuation is eliminated in processing；
(4) it is based on the object boundary line, the left and right breast region of interest is obtained by the method for polar coordinates inverse conversion
Skin line among domain on tomographic image.
A kind of detection method of above-mentioned skin of breast line, it is preferable that the local energy equation in the dynamic programming method
Formula be expressed as：
C (i)=dis (i, i-1)/max (dis)+g1(i)/max (g1) (1)
Wherein, i indicates that stage, the stage are each row of the two-dimensional transformations image, the two-dimensional transformations image
Each to be classified as the ray obtained by ray scanning method sequence, the gray value of the pixel on each row is ray scanning
The gray value for the pixel that the ray that method sequence obtains passes through in tomographic image among the left and right breast area of interest；C
(i) local energy for being point P (j) on the i-th stage；Dis (i, i-1) indicates that point Q (k) exists on the point P (j) to the (i-1)-th stage
The distance projected on i-th stage；Max (dis) indicates the maximum value in all distances in the two-dimensional transformations image；g1(i) table
Show that point P (j) is in the gradient image g after the processing of first level saturation on the i-th stage1In image gradient；mxa(g1) indicate
All the points are by the maximum value in first level saturation treated gradient image in two-dimensional transformations image.
A kind of detection method of above-mentioned skin of breast line, it is preferable that the gradient of the first level gradient saturation processing
Image g1Acquisition process it is as follows：
(1) Gaussian smoothing is carried out to the two-dimensional transformations image, obtains smoothed image；
(2) it is based on the smoothed image, obtains Initial Gradient image, and the gradient for calculating the Initial Gradient image is average
Value；
(3) it is based on the Initial Gradient image and two-dimensional transformations image, determines interference pixel, and by the Initial Gradient
The gradient average value of image assigns the interference pixel, and the Grad of rest of pixels point is constant, as gradient image g；
(4) the gradient distribution for counting the gradient image g, obtains gradient profile accumulation histogram；
(5) it by first threshold and the gradient profile accumulation histogram, obtains at the first level gradient saturation
The gradient image g of reason1。
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the process for obtaining Initial Gradient image is：It is based on
Partial derivative of the two-dimensional Gaussian kernel on the directions y of the two-dimensional transformations image, obtains the gradient of each pixel of the smoothed image
Value；By the Grad linear change of each pixel of the smoothed image to scheduled range, the as described Initial Gradient image.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the determination process of the interference pixel is as follows：
(1) the pixel M (m) that Grad in the Initial Gradient image is less than 512 is chosen；
(2) in the two-dimensional transformations image, the pixel M (m) is calculated in M pixel vertically upward
Average gray, the average gray are more than preset threshold value, then the pixel M (m) is interference pixel.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that straight by the first threshold and gradient profile accumulation
Fang Tu obtains the gradient image g of the first level gradient saturation processing1Process be：
(1) according to the gradient profile accumulation histogram, the first gradient value corresponding to the first threshold is obtained；
(2) if Grad is less than the first gradient value in the gradient image g, it is assigned a value of 0；If the gradient
Grad is more than the gradient average value of all Grad of gradient image g in image g, then is assigned a value of the gradient and be averaged
Value；
(3) by the first gradient value and the gradient average value, linear stretch is carried out to the gradient image g.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the first threshold is 1%.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that handled by second level saturation and eliminate the wave
Dynamic process is as follows：
(1) it according to the gradient profile accumulation histogram and second threshold, obtains by the second level saturation
The gradient image g of reason2；
(2) according to the gradient image g2, Dynamic Programming is carried out, optimal curve is obtained；
(3) with the optimal curve in the gradient image g2Position in each column is starting point, is determined successively in the ladder
Spend image g2First position with greatest gradient value in vertical downward direction in each column；By gradient image g1Correspondence image arranges
In the Grad of the downward all pixels point in the position be assigned a value of the greatest gradient value, that is, update the gradient image g1；
(4) according to the newer gradient image g1, Dynamic Programming is carried out, the object boundary line is obtained.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the second threshold is 10%.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that the greatest gradient value is 255.
The detection method of above-mentioned a kind of skin of breast line, it is preferable that determine the left and right breast region of interest successively respectively
The process of the skin line on tomographic image in domain other than the middle layer is as follows：
(1) in the layer described in addition to middle layer, the left and right breast area of interest middle layer both sides respectively press from
The order of middle layer from the near to the distant selects image layer as current layer successively, obtains the two-dimensional transformations image and ladder of the current layer
Spend image；
(2) testing result based on gaussian kernel function and the current layer preceding layer or latter tomographic image epithelium skin line is right
The current layer gradient image is improved；
(3) by the method for Dynamic Programming, object boundary line is obtained, the object boundary line is on the current tomographic image
Position of the skin line in the two-dimensional transformations image；
(4) judge whether the object boundary line of the current layer has fluctuation, if with fluctuation, it is full by second level
The fluctuation is eliminated with processing is changed；
(5) it is the skin on the current tomographic image by the method migration that the object boundary passes through polar coordinates inverse conversion
Line.
To solve the above-mentioned problems, the present invention also provides a kind of detection devices of skin of breast line, including：
Area-of-interest determination unit is suitable for inputting the initial three-dimensional sequence image of breast tissue, according to the breast
Characteristic point in the picture and feature locations are organized, determine the corresponding area-of-interest of left and right breast；
Enhancement unit is suitable for the statistical distribution according to skin line feature, enhances in the left and right breast area of interest
Skin of breast line on all tomographic images；
Middle layer skin line determination unit is suitable for respectively according to the middle layer figure of the left and right breast area of interest
Picture determines the skin line on tomographic image among the left and right breast area of interest by dynamic programming method；
Skin line determination unit, the skin being suitable on the intermediate tomographic image based on the left and right breast area of interest
Line determines the skin line on the tomographic image in the left and right breast area of interest other than the middle layer successively respectively.
Compared with prior art, enhance the breast in original image present invention incorporates the statistical distribution of skin of breast line gray scale
Room skin line；
Further, it is handled by the gradient saturation of different stage and eliminates artifact outside body of gland fibr tissue and skin line
Interference to skin line detection；
Further, the continuity features for introducing skin of breast line in adjacent scanning tomography, are accurately and efficiently realized
Full-automatic detection to skin of breast line.
Description of the drawings
Fig. 1 show a kind of flow diagram of the detection method of skin of breast line of the embodiment of the present invention；
Fig. 2 show the flow signal for the skin of breast line that the embodiment of the present invention enhances in left and right breast area of interest
Figure；
Fig. 3 show the embodiment of the present invention and determines tomographic image epithelium skin line among the left and right breast area of interest
Flow diagram；
The flow that Fig. 4 show the gradient image that the embodiment of the present invention obtains the first level gradient saturation processing is shown
It is intended to；
Fig. 5 show the embodiment of the present invention and is distributed accumulative histogram by the first threshold and gradient, obtains described the
The flow diagram of the gradient image of one rank gradient saturation processing；
Fig. 6 show the embodiment of the present invention and handles the flow signal for eliminating the fluctuation by second level gradient saturation
Figure；
Fig. 7 show the embodiment of the present invention and determines in the left and right breast area of interest other than the middle layer
Tomographic image on skin line flow diagram；
Fig. 8 show a kind of structural schematic diagram of the detection device of skin of breast line of the embodiment of the present invention.
Specific implementation mode
Many details are elaborated in the following description in order to fully understand the present invention.But the present invention can be with
Much implement different from other manner described here, those skilled in the art can be without prejudice to intension of the present invention the case where
Under do similar popularization, therefore the present invention is not limited to the specific embodiments disclosed below.
Secondly, the present invention is described in detail using schematic diagram, when describing the embodiments of the present invention, for purposes of illustration only, institute
It is example to state schematic diagram, should not limit the scope of protection of the invention herein.
The detection method and device of a kind of skin of breast line of the present invention are carried out in detail with reference to the accompanying drawings and examples
Explanation.The detection method of skin of breast line of the embodiment of the present invention is as shown in Figure 1, first, executing step S1, inputting breast tissue
Initial three-dimensional sequence image determines the sense of left and right breast according to breast tissue characteristic point in the picture and feature locations
Interest region.Specifically, in the present embodiment, by taking breast magnetic resonance image as an example, before carrying out image noise reduction processing, first
Establish human body coordinate system：The frontal axis of left and right directions is X-axis, and sagittal axis in the front-back direction is Y-axis, and the vertical axis of upper and lower directions is
Z axis.Standard deviation is calculated by the cross sections the MR image series (the Z values of i.e. every layer cross-sectional image are fixed) of the breast
Image, and standard difference image is divided automatically using the half of big law (Otsu) threshold value, combining form method with carry
Connected domain is taken to obtain final segmentation result.Then it regard segmentation result (image after dividing) as template, the breast
Each layer of MR cross-sectional image is all multiplied with the template, i.e., has carried out noise reduction to the breast magnetic resonance image, after noise reduction
Maximum value projection is done in each layer of cross section of breast magnetic resonance image along Z axis, obtained maximum value projection figure, and selected part drops
The cross-sectional image of breast magnetic resonance image after making an uproar does average value projection along Z axis, obtained average value perspective view.Described in calculating
The sum of the gray value of every a line all pixels point of average value perspective view chooses the maximum row of the sum of described gray value as initial
Lower boundary；On maximum value projection figure, all pixels point on initial lower boundary position is from left to right scanned, first gray scale becomes
Position where changing the pixel for not being 0 is first outer boundary in maximum value projection figure in X-direction, the last one
Grey scale change is not that the position where 0 pixel is second outer boundary in maximum value projection figure in X-direction, really
Area-of-interest the first outer boundary (left margin of left breast) in the X-axis direction and the second outer boundary (right side of right breast are determined
Boundary).It is cut-off rule by maximum value using the center line of the first outer boundary and the second outer boundary (i.e. right boundary) in X-direction
Perspective view is divided into two parts, i.e. first part and second part；To first part (left-hand component, the region where left breast)
It is scanned from top to bottom respectively with second part (right-hand component, the region where right breast), in the y that first part detects
The minimum non-zero pixels point of value is point P1, passing point P1And the straight line for being parallel to X-axis in maximum value projection figure is that first part is (left
Rim portion) the first outer boundary in Y direction；It is point P in the non-zero for the y value minimums that second part detects2, passing point P2And it is flat
The straight line of row X-axis in maximum value projection figure is the second outer boundary in second part (right-hand component) Y direction, that is, is determined
Area-of-interest the first outer boundary (coboundary of left breast) in the Y-axis direction and the second outer boundary (top of right breast
Boundary).
Then, according to the Y-coordinate value y of the initial lower boundary and Y-axis intersection point0, calculate the i-th of the breast 3-D view
The Y-coordinate value y of layer cross section and its Y-axis intersection pointi, wherein i indicates the level in the cross section of the breast 3-D view.It obtains
The intersection point of each layer of cross-sectional image and its Y-axis, by the intersection point and the straight line that is parallel to X-axis in this layer of cross section it is as every
Lower boundary in one layer of cross-sectional image in Y direction, the lower boundary in all cross-sectional images in Y direction is using three-dimensional
Plane fitting obtains the third outer boundary in the Y-axis direction, the lower boundary of the as described area-of-interest.
Then, in described first image Ixy' in, the boundary line of the area-of-interest is obtained by edge detection；It chooses
Point P on the boundary linei, calculate the point PiLine with m-th point of the front and the point PiWith m-th point behind
Line between included angle cosine value；According to the included angle cosine value, the inner edge of the area-of-interest in the X-axis direction is determined
Boundary.Selected point P respectively1With point P2N-layer is maximum value projection, point P along X-axis before and after sagittal plane where respectively1With point P2It is described
The characteristic point of breast tissue in the picture, point P1With point P2The position at place is also the position where the nipple of left and right breast, is obtained
To the first image I of two dimensionyz' and the second image I of two dimensionyz", wherein the two-dimensional image IyzFor the two-dimentional first image Iyz’
With the second image I of two dimensionyz”；The left two-dimensional image IyzWith right two-dimensional image IyzBy edge detection, determine described interested
The boundary of region in the Z-axis direction.By the above method, according to feature point P1And P2And the left side of the left breast of feature locations
Boundary, the right margin of right breast, the coboundary of left breast, the coboundary of right breast, the inner boundary of breast and lower boundary and Z-direction
On boundary, obtain the area-of-interest of left and right breast, and obtain the central point of the left and right breast area of interest.At this
In embodiment, 160 layers of three-dimensional magnetic resonance breast image are inputted, area-of-interest is the image between the 20th layer to the 120th layer.
Then, step S2 is executed, according to the statistical distribution of skin line feature, enhances the left and right breast area of interest
Skin of breast line on middle image.
Then, step S3 is executed, respectively according to the intermediate tomographic image of the left and right breast area of interest, passes through dynamic
Planing method determines the skin line on tomographic image among the left and right breast area of interest.In the present embodiment, in described
Interbed is the layer where the nipple of the left and right breast.Middle layer may be the figure of the most middle layer of the area-of-interest
Picture can also be a certain tomographic image near the most intermediate tomographic image of the area-of-interest.
Then, step S4 is executed, the skin line on the intermediate tomographic image based on the left and right breast area of interest divides
The skin line on the tomographic image in the left and right breast area of interest other than the middle layer is not determined successively.
Preferably, in the present embodiment, enhance the process of the skin of breast line of the left and right breast area of interest as schemed
Shown in 2, first, step S201 is executed, determine the data set of the statistical distribution for obtaining the skin line feature.In this implementation
In example, the statistical distribution of the skin line feature is distributed for gray-scale statistical.Method manually or semi-automatically will be used to obtain
Skin line in several three-dimensional magnetic resonance breast images of the skin line gray-scale statistical distribution is determined to get to for obtaining
Take the position of the skin of breast line of the statistical distribution of the skin line feature.Then, step S202 is executed, is collected based on the data
The position of middle skin of breast line and feature obtain the statistical distribution of the skin line feature.Specifically, in the present embodiment, root
According to the corresponding gray scale in position and the position of skin of breast line, the percentage of skin line grey value profile is obtained.Then, it executes
Step S203 respectively unites the gray scale of each pixel in the left and right breast area of interest and the skin line gray scale
The distribution percentage value of corresponding gray scale is multiplied in score cloth, to enhance the skin of breast line in left and right breast area of interest.
Preferred process such as Fig. 3 of tomographic image epithelium skin line among the left and right breast area of interest is determined in step S3
It is shown, first, step S301 is executed, according to ray scanning starting point, ray scanning radius and flying spot number, is sat based on pole
Tomographic image among the left and right breast area of interest is converted to two-dimensional transformations figure by mark conversion and ray scanning method respectively
Picture.Wherein, the ray scanning starting point is that the central point of the left and right breast is left and right described in the intermediate tomographic image
Subpoint on the lower boundary of breast area of interest；Distance of the ray scanning starting point apart from the left and right papilla of breast
Determine the ray scanning radius.In the present embodiment, for left breast, using the layer where left papilla of breast as middle layer figure
Picture, i.e., the 80th layer, in step S1 determine left breast central point in the intermediate tomographic image left breast area of interest
Lower boundary on subpoint be ray scanning starting point, with the ray scanning starting point apart from the left papilla of breast away from
It, will be in the left breast area of interest by polar coordinates conversion and ray scanning method from being the sweep radius plus 14mm
Between tomographic image be converted to two-dimensional transformations image, x-axis in the two-dimensional transformations image indicates that the serial number of flying spot, y-axis indicate
Distance of the point apart from ray scanning starting point on ray.Wherein, the number of rays of scanning is 181.Pass through ray scanning method
Obtained each flying spot, it is exactly two dimension that each flying spot, which is transformed by polar coordinate method in two-dimensional transformations image,
Each row in image, and the point on each row indicates the point that each flying spot passes through in intermediate tomographic image.By same
Tomographic image among the right breast area of interest is converted to two-dimensional transformations image by the step of sample.Through the above steps, two-dimentional
Skin line in breast tissue image (intermediate tomographic image) is resampled.
Then, step S302 is executed, the object boundary in the two-dimensional transformations image is obtained by the method for Dynamic Programming
Line, the object boundary line be skin line among the left and right breast area of interest on tomographic image after conversion
Position in two-dimensional transformations image.Specifically, in Dynamic Programming, (ray scanning method obtains each row of two-dimensional transformations image
The each flying spot obtained) it is considered as the stage, the point on each row of two-dimensional transformations image is considered as the candidate on the stage
Point, from the first stage, the stage (first row of two-dimensional transformations image to a last row) has minimum accumulation local energy to the end
Path is considered as optimal path, i.e. object boundary line (skin line of middle layer).It is every therefore, it is necessary to establish two conversions dimension image
The local energy equation of point on one row, which can be according to the gray scale, gradient or shape of breast 3-D view
Feature is established, can be there are many representation.Local energy is codetermined by internal energy and external energy, wherein internal energy
Amount determines that the flatness of optimal path (object boundary line), external energy determine that optimal path is located at the big position of gradient.
Dynamic Programming is carried out according to local energy equation, to find object boundary line, determines intermediate tomographic image epithelium skin line in two dimension
Convert the position in image.
According to above-mentioned described, it is thus necessary to determine that the local energy of the point on each row of two-dimensional transformations image, in the present embodiment
In, the formula of the local energy equation in the dynamic programming method is expressed as：
C (i)=dis (i, i-1)/max (dis)+g1(i)/max (g1) (1)
Wherein, i indicates that stage, the stage are each row of the two-dimensional transformations image, the two-dimensional transformations image
Each to be classified as the ray obtained by ray scanning method sequence, the gray value of the pixel on each row is ray scanning
The gray value for the pixel that the ray that method sequence obtains passes through in tomographic image among the left and right breast area of interest；C
(i) local energy for being point P (j) on the i-th stage；Dis (i, i-1) indicates that point Q (k) exists on the point P (j) to the (i-1)-th stage
The distance projected on i-th stage；Max (dis) indicates the maximum value in all distances in the two-dimensional transformations image；g1(i) table
Show that point P (j) is in the gradient image g after the processing of first level saturation on the i-th stage1In image gradient；max(g1) indicate
All the points are by the maximum value in first level saturation treated gradient image in two-dimensional transformations image.Max (dis) and
max(g1) for normalizing its corresponding entry in internal energy definition.The characteristics of according to skin of breast line in the picture, skin of breast
Line is located at the position with larger Grad that bright gray scale is changed to by dark gray.
Then, step S303 is executed, whether the object boundary line obtained in judgment step S302 has fluctuation, if having wave
It is dynamic, S304 is thened follow the steps, is handled by second level gradient saturation and eliminates the fluctuation, followed by executing step S305,
Based on the object boundary line, the left and right breast area of interest middle layer figure is obtained by the method for polar coordinates inverse conversion
As upper skin line；If the object boundary line does not have fluctuation, step S305 is directly executed, is based on the object boundary
Line obtains the skin line among the left and right breast area of interest on tomographic image by the method for polar coordinates inverse conversion.This hair
The neighbor pixel that the fluctuation of bright embodiment refers on curve is non-conterminous in the picture.
The point that Dynamic Programming obtains on curve on two-dimensional transformations image should be continuous on spatial position, still, by
In the neighbouring of body of gland fibr tissue in breast and skin of breast line, the interference detected to skin of breast line is caused, is thus caused
There is downward fluctuation in the curve that Dynamic Programming obtains on two-dimensional transformations image.Therefore, it is necessary to the fluctuations to object boundary line
It is detected.Specifically, in the present embodiment, if on object boundary line between all neighbor pixels in two-dimensional transformations image
Range difference on y-axis direction is less than or equal to 1 pixel, then shows that object boundary line is continuous, then follow the steps S305.If
Range difference on object boundary line between neighbor pixel in two-dimensional transformations image on y-axis direction is more than or equal to 2 pixels,
Then indicate that neighbor pixel is non-conterminous in the picture on object boundary line, is to have fluctuation, thens follow the steps S304, needs
It is handled by second level gradient saturation and eliminates the fluctuation.
Then, step S305 is executed, the object boundary line is based on, described in the method acquisition by polar coordinates inverse conversion
Skin line among left and right breast area of interest on tomographic image.Wherein, on intermediate tomographic image polar coordinates inverse conversion skin line
In each image column at least will there are one pixel, if not having, on the skin line of detection existing previous pixel and
Linear interpolation is carried out among the latter pixel, to obtain the skin line of left and right breast middle layer.
In dynamic programming process, the gradient image g of the processing of first level gradient saturation described in step S3021It is excellent
Acquisition process is selected as shown in figure 4, first, executing step S401, the two-dimensional transformations image that the step S301 is obtained is carried out high
This is smooth, obtains smoothed image.Then, step S402 is executed, the smoothed image is based on, obtains Initial Gradient image, and count
Calculate the gradient average value of the Initial Gradient image.Specifically, smoothed image is sought turning in two dimension based on two-dimensional Gaussian kernel
The gradient (partial derivative) on the y-axis direction of image is changed to get to the Grad of each pixel of smoothed image, by each picture of smoothed image
For the Grad linear change of vegetarian refreshments to the range of 0-1023, image at this time is Initial Gradient image.It should be noted that flat
The Grad of each pixel of sliding image can also change to 0-512 or other ranges.It calculates in Initial Gradient image and owns again
The gradient average value of pixel.Followed by execution step S403 is based on the Initial Gradient image and two-dimensional transformations image, really
Surely it interferes pixel, and assigns the gradient average value of the Initial Gradient image to the interference pixel, and rest of pixels point
Grad it is constant, as gradient image g.Specifically, according to the gray feature around skin of mammary gland line, in order to eliminate it is unrelated because
Influence of the element to dynamic programming process, therefore, it is necessary to some pixels influential on dynamic programming process (are interfered pixel
Point) it determines.The pixel M (m) that Grad in Initial Gradient image is less than 512 is chosen, for these pixels, in two dimension
It converts in image, calculates pixel M (m) in the average gray of M (5-20) a pixel vertically upward, the gray scale
It is 10-80 that average value, which is more than preset threshold range, then the pixel M (m) is interference pixel, and interference pixel is
Pixel inside mammary gland fibr tissue.It should be noted that if each pixel of smoothed image in step S402
Grad changes to the range of 0-512, then chooses the pixel that Grad in Initial Gradient image is less than 256.In the present embodiment
In, the value of M is 7mm (10 pixels), and threshold value is taken as 30.And assign the gradient average value of the Initial Gradient image to institute
State interference pixel, and the Grad of rest of pixels point is constant to get to gradient image g.Then, step S404, statistics are executed
The gradient of the gradient image g is distributed, and obtains gradient profile accumulation histogram.Then, step S405 is executed, first threshold is passed through
With the gradient profile accumulation histogram, the gradient image g of the first level gradient saturation processing is obtained1。
Specifically, the first order is obtained by the first threshold and gradient profile accumulation histogram in step S405
The gradient image g of other gradient saturation processing1Preferred process as shown in figure 5, first, step S501 is executed, according to the ladder
Profile accumulation histogram is spent, the first gradient value corresponding to the first threshold is obtained.In the present embodiment, the first threshold of setting
Value is 1%, in the gradient profile accumulation histogram, finds corresponding Grad, as first gradient value.Then, it executes
All Grad in the gradient image g are compared, wherein institute by step S502 with first gradient value and gradient average value
The average value that gradient average value is all Grad of gradient image g is stated, if described in Grad is less than in the gradient image g
The Grad is then assigned a value of 0 by first gradient value；If Grad is more than gradient average value in the gradient image g, by the ladder
Angle value is assigned a value of the gradient average value, remaining Grad remains unchanged, to obtain newer gradient image g.Then, it executes
Step S503 stretches newer gradient image g using first gradient value and gradient average value, obtains the first level
The gradient image g of gradient saturation processing1.For left and right breast, respectively according to above-mentioned local energy equation (1), can obtain
Object boundary line, that is, position of the skin line on intermediate tomographic image detected in two-dimensional transformations image.
In the present embodiment, the preferred mistake for eliminating the fluctuation is handled in step S304 by second level gradient saturation
Journey is as shown in fig. 6, first, executing step S601 and being obtained by institute according to the gradient profile accumulation histogram and second threshold
State the gradient image g of second level saturation processing2.Specifically, in the present embodiment, second threshold 10%, in gradient point
Grad corresponding with the second threshold, i.e. the second Grad are obtained in cloth accumulation histogram.Grad in gradient image g
Grad less than the second Grad is assigned 0, and Grad is more than the ladder of all gradients in gradient image g in gradient image g
The Grad of degree average value is assigned gradient average value, remaining Grad is constant to get to newer gradient image g, then profit
Linear stretch is carried out to newer gradient image g with the second Grad and gradient average value, is obtained full by the second level
With the gradient image g for changing processing2.Then, step S602 is executed, according to the gradient image g2, Dynamic Programming is carried out, is obtained most
Excellent curve.Specifically, in the present embodiment, Dynamic Programming is carried out in two-dimensional transformations image by formula (2), wherein described
Formula (2) is：
C (i)=dis (i, i-1)/max (dis)+g2(i)/max (g2) (2)
Wherein, i indicates that stage, the stage are each row of the two-dimensional transformations image, the two-dimensional transformations image
Each to be classified as the ray obtained by ray scanning method sequence, the gray value of the pixel on each row is ray scanning
The gray value for the pixel that the ray that method sequence obtains passes through in tomographic image among the left and right breast area of interest；C
(i) local energy for being point P (j) on the i-th stage；Dis (i, i-1) indicates that point Q (k) exists on the point P (j) to the (i-1)-th stage
The distance projected on i-th stage；Max (dis) indicates the maximum value in all distances in the two-dimensional transformations image；g2(i) table
Show that point P (j) is in the gradient image g after the processing of second level saturation on the i-th stage2In image gradient；max(g2) indicate
All the points are by the maximum value in second level saturation treated gradient image in two-dimensional transformations image.
Then, step S603 is executed, with the optimal curve in the gradient image g2Position in each column is starting point, according to
Secondary determination is in the gradient image g2First position with greatest gradient value in vertical downward direction in each column；By gradient map
As g1The Grad of the downward all pixels point in the position is assigned a value of the greatest gradient value in correspondence image row, i.e., described in update
Gradient image g1.Specifically, although the optimal curve obtained after being handled using second level gradient saturation can remove breast
The influence that interior body of gland fibr tissue detects skin of breast line still due to the influence of artifact outside skin of breast line, leads to second
The optimal curve obtained after rank gradient saturation and Dynamic Programming processing is located at the outside of practical skin of breast line.Therefore, with
The optimal curve is in the gradient image g2Position in each column is starting point, successively in gradient image g2In each column vertically downward
Find in the image column position where the Grad that first Grad is 255, and by gradient image g1In correspondence image row
The Grad of all pixels point of the position vertically downward is assigned a value of 255, and the Grad of remaining position is constant, that is, has updated described
Gradient image g1.Then, step S604 is executed, according to the newer gradient image g1, Dynamic Programming is carried out, the mesh is obtained
Mark boundary line.Specifically, Dynamic Programming is carried out according to formula (1), obtains object boundary line.
Specifically, in the present embodiment, it determines in the left and right breast area of interest other than the middle layer
The preferred process of skin line on tomographic image is as shown in fig. 7, first, execute step S701, in the layer in addition to middle layer
In, the left and right breast area of interest middle layer both sides select image successively by from the order of middle layer from the near to the distant respectively
Layer is used as current layer, obtains the two-dimensional transformations image and gradient image of the current layer.Specifically, in the present embodiment, with a left side
For breast, middle layer is the 80th layer, respectively forwardly and is backward successively handled, then successively handled forward since middle layer
When handle at first is the 79th layer, what is handled at first when successively processing backward is the 81st layer.To obtain the 79th layer of figure
For the skin layer of picture, then current layer is the 79th layer, and the 79th layer of two-dimensional transformations figure is obtained by step S301 and step S302
Picture and gradient image.
Then, step S702 is executed, based on gaussian kernel function and the current layer preceding layer or latter tomographic image epithelium skin
The testing result of line is improved the current layer gradient image.Specifically, in the 79th layer of two-dimensional transformations image, with
During the position of the object boundary line (skin line) detected on later layer (the 80th layer, middle layer) each row of two-dimensional transformations image is
The heart to get to middle layer object boundary line (skin line) and each row in the 79th layer of two-dimensional transformations image intersection point, by the 79th
Layer two-dimensional transformations image respective column on Grad be multiplied by one centered on the intersection point one-dimensional gaussian kernel function (sigma=
5) weighted value determined, to obtain improved 79th layer of gradient image.This is because the position of adjacent level skin of breast line
It is smaller to set variation, can largely retain the gradient of current layer skin line position using the improved method, and press down
Make the gradient in non-skin of breast line region so that subsequent dynamic program results are more accurate.It should be noted that if current layer is
81st layer, then the 80th layer is its preceding layer.
Then, it executes step S703 and object boundary line is obtained by the method for Dynamic Programming, the object boundary line is
Position of the skin line in the two-dimensional transformations image on the current tomographic image.Specifically, based on being obtained in step S702
Improved gradient image, the 79th layer of object boundary line (skin is obtained according to the method for the Dynamic Programming of step S302
Line).
Followed by execution step S704 judges whether the object boundary line of the current layer has fluctuation.Specifically, sentence
Whether disconnected 79th layer of object boundary line (skin line) has fluctuation, if so, thening follow the steps S705, is saturated by second level
Change processing and eliminate the fluctuation, specific process is as the method for eliminating the fluctuation on intermediate tomographic image epithelium skin line.If nothing
Fluctuation, thens follow the steps S706, is the 79th tomographic image by the method migration that the object boundary passes through polar coordinates inverse conversion
On skin line, specific process is as the method for the intermediate tomographic image of processing.According to the method described above, then with the 78th layer it is current
Layer, the 79th layer is preceding layer, obtains the skin line on the 78th tomographic image, and so on, the skin on each tomographic image can be obtained
Line.According to the method described above, the skin line being similarly obtained on each tomographic image in right breast area of interest.Through the above steps, it detects
Obtain the skin line of the breast tissue.
Corresponding to the detection method of above-mentioned skin of breast line, the embodiment of the present invention additionally provides a kind of dress of skin of breast line
It sets, as shown in figure 8, including determining territory element 1 interested, enhancement unit 2, middle layer skin line determination unit 3 and skin line
Determination unit 4.
The area-of-interest determination unit 1 is suitable for inputting the initial three-dimensional sequence image of breast tissue, according to the breast
Room tissue characteristic point in the picture and feature locations, determine the corresponding area-of-interest of left and right breast.
The enhancement unit 2 is suitable for the statistical distribution according to skin line feature, enhances the left and right breast area of interest
In skin of breast line.Specifically, it is determined that the data set of the statistical distribution for obtaining the skin line gray scale；Based on the number
According to the corresponding gray scale in position and the position for concentrating skin of breast line, the statistical distribution of the skin line gray scale is obtained；Respectively
It will be opposite in the statistical distribution of the feature of each pixel in the left and right breast area of interest and the skin line gray scale
The distribution percentage value of gray scale is answered to be multiplied.
The middle layer skin line determination unit 3 is suitable for respectively according to the middle layer of the left and right breast area of interest
Image determines the skin line on tomographic image among the left and right breast area of interest, wherein institute by dynamic programming method
State the layer where the nipple that middle layer is the left and right breast.
The skin line determination unit 4 is suitable for the skin on the intermediate tomographic image based on the left and right breast area of interest
Skin line determines the skin on the tomographic image in the left and right breast area of interest other than the middle layer successively respectively
Line.
The cooperation of each unit and the course of work can refer to above-mentioned skin of breast line in above-mentioned skin of breast line detector
The explanation of detection method, details are not described herein.
Although the invention has been described by way of example and in terms of the preferred embodiments, but it is not for limiting the present invention, any this field
Technical staff without departing from the spirit and scope of the present invention, may be by the methods and technical content of the disclosure above to this hair
Bright technical solution makes possible variation and modification, therefore, every content without departing from technical solution of the present invention, and according to the present invention
Technical spirit to any simple modifications, equivalents, and modifications made by above example, belong to technical solution of the present invention
Protection domain.
Claims (12)
1. a kind of detection method of skin of breast line, which is characterized in that include the following steps：
(1) the initial three-dimensional sequence image for inputting breast tissue, according to the characteristic point and feature of the breast tissue in the picture
Position determines the area-of-interest of left and right breast；
(2) according to the statistical distribution of skin line feature, enhance the skin of breast on image in the left and right breast area of interest
Line；
(3) respectively according to the intermediate tomographic image of the left and right breast area of interest, by dynamic programming method, determine described in
Skin line among left and right breast area of interest on tomographic image；
(4) skin line on intermediate tomographic image based on the left and right breast area of interest, determine successively respectively it is described it is left,
The skin line on tomographic image in right breast area of interest other than the middle layer；
Determine that the process of tomographic image epithelium skin line among the left and right breast area of interest is as follows：
According to ray scanning starting point, ray scanning radius and flying spot number, based on polar coordinates conversion and ray scanning side
Tomographic image among the left and right breast area of interest is converted to two-dimensional transformations image by method respectively；Wherein, the ray scanning
Starting point is central point left and right breast area of interest lower boundary described in the intermediate tomographic image of the left and right breast
On subpoint；Distance of the sweep radius by the ray scanning starting point apart from the left and right papilla of breast determines；
The object boundary line in the two-dimensional transformations image is obtained by the method for Dynamic Programming, the object boundary line is institute
State position of the skin line among left and right breast area of interest on tomographic image after conversion in two-dimensional transformations image；
Judge whether the object boundary line has fluctuation；
If with fluctuation, is handled by second level gradient saturation and eliminate the fluctuation, and be based on the object boundary
Line obtains the skin line among the left and right breast area of interest on tomographic image by the method for polar coordinates inverse conversion；If institute
State object boundary line and do not have fluctuation, be then based on the object boundary line, by the method for polar coordinates inverse conversion obtain it is described it is left,
Skin line among right breast area of interest on tomographic image.
2. a kind of detection method of skin of breast line as described in claim 1, which is characterized in that the enhancing left and right breast sense
The process of skin of breast line in interest region is：
(1) data set of the statistical distribution for obtaining the skin line feature is determined；
(2) position and the feature for concentrating skin of breast line based on the data, obtain the statistical distribution of the skin line feature；
(3) respectively by the system of the feature of each pixel and the skin line feature in the left and right breast area of interest
The distribution percentage value of corresponding feature is multiplied in score cloth.
3. a kind of detection method of skin of breast line as claimed in claim 2, which is characterized in that the statistics of the skin line feature
It is distributed as gray-scale statistical distribution.
4. a kind of detection method of skin of breast line as described in claim 1, which is characterized in that in the dynamic programming method
The formula of local energy equation is expressed as：
C (i)=dis (i, i-1)/max (dis)+g1(i)/max(g1) (1)
Wherein, i indicate the stage, the stage be the two-dimensional transformations image each row, the two-dimensional transformations image it is each
It is classified as the ray obtained by ray scanning method sequence, the gray value of the pixel on each row is to pass through ray scanning
The gray value for the pixel that the ray that method sequence obtains passes through in tomographic image among the left and right breast area of interest；C
(i) local energy for being point P (j) on the i-th stage；Dis (i, i-1) indicates that point Q (k) exists on the point P (j) to the (i-1)-th stage
The distance projected on i-th stage；Max (dis) indicates the maximum value in all distances in the two-dimensional transformations image；g1(i) table
Show that point P (j) is in the gradient image g after the processing of first level saturation on the i-th stage1In image gradient；max(g1) indicate
All the points are by the greatest gradient value in first level saturation treated gradient image in two-dimensional transformations image.
5. a kind of detection method of skin of breast line as claimed in claim 4, which is characterized in that the first level gradient saturation
Change the gradient image g of processing1Acquisition process it is as follows：
(1) Gaussian smoothing is carried out to the two-dimensional transformations image, obtains smoothed image；
(2) it is based on the smoothed image, obtains Initial Gradient image, and calculate the gradient average value of the Initial Gradient image；
(3) it is based on the Initial Gradient image and two-dimensional transformations image, determines interference pixel, and by the Initial Gradient image
Gradient average value assign the interference pixel, and the Grad of rest of pixels point is constant, as gradient image g；
(4) the gradient distribution for counting the gradient image g, obtains gradient profile accumulation histogram；
(5) by first threshold and the gradient profile accumulation histogram, the first level gradient saturation processing is obtained
Gradient image g1。
6. a kind of detection method of skin of breast line as claimed in claim 5, which is characterized in that described to obtain Initial Gradient image
Process be：Partial derivative based on two-dimensional Gaussian kernel on the directions y of the two-dimensional transformations image, it is each to obtain the smoothed image
The Grad of pixel；It is as described first by the Grad linear change of each pixel of the smoothed image to scheduled range
Beginning gradient image.
7. a kind of detection method of skin of breast line as claimed in claim 5, which is characterized in that the determination of the interference pixel
Process is as follows：
(1) the pixel M (m) that Grad in the Initial Gradient image is less than 512 is chosen；
(2) in the two-dimensional transformations image, calculate the pixel M (m) M pixel vertically upward gray scale
Average value, the average gray are more than preset threshold value, then the pixel M (m) is interference pixel.
8. a kind of detection method of skin of breast line as described in right wants 5, which is characterized in that pass through the first threshold and gradient
Profile accumulation histogram obtains the gradient image g of the first level gradient saturation processing1Process be：
(1) according to the gradient profile accumulation histogram, the first gradient value corresponding to the first threshold is obtained；
(2) all Grad in the gradient image g are compared with first gradient value and gradient average value, wherein described
Gradient average value is the average value of all Grad of gradient image g, if Grad is less than described the in the gradient image g
One Grad, then be assigned a value of 0；If Grad is more than the ladder of all Grad of gradient image g in the gradient image g
Average value is spent, then is assigned a value of the gradient average value；Otherwise, Grad remains unchanged, and obtains newer gradient image g；
(3) by the first gradient value and the gradient average value, linear stretch is carried out to the newer gradient image g.
9. a kind of detection method of skin of breast line as claimed in claim 8, which is characterized in that the first threshold is 1%.
10. a kind of detection method of skin of breast line as claimed in claim 5, which is characterized in that pass through second level saturation
The process that the fluctuation is eliminated in processing is as follows：
(1) it according to the gradient profile accumulation histogram and second threshold, obtains handling by the second level saturation
Gradient image g2；
(2) according to the gradient image g2, Dynamic Programming is carried out, optimal curve is obtained；
(3) with the optimal curve in the gradient image g2Position in each column is starting point, is determined successively in the gradient image
g2First position with greatest gradient value in vertical downward direction in each column；By the gradient image g1In correspondence image row
The Grad of the downward all pixels point in the position is assigned a value of the greatest gradient value, that is, updates the gradient image g1；
(4) according to the newer gradient image g1, Dynamic Programming is carried out, the object boundary line is obtained.
11. a kind of detection method of skin of breast line as claimed in claim 10, which is characterized in that the second threshold is
10%.
12. a kind of detection method of skin of breast line as claimed in claim 10, which is characterized in that the greatest gradient value is
255。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710657857.4A CN107563992B (en) | 2014-03-11 | 2014-03-11 | Method and device for detecting breast skin line |
| CN201410088614.XA CN104915942B (en) | 2014-03-11 | A kind of detection method and device of skin of breast line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410088614.XA CN104915942B (en) | 2014-03-11 | A kind of detection method and device of skin of breast line |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710657857.4A Division CN107563992B (en) | 2014-03-11 | 2014-03-11 | Method and device for detecting breast skin line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104915942A CN104915942A (en) | 2015-09-16 |
| CN104915942B true CN104915942B (en) | 2018-08-31 |
Family
ID=
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101664318A (en) * | 2008-09-03 | 2010-03-10 | 爱克发医疗保健公司 | Method for deriving the amount of dense tissue from a digital mammographic image representation |
| WO2013120834A1 (en) * | 2012-02-17 | 2013-08-22 | Agfa Healthcare | Method for defining a region of interest in a radiation image of a breast |
| WO2013127730A1 (en) * | 2012-02-27 | 2013-09-06 | Agfa Healthcare | Image alignment of breast images |
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101664318A (en) * | 2008-09-03 | 2010-03-10 | 爱克发医疗保健公司 | Method for deriving the amount of dense tissue from a digital mammographic image representation |
| WO2013120834A1 (en) * | 2012-02-17 | 2013-08-22 | Agfa Healthcare | Method for defining a region of interest in a radiation image of a breast |
| WO2013127730A1 (en) * | 2012-02-27 | 2013-09-06 | Agfa Healthcare | Image alignment of breast images |
Non-Patent Citations (1)
| Title |
|---|
| 乳腺X线图像结构扭曲检测的一种新方法;张胜君;《系统工程理论与实践》;20130331;第33卷(第3期);全文 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107480677B (en) | Method and device for identifying interest region in three-dimensional CT image | |
| EP2916738B1 (en) | Lung, lobe, and fissure imaging systems and methods | |
| US8447383B2 (en) | System and method for providing 2-dimensional computerized-tomography image corresponding to 2-dimensional ultrasound image | |
| US8270696B2 (en) | Image slice segmentation using midpoints of contour anchor points | |
| US10991102B2 (en) | Image processing apparatus and image processing method | |
| US20120065499A1 (en) | Medical image diagnosis device and region-of-interest setting method therefore | |
| CN109840913B (en) | Method and system for segmenting tumor in mammary X-ray image | |
| US9367924B2 (en) | Method and system for segmentation of the liver in magnetic resonance images using multi-channel features | |
| CN107680107B (en) | Automatic segmentation method of diffusion tensor magnetic resonance image based on multiple maps | |
| US8082019B2 (en) | System and method for magnetic resonance brain scan planning | |
| CN102890823A (en) | Motion object outline extraction and left ventricular image segmentation method and device | |
| US9542741B2 (en) | Method and system for automatic pelvis unfolding from 3D computed tomography images | |
| CN104881858B (en) | The extracting method and device of enhancing background tissues in a kind of breast | |
| CN105389811A (en) | Multi-modality medical image processing method based on multilevel threshold segmentation | |
| JP2017510427A (en) | Radiation image lung segmentation technology and bone attenuation technology | |
| KR102519515B1 (en) | Information processing device, information processing method, computer program | |
| US10521906B2 (en) | Method and image processor for evaluating a contrast agent-enhanced magnetic resonance slice image of a heart | |
| CN106537452A (en) | Device, system and method for segmenting an image of a subject. | |
| JP2009211138A (en) | Target area extraction method, device and program | |
| CN103927732B (en) | A kind of detection method of wall of the chest line | |
| US9147250B2 (en) | System and method for automatic magnetic resonance volume composition and normalization | |
| JP2008146278A (en) | Cell outline extraction device, cell outline extraction method and program | |
| CN105787978A (en) | Automatic medical image interlayer sketching method, device and system | |
| CN107563992B (en) | Method and device for detecting breast skin line | |
| CN104915942B (en) | A kind of detection method and device of skin of breast line |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 201815 Shanghai City, Jiading District Jiading Industrial Zone No. 1180 Building 8 Hing Road Patentee after: Shanghai Lianying Medical Technology Co., Ltd Address before: 201815 Shanghai City, Jiading District Jiading Industrial Zone No. 1180 Building 8 Hing Road Patentee before: SHANGHAI UNITED IMAGING HEALTHCARE Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |
Address after: 201807 2258 Chengbei Road, Jiading District, Shanghai Patentee after: Shanghai Lianying Medical Technology Co.,Ltd. Address before: 201815 No. 1180 Xingxian Road, Jiading Industrial Zone, Jiading District, Shanghai Patentee before: Shanghai Lianying Medical Technology Co.,Ltd. |