CN110322491A - A kind of algorithm of deformable mouse systemic map and mouse Image registration - Google Patents
A kind of algorithm of deformable mouse systemic map and mouse Image registration Download PDFInfo
- Publication number
- CN110322491A CN110322491A CN201910501435.7A CN201910501435A CN110322491A CN 110322491 A CN110322491 A CN 110322491A CN 201910501435 A CN201910501435 A CN 201910501435A CN 110322491 A CN110322491 A CN 110322491A
- Authority
- CN
- China
- Prior art keywords
- map
- skin
- bone
- registration
- image
- 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
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses the algorithms of a kind of deformable mouse systemic map and mouse Image registration, are mainly registrated target image using map flexible characteristics, the organic region in map is mapped in target individual, realize the division to target organ region.Posture, height and the weight and target of preparatory adjust automatically map skin and bone are close in registration.Remaining organ for meeting statistical variations rule can estimate corresponding location and shape according to map skin and bone use condition Gauss model；Remaining organ uses thin plate spline deformation map according to peripheral organs.Whole body map after obtaining experiment material extracts skin, bone and lung, with the filling of true gray value, will fill organ and target registration using three-dimensional image registration method, guidance map completes final deformation, obtains final result.The present invention solves the problems, such as body posture variation and organ individual morphology difference in mouse Image registration simultaneously, and has higher registration accuracy and robustness.
Description
Technical field
The invention belongs to toy medical image studying technological domain, be related to a kind of deformable mouse systemic map with it is small
The algorithm of mouse Image registration is focused on using the deformable feature of mouse systemic map and mouse Image registration, will be in map
Organic region be mapped in target image, and then from target image segmentation obtain mouse major organs.
Background technique
With the high speed development of Medical Imaging Technology, the research of toy image is manufactured experimently in preclinical cancer research and new drug
Middle important role.The progress of toy image to popularize for relevant image analysing computer work put forward new requirements, unit
The a large amount of image datas of time domestic demand processing, and the disadvantages of manually to handle that there are subjectivities strong, repeatability difference.Therefore it is badly in need of
The toy image analysis techniques automate, to objectify improve the precision and efficiency of toy image analysing computer.In the shadow of automation
As in analysis, Digital anatomy map plays an important role, by being registrated for map and target individual, provided for target individual
Anatomy reference provides anatomy positioning for the expression of lesion and gene information.In the preclinical study stage, mouse is most common
Experimental subjects.
Mouse number anatomical atlas includes three embryo's map, brain map and whole body map primary categories, and the present invention relates to
And mouse systemic map.Mouse systemic map mainly has four aspect applications in image analysing computer: being realized by registration and is schemed to individual
As the division and measurement of organic region；For positron emission Computed tomography (Positron Emission
Computed Tomography, PET), single photon emission computerized tomography,SPECT be imaged (Single Photon Emission
Computed Tomography, SPECT) and fluorescent molecular tomography (Fluorescence Molecular
Tomography, FMT) etc. function assessments image provide anatomy reference；Multiple image collection for same individual provides anatomy
Coordinate interrelation, multiple image collection include the different time points acquisition of the different images type collection and same individual of same individual
Two class situations；With the accumulation of mouse systemic gene information, anatomy positioning can be provided for gene expression information.
Although mouse systemic map there are many applications in image analysing computer, the Study of Registration of mouse systemic map is used
In, the morphological differences between map and target individual limits the precision, robustness and the degree of automation of atlas registration.Poor morphology
There are two types of different performances: the profile variation as caused by body gesture；The organ morphology as caused by the factors such as weight, age, population
Difference is reflected as the dimension scale and relative displacement and fat thickness etc. of organ.
To realize being registrated for mouse systemic map and target individual, a variety of registration Algorithms are proposed both at home and abroad, are broadly divided into
Nonlinear deformation registration, turning joint registration and statistical shape model registration.Nonlinear deformation registration passes through definition calibration earliest
Point be registrated with target individual, and subsequent appearance and body surface curved surface and 3-D image are registrated, but nonlinear mode of texturing can not retouch
Mouse systemic attitudes vibration is stated, organ distortion distortion is easy to appear.To solve body posture variation issue, propose living using mouse
Dynamic bone map is registrated, and is adjusted map posture by the rotation of skeletal joint, is driven internal device by nonlinear mode
Official's deformation, this method solve the registration problems of body gesture to a certain extent, but only realize the posture transformation to bone,
Remaining organ still uses nonlinear deformation mode, without fundamentally solving the problems, such as that internal distorts.In order to solve inside
Soft tissue organs distortion problem, Hongkai Wang et al. propose using statistical shape model to different weight, the age, gender,
The mouse systemic major organs of population are modeled, and mouse systemic statistical shape model is constructed, and pass through statistical shape model training
Resulting deformation rule registration, it is ensured that deformation rule of the mode of texturing of internal between training sample, basic topology knot
Structure is constant, and the metamorphosis of internal can accurately be more described than nonlinear deformation mode, is unlikely to distortion distortion occur to ask
Topic.But existing three classes method can not solve the problems, such as mouse body attitudes vibration and organ individual morphology difference simultaneously, be
This Hongkai Wang et al. constructs the mouse systemic map that attitudes vibration and organ morphology variation can be achieved at the same time, and map can
By changing the deformation of the state modulators maps such as bone posture, height and weight.The present invention is the base in the deformable mouse map
On plinth, a kind of and mouse Image registration algorithm is proposed.
Summary of the invention
The purpose of the present invention is to provide the algorithms of a kind of deformable mouse map and mouse Image registration, mainly solve
Technical problem is to utilize mouse systemic map flexible characteristics, while solving the variation of mouse body posture and organ in registration process
The problem of individual morphology difference, so that atlas registration has higher precision and robustness.The image that the present invention is applicable in includes various
Tomography medical image mode, such as CT image, nuclear magnetic resonance image and nuclear medical image.
Technical solution of the present invention:
A kind of algorithm of deformable mouse systemic map and mouse Image registration, steps are as follows:
The first step chooses dissection calibration point from mouse image.
Dissection calibration point need to choose the artis of bone and the central point of internal organs, choose calibration point quantity and do not want
It asks, but is to be able to determine mouse body posture, should respectively contain at least one dissection calibration point in mouse four limbs, remaining calibration
Point can be determined whether to add according to practical application effect by user.The selection mode of calibration point can be by manually choosing, also
It can be detected by automatic method.
Map skin and bone are registrated to target individual according to selected calibration point by second step.
Map skin and bone are filled with gray value similar in organ corresponding with target individual, use three dimensional grey scale image
Registration.Mode of texturing is deformed using cubic B-spline in registration process:
Wherein, T (x) is the transformation relation of corresponding points x registration front and back, xkFor control point, by the grid vertex of rule into
Row definition, β3It (x) is cubic B-spline multinomial, pkFor the motion vector at B-spline control point, σ is control point spacing, NkFor effect
Control point set at x point.
Similarity measure using mutual information (Mutual Information, MI) as image registration:
Wherein, IFAnd IMRespectively indicate fixed image and mobile image, LMAnd LFRespectively in mobile image and fixed image
The set for the strength information chosen at regular intervals, p are joint probability density, pFAnd pMRespectively fixed image and mobile image
Marginal probability density, marginal probability density acquired on f and m by joint probability density p, and f and m are respectively LMAnd LFOn
Variable-value.Joint probability density is estimated by B-spline Parzen windows, as follows:
Wherein, T (x) is the mode of texturing, ΩFIt is fixed image IMDomain, | ΩF| for number of voxel in image, wFWith
wMB-spline the Parzen windows, σ of respectively fixed image and mobile imageFAnd σMFor zoom factor, by LMAnd LF's
Width determines that these parameters arise directly from mobile image IM(x) and fixed image IF(x) intensity value ranges, or directly by
User is specified.
Similarity measure of the mutual information as image registration is not used only in registration, to match image skin and bone posture
It is quasi- more acurrate, the dissection calibration point information of the first step, corresponding mark in two images are added on the basis of the above similarity measure is measured
Auxiliary measurement index of the minimum range information of fixed point as similarity measure.Therefore registration map skin and bone are not only to examine
Consider image grayscale information, it is also contemplated that the positional relationship of known corresponding points obtains the map skin being registrated based on dissection calibration point
And bone.
Third step adjusts map body according to the obtained map skin and bone completed based on dissection calibration point registration
Postural change.
Postural change (the phase of mouse map is adjusted using the mode that Hongkai Wang et al. constructs deformable mouse map
Print paper: ADeformable Atlas of the Laboratory Mouse).Gesture stability skeleton is defined inside map,
Pass through bone Subspace Deformation mode (Skeletal Subspace Deformation, SSD), the posture of control atlas bone
Variation, control skeleton are to deform the control-rod established by implementation model, are not with without exception with the bone skeleton on anatomically significant
It reads.In the external control framework of map external definition, control atlas shoulder joint and hip joint in such a way that harmonics coordinates convert
The skin deformation at place, rest part skin use SSD control deformation.
Each control section in control skeleton is calculated by the map bone of registration front and back in registration result based on second step
Rigid deformation matrix in registration process, the rigid deformation of each control section in a manner of SSD control atlas bone curved surface
Deformation, mode of texturing are as follows:
p′i=(∑jωI, jTj)pi (4)
Wherein, piFor the four-dimensional homogeneous coordinates (x on i-th of vertex in mouse mapi, yi, zi, 1), TjIt is in control skeleton the
4 × 4 homogeneous transform matrix of j control section rigid deformation, ωI, jFor the influence power on the control section i-th of vertex in map
Weight, the weight information are defined by following formula:
Wherein, DI, jFor i-th of vertex to the shortest distance of j-th of control section, SiIt is that opposite vertexes i has dissection control
Point set.If vertex i is the vertex in skull, four limbs, claw or breastbone, SiThe as point set of the affiliated bone of vertex i.
If vertex i belongs to the vertex in backbone, rib cage, shoulder blade or clavicle, SiMiddle ωI, jThe part of > 0 can include multistage bone
Structure.The weight information need to use ωI, j/∑ωI, jNormalization, makes it meet ∑ ωI, j=1.
If the skin in mouse systemic map at shoulder joint and hip joint is directly deformed using SSD, it is easy to appear curved surface and collapses
Phenomenon is fallen into, the mode which need to use harmonics coordinates to convert deforms, by defining simple external control framework outside map,
Use the vertex of Simple framework as control point, corresponding skin curved surface is controlled by weighing factor of the vertex on skin curved surface
Deformation, deformation are defined as follows:
Wherein,It is the displacement vector on j-th of vertex in control framework,It is the position on i-th of vertex on skin curved surface
Move vector, hI, jIt is j-th of vertex in the control framework being calculated by harmonics coordinates, i-th of vertex on skin curved surface
The weight of generation.Skin curved surface in map at shoulder joint and hip joint is deformed using the harmonics coordinates, remaining
Skin curved surface is deformed using the SSD mode.
4th step adjusts map height and weight.
The variation of map height is caused by the variation of spine lengths, and mode of texturing meets linear scale:
Wherein, P is all apex coordinates after the variation of map height, P0For all apex coordinates of map initial configuration, O
Centre coordinate for mouse map expands to and P0Identical dimension can carry out matrix plus-minus operation, l0For map initial configuration
Spine lengths, l are the map spine lengths after height variation.
The variation of map weight is to lead to the variation of map skin due to the accumulation of subcutaneous fat, and skin caused by weight becomes
Shape vector VfIt can be obtained from sample by linear regression study, target individual first passes through in advance when carrying out the study of changes of weight
Standardization, sample have unified body size and posture form.Be accordingly used in control atlas changes of weight parameter need usingIt standardizes, wherein wkAnd lkFor the spine lengths and weight of k-th of individual mice, map caused by changes of weight is all
The variation of apex coordinate may be expressed as:
Wherein, w0For initial atlas weight, above-mentioned formula assumes P and P0Possess identical body length, uses following manner
Reflect the variation of all apex coordinates of map caused by height and changes of weight simultaneously:
P=S (l, W (w, P0)) (9)
Wherein, l and w is the input variable being independent of each other for two.In practice, the height of mouse and weight are to change simultaneously
, the variation relation of height and weight can be described by l=g (w), which can count from training sample and obtain, benefit
The variation of all apex coordinates of map caused by map height and changes of weight can be described in the following manner with the variation relation:
P=S (g (w), W (w, P0)) (10)
5th step maps remaining mouse internal by map skin after posture, height and changes of weight and bone.
Skin and bone are under the jurisdiction of statistical shape model (Statistical Shape Model, SSM) SSM in map1,
Lung, heart, liver, spleen and kidney are under the jurisdiction of statistical shape model SSM2, two statistical shape model form factor b1And b2Respectively
Gaussian distributed, and there are statistic correlations between the two, can use conditional Gaussian model (Conditional Gaussian
Model, CGM) description:
∑2|1=∑2+∑2,1(∑1)-1∑1,2 (13)
Wherein,For the mean value of conditional probability distribution, ∑2|1For the covariance matrix of conditional probability distribution,And ∑1
For b1Mean value and covariance matrix,And ∑2For b2Mean value and covariance matrix, ∑2,1And ∑1,2It is b1And b2Between
Cross-covariance.∑1,∑2, ∑2,1And ∑1,2Value can from training sample concentrate acquire.According to conditional Gaussian mould
The formula of type is the estimation that the low contrast organ shape and position are provided using skin and bone.For being unsatisfactory for counting
Learn deformation rule brain, according near brain skin and bone choose control point, use thin plate spline (Thin Plate
Spline, TPS) mode of texturing mapped.
Map bone, skin and lung are registrated in a manner of gray level image registration with target individual by the 6th step.
Both the mouse systemic map after posture, height and changes of weight, including all inside had been obtained after the completion of 4th step
Organ.Skin, bone and lung are extracted from the map after the variation, are filled with gray scale similar in organ corresponding with target image
Value is registrated mode with second step, carries out the registration of three dimensional grey scale image.Use cubic B-spline as anamorphose mode, uses
Mutual information is as registration similarity measure.With the deformation field that will definitely obtain three dimensional grey scale image registration, form is controlled using deformation field
Map deformation after posture, height and changes of weight, obtains final registration result.
7th step can be in the map deformation result of the 6th step if result and target individual gap that the 6th step obtains are larger
On the basis of, skin and bone are extracted, the process of the 4th step to the 6th step is repeated, until the registration result met the requirements.
Beneficial effects of the present invention: the deformable feature of mouse systemic map is utilized, previously according to mesh before image registration
Body gesture, height and the weight of mark individual adjustment map, then image registration is carried out with target individual.It solves simultaneously existing small
The problem of body posture variation is with organ individual morphology difference in mouse whole body atlas registration technology, and algorithm has higher registration
Precision and robustness.The present invention will play the role of actively promoting for toy image analysing computer research, improve preclinical toy and grind
The data analysis capabilities studied carefully promote the development of associated biomolecule medical research.
Detailed description of the invention
Fig. 1 is the algorithm flow chart of the present invention deformable mouse systemic map and mouse Image registration.
In figure: (a) target mouse CT images；(b) deformable mouse systemic map；(c) target individual dissects calibration point；
(d) map dissects calibration point；(e) the map skin and bone after posture, height and changes of weight；(f) posture, height and weight
Whole body map after variation；(g) the filling image of map bone, skin and lung；(h) registration result.
Specific embodiment
A specific embodiment of the invention is described in detail below in conjunction with technical solution and attached drawing.
As shown in Figure 1, a kind of algorithm of deformable mouse systemic figure map and mouse Image registration, target image are schemed with CT
As for, it is registrated using deformable mouse systemic map.Mainly include three parts: pre-adjusting deformable mouse systemic figure
Compose the form posture of skin and bone；Remaining internal is mapped according to deformed skin and bone；Pass through Three-Dimensional Gray figure
As with the deformed mouse systemic atlas registration of brigadier to target individual.Specific embodiment is as follows:
The first step chooses dissection calibration point from mouse image.It includes nose, cervical vertebra that target individual, which dissects calibration point (c),
Top, coccyx top, left fore elbow joint, right fore elbow joint, left hind knee joint and right hind knee joint totally 7 joints
Point, the present invention do not limit to the selection of this 7 calibration.Selection mode can be from target mouse CT images (a) or bone segmentation result
In in a manner of manually or automatic detection mode choose.Map dissects calibration point (d) can be from deformable mouse systemic map (a)
Defined in choose in artis.
Map skin and bone are registrated to target individual according to selected calibration point by second step.By map skin and bone
It is filled with gray value similar in organ corresponding with target individual, is registrated using three dimensional grey scale image.B three times is used in registration process
Batten uses mutual information and corresponding calibration point minimum range information to survey as the similitude of image registration as anamorphose mode
Degree.It obtains based on the map skin and bone after dissection calibration point registration.
Third step adjusts map according to the map skin and bone completed obtained by second step based on dissection calibration point registration
Body gesture variation.By registration front and back map bone can be calculated control skeleton in each control section in registration process
Rigid deformation matrix controls the deformation of whole body map bone curved surface the rigid deformation of each control section in a manner of SSD.
If the skin in mouse systemic map at shoulder joint and hip joint is directly deformed using SSD, it is easy to appear curved surface and collapses
It falls into, the mode which need to use harmonics coordinates to convert deforms, and simple external control framework is defined outside map, uses letter
The vertex of easy frame controls corresponding skin curved surface by weighing factor of the vertex on skin curved surface and deforms as control point.
The skin curved surface at remaining position is deformed using the SSD mode.
4th step adjusts map height and weight.Changed using target individual spine lengths Serial regulation map height.Make
With the registration result of map skin in the second step, vector difference between initial atlas, binding model training gained weight
The deformation vector that variation causes skin curved surface to change, is calculated map changes of weight coefficient.Finally obtain posture, height and body
Map skin and bone (e) change again after.
5th step maps device inside remaining mouse by map skin after posture, height and changes of weight and bone (e)
Official.Wherein meet the mouse internal of statistical variations rule, lung, heart, liver, spleen and kidney, using condition height
This model is estimated by high contrast organ skin and bone.It is unsatisfactory for the mouse internal of statistical variations rule, brain,
It is mapped using the mode of texturing of thin plate spline.Mouse systemic map after finally obtaining posture, height and changes of weight
Including each internal of mouse (f),.
Map bone, skin and lung are registrated in a manner of gray level image registration with target individual by the 6th step.It is described
Both the mouse systemic map (f) after posture, height and changes of weight had been obtained after the completion of 4th step, included all internals, from this
Skin, bone and lung are extracted in map after variation, are filled using gray value similar in organ corresponding with target image, are obtained figure
The filling image (g) of skin, bone and lung is composed, the registration of three dimensional grey scale image is then carried out.B sample three times is used in registration process
Item uses mutual information as the similarity measure of image registration as anamorphose mode.Three-Dimensional Gray figure can be obtained in registration
As the deformation field of registration, is deformed, obtained using the mouse systemic map (f) after deformation field control posture, height and changes of weight
Final registration result (h).
7th step can be in the map deformation result of the 6th step if result and target individual gap that the 6th step obtains are larger
On the basis of, skin and bone are extracted, the process of the 4th step to the 6th step is repeated, until the registration result met the requirements.
Claims (1)
1. a kind of algorithm of deformable mouse systemic map and mouse Image registration, which is characterized in that steps are as follows:
The first step chooses dissection calibration point from mouse image
Dissection calibration point need to choose the artis of bone and the central point of internal organs, in order to determine mouse body posture, mouse
Dissection calibration point should be respectively contained at least one in four limbs, remaining calibration point is determined whether to add according to practical application effect by user
Add；
Map skin and bone are registrated to target individual according to selected calibration point by second step
Map skin and bone are filled with gray value similar in organ corresponding with target individual, matched using three dimensional grey scale image
It is quasi-；Mode of texturing is deformed using cubic B-spline in registration process:
Wherein, T (x) is the transformation relation of corresponding points x registration front and back, xkFor control point, determined by the grid vertex of rule
Justice, β3It (x) is cubic B-spline multinomial, pkFor the motion vector at B-spline control point, σ is control point spacing, NkTo act on x
Control point set at point；
Use mutual information as the similarity measure of image registration:
Wherein, IFAnd IMRespectively indicate fixed image and mobile image, LMAnd LFRespectively one is pressed in mobile image and fixed image
The set for the strength information that fixed interval is chosen, p is joint probability density, pFAnd pMThe side of respectively fixed image and mobile image
Edge probability density, marginal probability density are acquired on f and m by joint probability density p, and f and m are respectively LMAnd LFOn variable
Value；Joint probability density is estimated by B-spline Parzen windows, as follows:
Wherein, T (x) is the mode of texturing, ΩFIt is fixed image IMDomain, | ΩF| for number of voxel in image, wFAnd wMPoint
B-spline the Parzen windows, σ of image and mobile image Wei not fixedFAnd σMFor zoom factor, by LMAnd LFWidth
Degree determines that these parameters arise directly from mobile image IM(x) and fixed image IF(x) intensity value ranges, or directly by with
Family is specified；
Similarity measure of the mutual information as image registration is not used only in registration, is that image skin and bone posture registration are more quasi-
Really, the dissection calibration point information of the first step is added on the basis of the above similarity measure is measured, and corresponds to calibration point in two images
Auxiliary measurement index of the minimum range information as similarity measure；Registration map skin and bone not only consider that image grayscale is believed
Breath, it is also contemplated that the positional relationship of known corresponding points obtains the map skin and bone being registrated based on dissection calibration point；
Third step adjusts map body gesture according to the obtained map skin and bone completed based on dissection calibration point registration
Variation
Gesture stability skeleton is defined inside map, passes through bone Subspace Deformation mode, the postural change of control atlas bone, control
Skeleton processed is to deform the control-rod established by implementation model；Control framework outside map external definition, is become by harmonics coordinates
The skin deformation at mode control atlas shoulder joint and hip joint changed, rest part skin use SSD control deformation；
Registration result based on second step is calculated each control section in control skeleton by the map bone of registration front and back and is matching
Rigid deformation matrix during standard, the rigid deformation of each control section, control atlas bone curved surface is deformed in a manner of SSD,
Mode of texturing is as follows:
p′i=(∑jωI, jTj)pi (4)
Wherein, piFor the four-dimensional homogeneous coordinates (x on i-th of vertex in mouse mapi, yi, zi, 1), TjIt is j-th in control skeleton
4 × 4 homogeneous transform matrix of control section rigid deformation, ωI, jIt, should for the weighing factor on the control section i-th of vertex in map
Weight information is defined by following formula:
Wherein, DI, jFor i-th of vertex to the shortest distance of j-th of control section, SiIt is the point set that opposite vertexes i has dissection control
It closes；If vertex i is the vertex in skull, four limbs, claw or breastbone, SiThe as point set of the affiliated bone of vertex i；If top
Point i belongs to the vertex in backbone, rib cage, shoulder blade or clavicle, SiMiddle ωI, jThe part of > 0 can include multistage skeletal structure；Institute
ω need to be used by stating weight informationI, j/∑ωI, jNormalization, makes it meet ∑ ωI, j=1；
If the skin in mouse systemic map at shoulder joint and hip joint is directly deformed using SSD, be easy to appear curved surface collapse it is existing
As the mode that the partial skin need to use harmonics coordinates to convert deforms, and by defining simple external control framework outside map, uses
The vertex of Simple framework controls corresponding skin curved surface by weighing factor of the vertex on skin curved surface and becomes as control point
Shape, deformation are defined as follows:
Wherein,It is the displacement vector on j-th of vertex in control framework,It is the displacement arrow on i-th of vertex on skin curved surface
Amount, hI, jIt is that j-th of vertex in the control framework being calculated by harmonics coordinates, i-th of vertex on skin curved surface generates
Weight；Skin curved surface in map at shoulder joint and hip joint is deformed using the harmonics coordinates, remaining skin
Curved surface is deformed using the SSD mode；
4th step adjusts map height and weight
The variation of map height is caused by the variation of spine lengths, and mode of texturing meets linear scale:
Wherein, P is all apex coordinates after the variation of map height, P0For all apex coordinates of map initial configuration, O is small
The centre coordinate of mouse map expands to and P0Identical dimension carries out matrix plus-minus operation；l0Backbone for map initial configuration is long
Degree, l are the map spine lengths after height variation；
The variation of map weight is to lead to the variation of map skin due to the accumulation of subcutaneous fat, skin deformation caused by weight to
Measure VfIt is to be obtained from sample by linear regression study, target individual first passes through standard when carrying out the study of changes of weight in advance
Change, sample has unified body size and posture form；Be accordingly used in control atlas changes of weight parameter need using
It standardizes, wherein wkAnd lkFor the spine lengths and weight of k-th of individual mice, all vertex of map caused by changes of weight are sat
Target variation indicates are as follows:
Wherein, w0For initial atlas weight, above-mentioned formula assumes P and P0Possess identical body length, simultaneously using following manner
Reflect the variation of all apex coordinates of map caused by height and changes of weight:
P=S (l, W (w, P0)) (9)
Wherein, l and w is the input variable being independent of each other for two；In practice, the height of mouse and weight change simultaneously,
Height and the variation relation of weight are described by l=g (w), which counts from training sample and obtain, and utilize the variation
Relationship can describe the variation of all apex coordinates of map caused by map height and changes of weight in the following manner:
P=S (g (w), W (w, P0)) (10)
5th step maps remaining mouse internal by map skin after posture, height and changes of weight and bone
Skin and bone are under the jurisdiction of statistical shape model SSM in map1, lung, heart, liver, spleen and kidney are under the jurisdiction of statistics shape
Shape model SSM2, two statistical shape model form factor b1And b2Gaussian distributed respectively, and there are statistical correlations between the two
Property, it is described with conditional Gaussian model:
∑2|1=∑2+∑2,1(∑1)-1∑1,2 (13)
Wherein,For the mean value of conditional probability distribution, ∑2|1For the covariance matrix of conditional probability distribution,And ∑1For b1's
Mean value and covariance matrix,And ∑2For b2Mean value and covariance matrix, ∑2,1And ∑1,2It is b1And b2Between mutual association side
Poor matrix；∑1,∑2, ∑2,1And ∑1,2Value from training sample concentration acquire；According to the formula of conditional Gaussian model
The estimation of the low contrast organ shape and position is provided using skin and bone；For being unsatisfactory for statistics deformation rule
The brain of rule, according near brain skin and bone choose control point, mapped using the mode of texturing of thin plate spline；
Map bone, skin and lung are registrated in a manner of gray level image registration with target individual by the 6th step
Both the mouse systemic map after posture, height and changes of weight, including all internals had been obtained after the completion of 4th step；From this
Skin, bone and lung are extracted in map after variation, are filled with gray value similar in organ corresponding with target image, same to second step
Registration mode carries out the registration of three dimensional grey scale image；Use cubic B-spline as anamorphose mode, use mutual information as
It is registrated similarity measure；With will definitely three dimensional grey scale image registration deformation field, utilize deformation field control form posture, height and body
Map deformation change again after, obtains final registration result；
7th step, if result and target individual gap that the 6th step obtains are larger, on the basis of the map deformation result of the 6th step,
Skin and bone are extracted, the process of the 4th step to the 6th step is repeated, until the registration result met the requirements.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910501435.7A CN110322491B (en) | 2019-06-11 | 2019-06-11 | Algorithm for registering deformable mouse whole-body atlas and mouse image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910501435.7A CN110322491B (en) | 2019-06-11 | 2019-06-11 | Algorithm for registering deformable mouse whole-body atlas and mouse image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110322491A true CN110322491A (en) | 2019-10-11 |
| CN110322491B CN110322491B (en) | 2022-03-04 |
Family
ID=68119510
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910501435.7A Active CN110322491B (en) | 2019-06-11 | 2019-06-11 | Algorithm for registering deformable mouse whole-body atlas and mouse image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110322491B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113616273A (en) * | 2021-08-04 | 2021-11-09 | 长安大学 | Positioning block manufacturing method and system for precise replacement of artificial knee joint |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101013503A (en) * | 2007-01-26 | 2007-08-08 | 清华大学 | Method for segmenting abdominal organ in medical image |
| WO2013019985A2 (en) * | 2011-08-03 | 2013-02-07 | The Regents Of The University Of California | 2d/3d registration of a digital mouse atlas with x-ray projection images and optical camera photos |
| CN103514606A (en) * | 2013-10-14 | 2014-01-15 | 武汉大学 | Heterology remote sensing image registration method |
| CN104867104A (en) * | 2015-05-20 | 2015-08-26 | 天津大学 | Method for obtaining anatomical structural atlas for target mouse based on XCT image non-rigid registration |
| CN106530338A (en) * | 2016-10-31 | 2017-03-22 | 武汉纺织大学 | Method and system for matching MR image feature points before and after nonlinear deformation of biological tissue |
| CN106920228A (en) * | 2017-01-19 | 2017-07-04 | 北京理工大学 | The method for registering and device of brain map and brain image |
| CN108053431A (en) * | 2018-02-24 | 2018-05-18 | 中原工学院 | A kind of non-rigid medical image registration method based on gradient distribution |
| CN108428245A (en) * | 2018-02-11 | 2018-08-21 | 中国科学院苏州生物医学工程技术研究所 | Sliding method for registering images based on self-adapting regular item |
| CN109242865A (en) * | 2018-09-26 | 2019-01-18 | 上海联影智能医疗科技有限公司 | Medical image auto-partition system, method, apparatus and storage medium based on multichannel chromatogram |
-
2019
- 2019-06-11 CN CN201910501435.7A patent/CN110322491B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101013503A (en) * | 2007-01-26 | 2007-08-08 | 清华大学 | Method for segmenting abdominal organ in medical image |
| WO2013019985A2 (en) * | 2011-08-03 | 2013-02-07 | The Regents Of The University Of California | 2d/3d registration of a digital mouse atlas with x-ray projection images and optical camera photos |
| CN103514606A (en) * | 2013-10-14 | 2014-01-15 | 武汉大学 | Heterology remote sensing image registration method |
| CN104867104A (en) * | 2015-05-20 | 2015-08-26 | 天津大学 | Method for obtaining anatomical structural atlas for target mouse based on XCT image non-rigid registration |
| CN106530338A (en) * | 2016-10-31 | 2017-03-22 | 武汉纺织大学 | Method and system for matching MR image feature points before and after nonlinear deformation of biological tissue |
| CN106920228A (en) * | 2017-01-19 | 2017-07-04 | 北京理工大学 | The method for registering and device of brain map and brain image |
| CN108428245A (en) * | 2018-02-11 | 2018-08-21 | 中国科学院苏州生物医学工程技术研究所 | Sliding method for registering images based on self-adapting regular item |
| CN108053431A (en) * | 2018-02-24 | 2018-05-18 | 中原工学院 | A kind of non-rigid medical image registration method based on gradient distribution |
| CN109242865A (en) * | 2018-09-26 | 2019-01-18 | 上海联影智能医疗科技有限公司 | Medical image auto-partition system, method, apparatus and storage medium based on multichannel chromatogram |
Non-Patent Citations (2)
| Title |
|---|
| M.BAIKER等: "FULLY AUTOMATED WHOLE-BODY REGISTRATION IN MICE USING AN ARTICULATED SKELETON ATLAS", 《2007 4TH IEEE INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING: FROM NANO TO MACRO》 * |
| 屠睿博 等: "基于随机森林算法的小鼠micro-CT影像中骨骼关节特征点定位", 《中国生物医学工程学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113616273A (en) * | 2021-08-04 | 2021-11-09 | 长安大学 | Positioning block manufacturing method and system for precise replacement of artificial knee joint |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110322491B (en) | 2022-03-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110335358B (en) | Personalized deformation method of deformable digital human anatomy model | |
| CN106485695B (en) | Medical image Graph Cut dividing method based on statistical shape model | |
| CN104851123B (en) | A kind of three-dimensional face change modeling method | |
| Kroon | Segmentation of the mandibular canal in cone-beam CT data. | |
| CN106061379B (en) | Processing equipment, processing method and program | |
| CN108898606A (en) | Automatic division method, system, equipment and the storage medium of medical image | |
| CN103229210B (en) | Image registration device | |
| CN104867104B (en) | Target mouse anatomical structure collection of illustrative plates acquisition methods based on the non-rigidity registration of XCT images | |
| CN107680110B (en) | Inner ear three-dimensional level set segmentation method based on statistical shape model | |
| CN110338840A (en) | The display processing method and 3-D supersonic imaging method and system of three-dimensional imaging data | |
| CN108830852B (en) | Three-dimensional ultrasonic tumor auxiliary measurement system and method | |
| CN107220965A (en) | A kind of image partition method and system | |
| CN106846330A (en) | Human liver's feature modeling and vascular pattern space normalizing method | |
| CN104545999B (en) | Method and device for measuring bladder volume through ultrasound images | |
| CN106548476B (en) | Lung's three-dimensional feature Method On Shape is counted using medical image | |
| CN114119872A (en) | Method for analyzing 3D printing intraspinal plants based on artificial intelligence big data | |
| Claes | A robust statistical surface registration framework using implicit function representations-application in craniofacial reconstruction | |
| CN107424162A (en) | A kind of image partition method and system | |
| Wang et al. | Deformable torso phantoms of Chinese adults for personalized anatomy modelling | |
| CN115830016A (en) | Medical image registration model training method and equipment | |
| KR20140061380A (en) | Image processing method | |
| CN110322491A (en) | A kind of algorithm of deformable mouse systemic map and mouse Image registration | |
| CN112466441A (en) | Organ delineation method and device | |
| WO2018201437A1 (en) | Image segmentation method and system | |
| TW201221117A (en) | Enhancing segmentation for molecular image quantification system and the method thereof |
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 |