CN108090909A - A kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model - Google Patents
A kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model Download PDFInfo
- Publication number
- CN108090909A CN108090909A CN201711347973.2A CN201711347973A CN108090909A CN 108090909 A CN108090909 A CN 108090909A CN 201711347973 A CN201711347973 A CN 201711347973A CN 108090909 A CN108090909 A CN 108090909A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- image
- msubsup
- function
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/149—Segmentation; Edge detection involving deformable models, e.g. active contour models
-
- 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
- G06T2207/10132—Ultrasound image
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of ultrasonic contrast image partition methods based on statistics Partial Differential Equation Model, and including pre-treatment step and online segmentation step, image and initial profile to be split are obtained in pre-treatment step;Using image to be split and initial profile as input quantity in online segmentation step, obtain energy function, Edge-stopping function is introduced in energy function with constraint length item, and increase LoG image energy functions, numerical solution is carried out to energy function, quickly the object edge of single or multiple different ROI images is captured and split.ROI in ultrasonic contrast image is split, by the segmentation result for observing entire sequence, doctor can be aided in obtain focal area and same level reference zone, the time-activity curve of lump such as in the superficial organs such as liver, kidney abdomen organa parenchymatosum and mammary gland, thyroid gland, so as to analyze to obtain quantitative index.
Description
Technical field
The present invention relates to a kind of ultrasonic contrast image partition method, more particularly to a kind of based on statistics Partial Differential Equation Model
Ultrasonic contrast image partition method.
Background technology
The cutting techniques of image in the qualitative and quantitative analysis of medical ultrasonic contrastographic picture in occupation of highly important
Position.In the application of clinical ultrasound radiography, ultrasonic doctor is often by the different area-of-interest of subjective analysis (Region of
Interest, ROI) the perfusion intensity of dynamic image, the uniformity, reperfusion mode (diffusivity enhancing/centrality enhancing) judge
The blood supply situation of lesion.However, it is that ultrasonic contrast obtains the result is that a video or Dicom sequences, and there are noise, gray scales point
The problems such as cloth is uneven, the soft edge of target object, ultrasonic doctor is difficult to subjective analysis.
The content of the invention
It is contemplated that at least solving technical problem in the prior art, especially innovatively propose a kind of based on system
Count the ultrasonic contrast image partition method of Partial Differential Equation Model.
In order to realize the above-mentioned purpose of the present invention, the present invention provides a kind of ultrasonic contrasts based on statistics Partial Differential Equation Model
Image partition method, including pre-treatment step and online segmentation step,
The pre-treatment step is based on ultrasonic contrast image, obtains image and initial profile to be split;
The online segmentation step obtains energy function, in the energy using image to be split and initial profile as input quantity
Edge-stopping function is introduced in flow function with constraint length item, and increases LoG image energy functions;Numerical value is carried out to energy function
It solves, completes that the object edge of single or multiple difference ROI is captured and split.
In the another kind of preferred embodiment of the present invention, the energy function is:
Wherein, φ is level set function;u1Simplify description for the gray average function in active contour interior zone;u2
Simplify description for the gray average function in active contour perimeter;For the variance function in active contour interior zone
Simplify description;Simplify description for the variance function in active contour perimeter.
Preferred embodiment is planted again in the present invention, the Edge-stopping function is:
Wherein,It is for standard deviationGaussian kernel functionWith image I's to be split
Convolution algorithm formula, takesX is the point in image to be split, and y is the point in the field of point x;For gradient operator.
Preferred embodiment is planted again in the present invention, the ElgdEnergy term is:
Wherein, u1(x) and u2(x) be respectively active contour interior zone Ω1With perimeter Ω2Gray average;WithThe respectively interior zone Ω of active contour1With perimeter Ω2Gray variance;Y is in the neighborhood of x
Point, radius of neighbourhood r;Kσ(x-y) it is gaussian kernel function, is defined as
WithRespectively Ω1And Ω2The probability density function in region, point
It is not:
Wherein, I (y) is the gray value of image I midpoint y.
Preferred embodiment is planted again in the present invention, the ERS(φ) energy term is:
ERS(φ)=μ P (φ)+vL*(φ)+ωMΔ(φ);
Wherein, the P (φ) is regularization term;MΔ(φ) is the LoG item of image energy functions after optimization;L*(φ) is increase
The new length item of Edge-stopping function, is defined as:
To be carried out first with extra large formula function pair level set function φ (x)
Processing, then seek the gradient image of H (φ (x));μ, v and ω are every weight coefficient.
Preferred embodiment is planted again in the present invention, the regularization term P (φ) is:
Wherein, φ (x) is the level set function of x.
Preferred embodiment is planted again in the present invention, the LoG MΔ(φ) is defined as:
MΔ(φ)=∫ΩH(φ)·M(x)dx;
Wherein, M (x) is gray value of the LoG images at point x, is defined as:
In above formula, β is positive number, and M is LoG images;Δ(Gσ* I) it is second order partial differential edge detection operator.
Preferred embodiment is planted again in the present invention, the numerical solution process of the energy function obtains ladder first to calculate
Flow equation is spent, then iterative numerical computing is carried out to gradient flow equation;
Gradient flow equation is:
Wherein, λ1,λ2, μ, v and ω are every weight coefficient;It is curvature, Δ is Laplace operator, δ
(φ) is the one-dimensional Dirac function of level set function;ei(x) it is defined as:
I is 1 or 2, e1And e2Respectively e1(x) and e2(x) write a Chinese character in simplified form;
The condition of convergence of iterative numerical computing is:
Or iterations iter≤P;
Wherein, P is greatest iteration step number, and iter is iterations sequence number, and Γ is pixel number threshold value;For the pixel number of+1 active contour of pth,For pth time active contour.
Preferred embodiment is planted again in the present invention, the detailed process of the pre-treatment step is:
S1:Ultrasonic contrast image is converted into original sequence;
S2:Down-sampling processing is carried out using quadratic interpolattion to original sequence;
And/or image noise reduction processing is carried out using Gassian low-pass filter;
And/or picture smooth treatment is carried out using guiding filtering;
Obtain treated image sequence;
S3:One or more different ROI regions are chosen in image sequence after treatment, ROI image is obtained, utilizes Hai Shi
Function asks for ROI image binary map;
S4:The profile of ROI image binary map is asked for using one-dimensional Dirac function, is initial profile;
S5:By the computing of ROI image binary map and treated image sequence does dot product, operation result is figure to be split
Picture.
Preferred embodiment is planted again in the present invention, the specific implementation procedure of the online segmentation step includes:
S10:By image to be split and the input quantity that initial profile is RSLGD model energy functions, initiation parameter;
S20:Calculate active contour interior zone and the average gray and variance of perimeter;
S30:Calculated level set function;
S40:Judge whether the iterative numerical computing of gradient flow equation restrains, if so, stopping iteration, after output segmentation
Image ordered series of numbers and level set function;If it is not, making p=p+1, S20 is transferred to, p is iterations, continues iterative cycles;
S50:Image sequence after segmentation is added in original sequence.
In conclusion by adopting the above-described technical solution, the beneficial effects of the invention are as follows:
On the basis of local Gaussian distributed model, RSLGD models are established, a kind of edge is introduced in the energy function of the model
Stop function and carry out constraint length item, and pass through and add LoG image energy functions, so as to quickly capture the target side in multiple ROI
Edge is simultaneously split, and by segmentation the result is shown in original video or sequence, beneficial to ultrasonic doctor subjective analysis.It is whole by observing
The segmentation result of a sequence can aid in doctor to obtain focal area and same level reference zone, such as real in liver, kidney abdomen
The time-activity curve of lump in the superficial organs such as matter organ and mammary gland, thyroid gland, so as to analyze to obtain quantitative index, is such as opened
Beginning Enhanced time, enhancing duration, peak strength, clean up time and area under a curve etc. at peak time.
Description of the drawings
Fig. 1 is the processing procedure schematic diagram of the embodiment of the invention;
Fig. 2 is the RSLGD algorithm flow charts of the embodiment of the invention.
Specific embodiment
The embodiment of the present invention is described below in detail, the example of the embodiment is shown in the drawings, wherein from beginning to end
Same or similar label represents same or similar element or has the function of same or like element.Below with reference to attached
The embodiment of figure description is exemplary, and is only used for explaining the present invention, and is not considered as limiting the invention.
In the description of the present invention, it is to be understood that term " longitudinal direction ", " transverse direction ", " on ", " under ", "front", "rear",
The orientation or position relationship of the instructions such as "left", "right", " vertical ", " level ", " top ", " bottom " " interior ", " outer " is based on attached drawing institutes
The orientation or position relationship shown is for only for ease of the description present invention and simplifies description rather than instruction or imply signified dress
It puts or element must have specific orientation, with specific azimuth configuration and operation, therefore it is not intended that limit of the invention
System.
In the description of the present invention, unless otherwise prescribed with limit, it is necessary to explanation, term " installation ", " connected ",
" connection " should be interpreted broadly, for example, it may be the connection inside mechanical connection or electrical connection or two elements, it can
To be to be connected directly, can also be indirectly connected by intermediary, it for the ordinary skill in the art, can basis
Concrete condition understands the concrete meaning of above-mentioned term.
The invention discloses a kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model, Fig. 1 is the present invention
The processing procedure schematic diagram of one specific embodiment, including pre-treatment step and online segmentation step,
Pre-treatment step is based on ultrasonic contrast image, obtains image and initial profile to be split;
Online segmentation step based on RSLGD models and obtains energy letter using image to be split and initial profile as input quantity
Number introduces Edge-stopping function with constraint length item in energy function, and increases LoG image energy functions, completes to single
Or the object edge of multiple and different ROI is captured and split.
RSLGD models are to be distributed (Region-Scalable and Local based on the telescopic local Gaussian in region
Gaussian-Distribution, RSLGD) parted pattern.
In the preferred embodiment of the present invention, the specific implementation procedure of pre-treatment step includes:
S1:Ultrasonic contrast image is converted into original sequence;
Assuming that the data after targeted ultrasound contrast agent imaging are video or Dicom sequences, A={ A are denoted as1,A2,…,An,
Wherein n represents the number of video frame number or Dicom sequences, and to be more than or equal to 1 positive integer, video or Dicom sequences are turned
For original sequence B={ B1,B2,…,Bn}。
S2:Down-sampling processing is carried out using quadratic interpolattion to original sequence;
And/or image noise reduction processing is carried out using Gassian low-pass filter;
And/or picture smooth treatment is carried out using guiding filtering;
Obtain treated image sequence;
Down-sampling f is carried out to original sequence B using quadratic interpolattion and/or using Gassian low-pass filter to f (B)
Carry out image noise reduction g and/or using guiding filtering to g (f (B)) progress image smoothing h, the image sequence C=that obtains that treated
{C1,C2,…,Cn, i.e. Ck=h (g (f (Bk))), wherein, f, g, h is mutual indepedent, k=1,2 ..., n, k for video frame or
The number sequence number of Dicom sequences.
S3:One or more different ROI regions are chosen in image sequence after treatment, ROI image is obtained, utilizes Hai Shi
Function asks for ROI image binary map;
Assuming that image C after treatmentkM ROI region of middle selection, is denoted as R={ R respectively1,R2,…,Rm, andJ is the number of single ROI region, i.e., is separated from each other between each ROI region, non-intersect, is obtained
ROI image Dk。
It utilizes Hai Shi functions (Heaviside Function, HF)D is obtainedkCorresponding binary map
Bwk=H (Dk)。
S4:The profile of ROI image binary map is asked for using one-dimensional Dirac function, is initial profile;
It utilizes one-dimensional Dirac function (Dirac Function, DF)Bw is obtainedkProfile Ck 0=δ
(Bwk), and using the profile as the initial profile of RSLGD algorithms.
S5:By the computing of ROI image binary map and treated image sequence does dot product, operation result is figure to be split
Picture.
To accelerate algorithm iteration efficiency, image to be split is sought using binary map, image to be split for ROI image binary map with
Treated, and image sequence is the operation result of dot product, expression formula Ik=Ck.*Bwk。
Initial data is pre-processed, such as down-sampling, image noise reduction, image smoothing and image enhancement operation, after being beneficial to
Continuous cutting operation.
In the preferred embodiment of the present invention, the process for establishing RSLGD models is:
Assuming that image I to be splitk:Ω → R, i.e., image I to be splitkThe gray value of point in middle region Ω is real number, and k is
The number sequence number of video frame or Dicom sequences, the meaning of R are that the gray value of pixel is real number;Ω is in image I to be split
Regional area, two non-conterminous region Ω are classified as by zero level set function1And Ω2, x ∈ Ω;For in the Ω of region
Each point x Gaussian distributed.Make φ0For zero level set function, then φ0={ x:φ (x)=0 } region Ω is divided to for two
A non-conterminous region:Ω1={ x:φ (x) > 0 } and Ω2={ x:φ (x) < 0 }, I is represented respectivelykProspect and background area
Domain, therefore, image IiOptimum segmentation can be solved by maximum Likelihood.(Local is distributed in local Gaussian
Gaussian Distribution, LGD) model on the basis of, introduce Edge-stopping function and carry out constraint length item, and pass through and add
Add LoG image energy functions, so as to quickly capture object edge and the segmentation in multiple ROI.Therefore the energy letter of RSLGD models
Number is defined as:
Wherein, φ is level set function;u1Simplify description for the gray average function in active contour interior zone;u2
Simplify description for the gray average function in active contour perimeter;For the gray variance in active contour interior zone
Function simplifies description;Simplify description for the gray variance function in active contour perimeter.
Initial profile is the contour curve of an initialization, and the energy of initial profile is not necessarily minimum, it is necessary to pass through
Evolution obtains object edge.The curve of intermediate iteration process generation, initial profile and final objective contour can be known as
Active contour.RSLGD models are the model basis that (Local Gaussian Distribution, LGD) is distributed in local Gaussian
On, it introduces a kind of Edge-stopping function and carrys out constraint length item, and pass through and add LoG image energy functions, it is more so as to quickly capture
Object edge and segmentation in a ROI.
LoG be Laplacian of Gaussian, Gauss-Laplace.
In the preferred embodiment of the present invention, Edge-stopping function is defined as:
Wherein,It is for standard deviationGaussian kernel functionWith image I's to be split
Convolution algorithm formula, takesX is the point in image to be split, and y is the point in the field of point x;For gradient operator.
In the preferred embodiment of the present invention, ElgdEnergy term is defined as:
For each x ∈ Ω, O is madex={ y:| | x-y | |≤r be x neighborhood, r is the radius of neighborhood.From the angle of statistics
Degree sees that the regional area fitting energy of active contour can be described with Gaussian Profile, and be derived using Level Set Theory,
ElgdEnergy term obtains above-mentioned definition.
Wherein, u1(x) and u2(x) be respectively active contour interior zone Ω1With perimeter Ω2Gray average;The respectively interior zone Ω of active contour1With perimeter Ω2Gray variance;Ω is image to be detected
Regional area in I is classified as two non-conterminous region Ω by zero level set function1And Ω2, x ∈ Ω, y are the neighborhood of x
Interior point;H (φ (y)) is the handling result using Hai Shi function pair level set function φ (y), for binaryzation;Kσ(x-y) it is
Gaussian kernel function is defined as
u1(x) it is:
u2(x) it is:
For:
For:
And have:
M1(φ (y))=H (φ (y)), M2(φ (y))=1-H (φ (y));
WithRespectively Ω1And Ω2The probability density function in region, point
It is not defined as:
Wherein, I (y) is the gray value of image I midpoint y.
In the preferred embodiment of the present invention, ERS(φ) energy term is defined as:
ERS(φ)=μ P (φ)+vL*(φ)+ωMΔ(φ);
Above formula can promote the marginal information of image and more accurately detect edge.
Wherein, to ensure the systematicness of level set function, regularization term is introducedIt can
Stablize evolutionary process.Meanwhile be smooth active contour, introduce a kind of Edge-stopping functionTo change in LGD models
Length itemI.e. new length item is:
In formulaIt is that a standard deviation is's
Gaussian kernel functionWith the convolution algorithm of image I, can use
MΔ(φ) is the LoG item of image energy functions after optimization;LoG items after optimization are defined as MΔ(φ)=∫ΩH
(φ) M (x) dx, M (x) therein are gray value of the LoG images at point x, can be iterated and are calculated by following formula:
Wherein, β is a positive number, can 0 < β≤1 of value;M is LoG images;Δ(Gσ* I) it is famous second order partial differential
Edge detection operator, mathematic(al) representation are:
For to be handled first with extra large formula function pair level set function φ (x), then seek the gradient of H (φ (x))
Image;μ, v and ω are every weight coefficient, and the value range of μ and ω are that 0 to 1, v values are 2552× T, wherein T values are
0.0001 to 0.001;φ (x) is the level set function of x;Δ(Gσ* I) it is second order partial differential edge detection operator.
In the preferred embodiment of the present invention, ladder is obtained first to calculate to the numerical solution process of energy function
Flow equation is spent, then iterative numerical computing is carried out to gradient flow equation;
Energy function ERSLGDSeek image IiOptimum segmentation problem, can be translated into solve energy functional minimum ask
Topic.First using the method for value solving of display Euler, then the Euler-Lagrange equations of energy functional are obtained by the calculus of variations,
Gradient flow equation is calculated.
Gradient flow equation is:
Wherein, λ1,λ2, μ, v and ω are every weight coefficient;λ1And λ2It is determined according to actual scene;It is bent
Rate, Δ are Laplace operator, and δ (φ) is the one-dimensional Dirac function of level set function;ei(x) it is defined as:
I is 1 or 2, e1And e2Respectively e1(x) and e2(x) write a Chinese character in simplified form;
For the gradient current partial differential equation of energy functional, iterative numerical computing, numerical value are carried out using finite difference calculus
The condition of convergence of interative computation is:
Or iterations iter≤P;
Wherein, P is greatest iteration step number, and Γ is pixel number threshold value,For+1 active contour of pth
Pixel number,For the pixel number of pth time active contour, iter is iterations sequence number;Represent the pixel number of+1 active contour of pth and the picture of pth time active contour
The absolute value of the difference of vegetarian refreshments number is less than pixel number threshold value Γ.
In the preferred embodiment of the present invention, the specific implementation procedure of online segmentation step includes:
S10:Using image to be split and initial profile as the input quantity of RSLGD model energy functions, initiation parameter;Just
Beginningization parameter includes:λ1=1, λ2=1, ω=1, μ=0.1, v=0.0005 × 2552, P=200;
S20:Calculate active contour interior zone and the average gray and variance of perimeter;
S30:Calculated level set function φ (x);
S40:Judge whether the iterative numerical computing of gradient flow equation restrains, if so, stopping iteration, after output segmentation
Image ordered series of numbers and level set function;If it is not, making p=p+1, p is iterations, is transferred to S20, continues iterative cycles;
S50:Image sequence after segmentation is added in original sequence.
Using RSLGD algorithms, ROI in ultrasonic contrast image is split, by observing the segmentation result of entire sequence,
Doctor can be aided in obtain focal area and same level reference zone, such as in liver, kidney abdomen organa parenchymatosum and mammary gland, first shape
The time-activity curve of lump in the superficial organs such as gland, so as to analyze to obtain quantitative index (including starting Enhanced time, enhancing
Duration, peak time, clean up time, area under a curve etc. at peak strength).
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or the spy for combining the embodiment or example description
Point is contained at least one embodiment of the present invention or example.In the present specification, schematic expression of the above terms is not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiments or example in combine in an appropriate manner.
Although an embodiment of the present invention has been shown and described, it will be understood by those skilled in the art that:Not
In the case of departing from the principle of the present invention and objective a variety of change, modification, replacement and modification can be carried out to these embodiments, this
The scope of invention is limited by claim and its equivalent.
Claims (10)
1. a kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model, which is characterized in that including pre-treatment step
With online segmentation step,
The pre-treatment step is based on ultrasonic contrast image, obtains image and initial profile to be split;
The online segmentation step obtains energy function, in energy function using image to be split and initial profile as input quantity
Edge-stopping function is introduced with constraint length item, and increases LoG image energy functions;Numerical solution is carried out to energy function, it is complete
The object edge of paired single or multiple difference ROI is captured and split.
2. the ultrasonic contrast image partition method as described in claim 1 based on statistics Partial Differential Equation Model, which is characterized in that institute
Stating energy function is:
<mrow>
<msup>
<mi>E</mi>
<mrow>
<mi>R</mi>
<mi>S</mi>
<mi>L</mi>
<mi>G</mi>
<mi>D</mi>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>E</mi>
<mrow>
<mi>l</mi>
<mi>g</mi>
<mi>d</mi>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msup>
<mi>E</mi>
<mrow>
<mi>R</mi>
<mi>S</mi>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, φ is level set function;u1Simplify description for the gray average function in active contour interior zone;u2For work
Simplifying for gray average function in driving wheel exterior feature perimeter describes;For the gray variance function in active contour interior zone
Simplify description;Simplify description for the gray variance function in active contour perimeter.
3. the ultrasonic contrast image partition method as claimed in claim 2 based on statistics Partial Differential Equation Model, which is characterized in that institute
Stating Edge-stopping function is:
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>I</mi>
<mo>|</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>min</mi>
<mo>{</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<mo>|</mo>
<mo>&dtri;</mo>
<msub>
<mi>G</mi>
<mover>
<mi>&sigma;</mi>
<mo>~</mo>
</mover>
</msub>
<mo>*</mo>
<mi>I</mi>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>,</mo>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<msub>
<mi>G</mi>
<mover>
<mi>&sigma;</mi>
<mo>~</mo>
</mover>
</msub>
<mo>*</mo>
<mi>I</mi>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
</mrow>
<msup>
<mover>
<mi>&sigma;</mi>
<mo>~</mo>
</mover>
<mn>2</mn>
</msup>
</mfrac>
<mo>)</mo>
</mrow>
<mo>}</mo>
<mo>;</mo>
</mrow>
Wherein,It is for standard deviationGaussian kernel functionIt is transported with the convolution of image I to be split
Formula takesX is the point in image to be split, and y is the point in the field of point x;▽ is gradient operator.
4. the ultrasonic contrast image partition method as claimed in claim 3 based on statistics Partial Differential Equation Model, which is characterized in that institute
State ElgdEnergy term is:
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msup>
<mi>E</mi>
<mrow>
<mi>l</mi>
<mi>g</mi>
<mi>d</mi>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<mo>-</mo>
<mo>&Integral;</mo>
<mo>&lsqb;</mo>
<msub>
<mo>&Integral;</mo>
<msub>
<mi>&Omega;</mi>
<mn>1</mn>
</msub>
</msub>
<msub>
<mi>K</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>-</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mi>log</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mi>x</mi>
</mrow>
</msub>
<mo>(</mo>
<mrow>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>&rsqb;</mo>
<mo>&CenterDot;</mo>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>x</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mo>&Integral;</mo>
<mo>&lsqb;</mo>
<msub>
<mo>&Integral;</mo>
<msub>
<mi>&Omega;</mi>
<mn>2</mn>
</msub>
</msub>
<msub>
<mi>K</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>-</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mi>log</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mrow>
<mn>2</mn>
<mo>,</mo>
<mi>x</mi>
</mrow>
</msub>
<mo>(</mo>
<mrow>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>&rsqb;</mo>
<mo>&CenterDot;</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>H</mi>
<mo>(</mo>
<mrow>
<mi>&phi;</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>x</mi>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>;</mo>
</mrow>
Wherein, u1(x) and u2(x) be respectively active contour interior zone Ω1With perimeter Ω2Gray average;WithThe respectively interior zone Ω of active contour1With perimeter Ω2Gray variance;Y be x neighborhood in point, neighborhood
Radius is r;H (φ (y)) be utilize Hai Shi function pair level set function φ (y) handling result, Kσ(x-y) it is Gaussian kernel letter
Number, is defined as
WithRespectively Ω1And Ω2The probability density function in region, respectively
For:
<mrow>
<msub>
<mi>p</mi>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mi>x</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>I</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<msqrt>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
</mrow>
</msqrt>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&CenterDot;</mo>
<mi>exp</mi>
<mo>&lsqb;</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>u</mi>
<mn>1</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>-</mo>
<mi>I</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&sigma;</mi>
<mn>1</mn>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
<mo>;</mo>
</mrow>
<mrow>
<msub>
<mi>p</mi>
<mrow>
<mn>2</mn>
<mo>,</mo>
<mi>x</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>I</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>,</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<msqrt>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
</mrow>
</msqrt>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&CenterDot;</mo>
<mi>exp</mi>
<mo>&lsqb;</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>u</mi>
<mn>2</mn>
</msub>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>-</mo>
<mi>I</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&sigma;</mi>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
<mo>;</mo>
</mrow>
Wherein, I (y) is the gray value of image I midpoint y.
5. the ultrasonic contrast image partition method as claimed in claim 4 based on statistics Partial Differential Equation Model, which is characterized in that institute
State ERS(φ) energy term is:
ERS(φ)=μ P (φ)+vL*(φ)+ωMΔ(φ);
Wherein, the P (φ) is regularization term;MΔ(φ) is the LoG item of image energy functions after optimization;L*(φ) is to add side
Edge stops the new length item of function, is defined as:
L*(φ)=∫ g (| ▽ I |) | ▽ H (φ (x)) | dx, ▽ H (φ (x)) are first with extra large formula function pair level set function φ
(x) handled, then seek the gradient image of H (φ (x));μ, v and ω are weight coefficient.
6. the ultrasonic contrast image partition method as claimed in claim 5 based on statistics Partial Differential Equation Model, which is characterized in that institute
Stating regularization term P (φ) is:
<mrow>
<mi>P</mi>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>&Integral;</mo>
<msup>
<mrow>
<mo>(</mo>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>&phi;</mi>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>|</mo>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mi>d</mi>
<mi>x</mi>
<mo>;</mo>
</mrow>
Wherein, φ (x) is the level set function of x.
7. the ultrasonic contrast image partition method as claimed in claim 5 based on statistics Partial Differential Equation Model, which is characterized in that institute
State LoG MΔ(φ) is defined as:
MΔ(φ)=∫ΩH(φ)·M(x)dx;
Wherein, M (x) is gray value of the LoG images at point x, is defined as:
<mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>M</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>t</mi>
</mrow>
</mfrac>
<mo>=</mo>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>I</mi>
<mo>|</mo>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mi>M</mi>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>g</mi>
<mo>(</mo>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>I</mi>
<mo>|</mo>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<mi>M</mi>
<mo>-</mo>
<mi>&beta;</mi>
<mo>&times;</mo>
<mi>&Delta;</mi>
<mo>(</mo>
<mrow>
<msub>
<mi>G</mi>
<mi>&sigma;</mi>
</msub>
<mo>*</mo>
<mi>I</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
In above formula, β is positive number, and M is LoG images;Δ(Gσ* I) it is second order partial differential edge detection operator.
8. the ultrasonic contrast image partition method as claimed in claim 5 based on statistics Partial Differential Equation Model, which is characterized in that institute
It states the numerical solution process of energy function and obtains gradient flow equation first to calculate, then iterative numerical fortune is carried out to gradient flow equation
It calculates;
Gradient flow equation is:
<mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>&phi;</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>t</mi>
</mrow>
</mfrac>
<mo>=</mo>
<mo>-</mo>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>)</mo>
</mrow>
<mo>&CenterDot;</mo>
<mrow>
<mo>(</mo>
<msub>
<mi>&lambda;</mi>
<mn>1</mn>
</msub>
<msub>
<mi>e</mi>
<mn>1</mn>
</msub>
<mo>-</mo>
<msub>
<mi>&lambda;</mi>
<mn>2</mn>
</msub>
<msub>
<mi>e</mi>
<mn>2</mn>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&mu;</mi>
<mo>&lsqb;</mo>
<mi>&Delta;</mi>
<mi>&phi;</mi>
<mo>-</mo>
<mi>d</mi>
<mi>i</mi>
<mi>v</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<mo>&dtri;</mo>
<mi>&phi;</mi>
</mrow>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>&phi;</mi>
<mo>|</mo>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>+</mo>
<mi>v</mi>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>)</mo>
</mrow>
<mo>&CenterDot;</mo>
<mi>d</mi>
<mi>i</mi>
<mi>v</mi>
<mrow>
<mo>(</mo>
<mi>g</mi>
<mo>(</mo>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>I</mi>
<mo>|</mo>
</mrow>
<mo>)</mo>
<mo>&CenterDot;</mo>
<mfrac>
<mrow>
<mo>&dtri;</mo>
<mi>&phi;</mi>
</mrow>
<mrow>
<mo>|</mo>
<mo>&dtri;</mo>
<mi>&phi;</mi>
<mo>|</mo>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&omega;</mi>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<mi>&phi;</mi>
<mo>)</mo>
</mrow>
<mo>&CenterDot;</mo>
<mi>M</mi>
<mo>;</mo>
</mrow>
Wherein, λ1,λ2, μ, v and ω are every weight coefficient;Div (▽ φ/| ▽ φ |) is curvature, and Δ is Laplace operator, δ
(φ) is the one-dimensional Dirac function of level set function;ei(x) it is defined as:
<mrow>
<msub>
<mi>e</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mo>&Integral;</mo>
<msub>
<mi>&Omega;</mi>
<mi>i</mi>
</msub>
</msub>
<msub>
<mi>K</mi>
<mi>&sigma;</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>-</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>&lsqb;</mo>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&sigma;</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>-</mo>
<mi>I</mi>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&sigma;</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
<mi>d</mi>
<mi>y</mi>
<mo>;</mo>
</mrow>
I is 1 or 2, e1And e2Respectively e1(x) and e2(x) write a Chinese character in simplified form;
The condition of convergence of iterative numerical computing is:
Or iterations iter≤P;
Wherein, P is greatest iteration step number, and iter is iterations sequence number, and Γ is pixel number threshold value;For
The pixel number of p+1 active contour,For pth time active contour.
9. the ultrasonic contrast image partition method as described in claim 1 based on statistics Partial Differential Equation Model, which is characterized in that institute
The detailed process for stating pre-treatment step is:
S1:Ultrasonic contrast image is converted into original sequence;
S2:Down-sampling processing is carried out using quadratic interpolattion to original sequence;
And/or image noise reduction processing is carried out using Gassian low-pass filter;
And/or picture smooth treatment is carried out using guiding filtering;
Obtain treated image sequence;
S3:One or more different ROI regions are chosen in image sequence after treatment, ROI image is obtained, utilizes Hai Shi functions
Ask for ROI image binary map;
S4:The profile of ROI image binary map is asked for using one-dimensional Dirac function, is initial profile;
S5:By the computing of ROI image binary map and treated image sequence does dot product, operation result is image to be split.
10. the ultrasonic contrast image partition method as described in claim 1 based on statistics Partial Differential Equation Model, which is characterized in that
The specific implementation procedure of the online segmentation step includes:
S10:Using image to be split and initial profile as the input quantity of RSLGD model energy functions, initiation parameter;
S20:Calculate active contour interior zone and the average gray and variance of perimeter;
S30:Calculated level set function;
S40:Judge whether the iterative numerical computing of gradient flow equation restrains, if so, stop iteration, the image after output segmentation
Ordered series of numbers and level set function;If it is not, making p=p+1, p is iterations, is transferred to S20, continues iterative cycles;
S50:Image sequence after segmentation is added in original sequence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711347973.2A CN108090909B (en) | 2017-12-15 | 2017-12-15 | Ultrasonic contrast image segmentation method based on statistical partial differential model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711347973.2A CN108090909B (en) | 2017-12-15 | 2017-12-15 | Ultrasonic contrast image segmentation method based on statistical partial differential model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108090909A true CN108090909A (en) | 2018-05-29 |
CN108090909B CN108090909B (en) | 2020-09-08 |
Family
ID=62176688
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711347973.2A Active CN108090909B (en) | 2017-12-15 | 2017-12-15 | Ultrasonic contrast image segmentation method based on statistical partial differential model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108090909B (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108986114A (en) * | 2018-07-11 | 2018-12-11 | 中南大学 | A kind of abdominal CT sequence image liver automatic division method based on level set and shape description symbols |
CN110211140A (en) * | 2019-06-14 | 2019-09-06 | 重庆大学 | Abdominal vascular dividing method based on 3D residual error U-Net and Weighted Loss Function |
CN110613486A (en) * | 2019-09-30 | 2019-12-27 | 深圳大学总医院 | Method and device for detecting breast ultrasound image |
CN110992370A (en) * | 2019-11-19 | 2020-04-10 | 东软医疗系统股份有限公司 | Pancreas tissue segmentation method and device and terminal equipment |
CN111127479A (en) * | 2019-12-17 | 2020-05-08 | 昆明理工大学 | Level set image segmentation method based on curve area |
CN111462145A (en) * | 2020-04-01 | 2020-07-28 | 重庆大学 | Active contour image segmentation method based on double-weight symbol pressure function |
CN111539926A (en) * | 2020-04-20 | 2020-08-14 | 京东方科技集团股份有限公司 | Image detection method and device |
CN111583326A (en) * | 2020-04-07 | 2020-08-25 | 青岛大学附属医院 | Method and system for measuring human kidney parenchyma area |
CN113012041A (en) * | 2021-04-06 | 2021-06-22 | 重庆大学 | New method and system for CT/MRI ultrasound simulation |
CN113177958A (en) * | 2021-06-29 | 2021-07-27 | 之江实验室 | Ultrasonic microscopic image processing method for specific depth information |
CN116645576A (en) * | 2023-07-26 | 2023-08-25 | 深圳华声医疗技术股份有限公司 | Ultrasonic image tag processing method and device, ultrasonic equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105321184A (en) * | 2015-11-26 | 2016-02-10 | 北京交通大学 | Improved edge level set-based method and system for segmenting noisy image |
CN106056576A (en) * | 2016-03-09 | 2016-10-26 | 华侨大学 | Method for segmenting aorta in CT image combining edge and area characteristics |
CN107274414A (en) * | 2017-05-27 | 2017-10-20 | 西安电子科技大学 | Image partition method based on the CV models for improving local message |
-
2017
- 2017-12-15 CN CN201711347973.2A patent/CN108090909B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105321184A (en) * | 2015-11-26 | 2016-02-10 | 北京交通大学 | Improved edge level set-based method and system for segmenting noisy image |
CN106056576A (en) * | 2016-03-09 | 2016-10-26 | 华侨大学 | Method for segmenting aorta in CT image combining edge and area characteristics |
CN107274414A (en) * | 2017-05-27 | 2017-10-20 | 西安电子科技大学 | Image partition method based on the CV models for improving local message |
Non-Patent Citations (1)
Title |
---|
KEYAN DINGA等: "Active contours driven by region-scalable fitting and optimized Laplacian of Gaussian energy for image segmentation", 《SIGNAL PROCESSING》 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108986114B (en) * | 2018-07-11 | 2022-03-29 | 中南大学 | Automatic abdominal CT sequence image liver segmentation method based on level set and shape descriptor |
CN108986114A (en) * | 2018-07-11 | 2018-12-11 | 中南大学 | A kind of abdominal CT sequence image liver automatic division method based on level set and shape description symbols |
CN110211140A (en) * | 2019-06-14 | 2019-09-06 | 重庆大学 | Abdominal vascular dividing method based on 3D residual error U-Net and Weighted Loss Function |
CN110613486A (en) * | 2019-09-30 | 2019-12-27 | 深圳大学总医院 | Method and device for detecting breast ultrasound image |
CN110613486B (en) * | 2019-09-30 | 2022-04-22 | 深圳大学总医院 | Method and device for detecting breast ultrasound image |
CN110992370A (en) * | 2019-11-19 | 2020-04-10 | 东软医疗系统股份有限公司 | Pancreas tissue segmentation method and device and terminal equipment |
CN110992370B (en) * | 2019-11-19 | 2023-07-04 | 东软医疗系统股份有限公司 | Pancreas tissue segmentation method and device and terminal equipment |
CN111127479A (en) * | 2019-12-17 | 2020-05-08 | 昆明理工大学 | Level set image segmentation method based on curve area |
CN111462145B (en) * | 2020-04-01 | 2023-04-07 | 重庆大学 | Active contour image segmentation method based on double-weight symbol pressure function |
CN111462145A (en) * | 2020-04-01 | 2020-07-28 | 重庆大学 | Active contour image segmentation method based on double-weight symbol pressure function |
CN111583326A (en) * | 2020-04-07 | 2020-08-25 | 青岛大学附属医院 | Method and system for measuring human kidney parenchyma area |
CN111539926A (en) * | 2020-04-20 | 2020-08-14 | 京东方科技集团股份有限公司 | Image detection method and device |
CN111539926B (en) * | 2020-04-20 | 2024-04-26 | 京东方科技集团股份有限公司 | Image detection method and device |
CN113012041A (en) * | 2021-04-06 | 2021-06-22 | 重庆大学 | New method and system for CT/MRI ultrasound simulation |
CN113012041B (en) * | 2021-04-06 | 2023-12-12 | 重庆大学 | Novel CT/MRI (computed tomography/magnetic resonance imaging) ultrasound simulation method and system |
CN113177958A (en) * | 2021-06-29 | 2021-07-27 | 之江实验室 | Ultrasonic microscopic image processing method for specific depth information |
CN116645576A (en) * | 2023-07-26 | 2023-08-25 | 深圳华声医疗技术股份有限公司 | Ultrasonic image tag processing method and device, ultrasonic equipment and storage medium |
CN116645576B (en) * | 2023-07-26 | 2023-09-19 | 深圳华声医疗技术股份有限公司 | Ultrasonic image tag processing method and device, ultrasonic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108090909B (en) | 2020-09-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108090909A (en) | A kind of ultrasonic contrast image partition method based on statistics Partial Differential Equation Model | |
Chakravarty et al. | RACE-net: a recurrent neural network for biomedical image segmentation | |
Barbosa et al. | B-spline explicit active surfaces: an efficient framework for real-time 3-D region-based segmentation | |
Beichel et al. | Liver segmentation in contrast enhanced CT data using graph cuts and interactive 3D segmentation refinement methods | |
US9990743B2 (en) | Suppression of vascular structures in images | |
US7773806B2 (en) | Efficient kernel density estimation of shape and intensity priors for level set segmentation | |
US8929636B2 (en) | Method and system for image segmentation | |
Athavale et al. | Multiscale properties of weighted total variation flow with applications to denoising and registration | |
Wang et al. | Fast and accurate smoothing method using a modified Allen–Cahn equation | |
CN104867151B (en) | A kind of eye fundus image blood vessel segmentation method based on ductility active contour model | |
Jia et al. | A novel dynamic multilevel technique for image registration | |
Manniesing et al. | Local speed functions in level set based vessel segmentation | |
CN107492105A (en) | A kind of variation dividing method based on more statistical informations | |
Schwarzkopf et al. | Volumetric nonlinear anisotropic diffusion on GPUs | |
Pan et al. | Current-and varifold-based registration of lung vessel and airway trees | |
Qian et al. | Robust segmentation of 4D cardiac MRI-tagged images via spatio-temporal propagation | |
Alom et al. | Automatic slice growing method based 3D reconstruction of liver with its vessels | |
Romdhane et al. | 3D medical images denoising | |
Zhu et al. | Non-rigid multi-modal brain image registration based on two-stage generative adversarial nets | |
Ciurte et al. | An efficient segmentation method for ultrasound images based on a semi-supervised approach and patch-based features | |
Fritz | Diffusion-based applications for interactive medical image segmentation | |
Haque et al. | Automated registration of 3d ultrasound and ct/mr images for liver | |
Vera et al. | Smart operator for the human liver automatic segmentation, present in medical images | |
Chen et al. | A Force Feedback Pedical Screw Impantation Surgical Trainer with Automatic Vertebra Segmentation | |
Sarti et al. | Evolutionary partial differential equations for biomedical image processing |
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 |