CN106846296A - A kind of cell image tracks intelligent algorithm - Google Patents
A kind of cell image tracks intelligent algorithm Download PDFInfo
- Publication number
- CN106846296A CN106846296A CN201611186286.2A CN201611186286A CN106846296A CN 106846296 A CN106846296 A CN 106846296A CN 201611186286 A CN201611186286 A CN 201611186286A CN 106846296 A CN106846296 A CN 106846296A
- Authority
- CN
- China
- Prior art keywords
- target
- time point
- coordinate
- image
- axis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10064—Fluorescence image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10101—Optical tomography; Optical coherence tomography [OCT]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30241—Trajectory
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of cell image tracking intelligent algorithm, the three-dimensional tracking to neutrophil leucocyte in zebrafish embryo is realized.Picking rate reach cameras frame rate, to biological sample light radiate it is small.Various video procession algorithms are used for the detection of neutrophil leucocyte and identification in zebrafish embryo, the change in location and morphological feature of cell is obtained.For next step data analysis lays the foundation.And neutrophil leucocyte structure, morphological feature, movement locus and inflammation and pharmaceutically-active correlation are analyzed, it is that study of disease change and drug action mechanism establish experiment and theoretical foundation.
Description
Technical field
Field is recognized the present invention relates to optical imaging system and image processing mode, is related to a kind of instantaneous orthogonal optical to project
Chromatographic imaging system.
Background technology
Disclose the main line that the essence of life, concern human health are contemporary science and technology development.Carried out into by biological specimen
As studying the design feature and physiological function of sample, the purpose that objective basis are provided for medical clinic applicationses is can reach.Optics
Be combined for CT technologies and microtechnic by projection tomography (Optical Projection Tomography, OPT) technology,
The new technology of three-dimensional imaging can be carried out to fluorescence and non-fluorescence material.OPT has the resolution ratio higher than nuclear magnetic resonance, up to micro-
Rice magnitude；With the imaging depth bigger than confocal microscopic imaging, sample that can be to thickness up to more than ten millimeters is imaged, also may be used
Tissue specificity or gene specific coloring treatment are carried out to sample with using colored or fluorescent dye.When OPT is imaged, first gather
Data for projection of the sample difference tomography under different angles, then these data applications Radon inverse transformations are recombinated by computer
Image.OPT has important value in the research of the aspects such as tissue development, gene expression and medical diagnosis.
Understanding biosystem not only needs to study the spatial distribution of cell, in addition it is also necessary to studies them and becomes with the dynamic of time
Change.The dynamic for studying cell is the situation of movement of cell, and this changes to study of disease, cancer metastasis is particularly important.Tissue
In be forced bad place and have inflammatory reaction, immunocyte can change, and at this moment leucocyte can increase.Signaling situation can be with
Launch according to the threedimensional model in tissue.In the model caught an illness, the inner cell movement of research living organisms, this is to biological medicine
Study mechanism has important practical significance.
Position and track cell using instantaneous orthogonal imaging technique (gathering two images simultaneously in orthogonal CCD)
Motion.Research cell detection and recognizer, extract many reference amounts morphological feature of cell.It is biological fabric study and medicine
Operation mechanism provides new technological means and thinking.
The content of the invention
In view of this, in order to reach the effect of above-mentioned technical proposal, the present invention provide it is a kind of solve or part solve it is above-mentioned
A kind of cell image tracking intelligent algorithm of problem：
Step one, IMAQ：
Zebrafish embryo is placed and starts rotation on a spinstand, while rotation, by instantaneous orthogonal optical imagery
System carries out IMAQ to cell object in zebrafish embryo；It is right by obtaining projected image pair after complete IMAQ
Projected image carries out figure by the way that mathematical morphology is smooth to carrying out image enhaucament, image denoising and image segmentation with size selection
As segmentation, multiple neutrophil leucocyte images are obtained, by tracking all neutrophil leucocytes in multiple neutrophil leucocyte images
Movement locus, the area of the multiple neutrophil leucocyte images of setting segmentation is less than a particular value；And split with watershed algorithm
The situation for contacting with each other and overlapping；
There is the coordinate position that record current time all neutrophil leucocytes in the multiple neutrophil leucocyte images of collection；Adopt
It is separated by two field pictures shooting time at intervals of a set time during collection multiple neutrophil leucocyte image；
Step 2, track algorithm：
The multiple neutrophil leucocyte images of analysis, and using the one of neutrophil leucocyte in the object for analyzing as mesh
Mark；
Any given time point k is time point k+1 by the future time point after a set time；Time point k+1 is passed through
After the future time point after a set time be time point k+2；
Target during by time point k is connected with the target nearest in time point k+1, by comparison prediction position
The geometric distance difference with physical location is put, judges whether to list the trajectory lists of target in；
The measurement in step 3, cell morphology characteristic, movement velocity and direction：
Behind the track of reconstruct tracking target, following characteristics are calculated：
The total length in the track of target, is the total displacement of a neutrophil leucocyte；
The track total displacement of target, is a distance for the beginning and end of neutrophil leucocyte；
The average speed value and the direction of motion of target；
Tracking time interval；
According to the object judgement direction of motion：From target in time point k, during time point k+1, three during time point k+2
After position carries out line, the included angle cosine value of two neighboring displacement increment is calculated, obtain the direction of motion；
If 1 direction of motion direction value of included angle cosine value is 1；The referred to as unidirectional direction of motion；If included angle cosine value
For 0 direction of motion direction value is 0；The referred to as complete randomness direction of motion；
It is that, with the unidirectional direction of motion, the target remote apart from wound will not that the target near apart from wound can be affected
It is exactly the complete randomness direction of motion to be affected, for judging whether target can move towards the wound of zebrafish embryo；
Step 4, cell detection and identification：
The X-axis of definition measurement projected coordinate system, Y-axis, Z axis；Z axis are vertical coordinates；X-axis is abscissa from multiple neutrality grains
The left side of cell image starts measurement, and Y-axis is another abscissa, orthogonal to Y-axis；
The projected image in projected image and YZ axles that projected image is divided on XZ axles with Z axis；
The projected image in projected image and YZ axles in each set time on XZ axles respectively gathers piece image,
K is obtained to reduce light radiation when three-dimensional tracking is carried out to zebrafish embryo, and in time period between 0 to time point k
Projected image pair, by image segmentation and triangulation, measure coordinate of the target in time point k is Xk, Yk, Zk, wherein：Xk、
Yk、ZkRespectively target in time point k in X-axis, in Y-axis, the measure coordinate on Z axis；Xk+1、Yk+1Respectively target when
Between point k+1 when in X-axis, the measure coordinate value in Y-axis；
Prediction coordinate of the target in time point k is P (xk, yk, zk), wherein：P(xk)、P(yk)、P(zk) it is respectively target
In time point k in X-axis, in Y-axis, the prediction coordinate value on Z axis；
Coordinate prediction is carried out using formula (1)；
And Pk(zk+1)=Zk+1Formula (1)
P in formula (1)k(zk+1)=Zk+1In Pk(zk+1) it is prediction coordinate of the target in time point k+1 on Z axis
Value；Zk+1It is measure coordinate value of the target in time point k+1 on Z axis；Pk(zk+1)=Zk+1Represent target in time point k+1
When prediction coordinate value on Z axis, it is equal with measure coordinate value of the target in time point k+1 on Z axis；
In formula (1) is dot product symbol
When θ in formula (1) is rotating acquisition in each time point k arbitrary neighborhood collection picture frame angle step,
XAOR, YAORIt is the coordinate position of XY axle of the rotary shaft in projected coordinate system；
Projected image detection single target in projected image and YZ axles on XZ axles, it is adjustable, non-linear using size
Cap functional image Processing Algorithm high is identified target；
The effective coverage of single target is judged according to the segmentation of multiple target sizes empirical values, (fitting is exactly plane for fitting
Upper a series of point, is coupled together with a smooth curve) geometric center of multiple targets is calculated, in the seat of XZ coordinate systems
(x, z) is designated as, is (y, z) in the coordinate of YZ coordinate systems；
When being identified to the target in projected image pair to k, if close Z coordinate value is then considered same mesh
Mark；There is close Z coordinate value in the presence of 2 or more targets in XZ coordinate systems and YZ coordinate systems, as turntable is rotated, if
It is different target, the change of the Z coordinate value in ensuing image is unsatisfactory for the boundary condition of setting so as to exclude；
Step 5, reconstruct carry out the movement locus of the segmentation object of measurement projected coordinate system：
Connected in each target of any given frame target nearest with next frame；The target of all identifications is in kth
The coordinate X of the detection at individual time pointk, Yk, ZkList of coordinates transforms to+1 coordinate system P (x at time point of kthk+1, yk+1,
zk+1), P (xk+1, yk+1, zk+1) and in+1 coordinate X of time point detection of kthk+1, Yk+1, Zk+1It is compared, the space of target
Same target is considered when distance is in the range of some, each target can only add a track；
P(xk+1, yk+1, zk+1) it is that target is being in k-th coordinate X at time point according to the target for recognizingk, Yk, ZkIn kth
+ 1 prediction coordinate at time point；
If corresponding target is not identified at+1 time point of kth, changed commanders+2 time points of kth using suitable change
The coordinate at+2 time points of prediction coordinate and kth be compared；If corresponding target is not still identified, target
Tracking terminates at k-th time point.
Brief description of the drawings
Fig. 1 is the Technology Roadmap of tracking cell intelligent algorithm
Fig. 2 is the track algorithm flow chart of tracking cell intelligent algorithm
Fig. 3 is that the focal plane of orthogonal optical projection chromatographic imaging system (shows orthogonal imaging system for intersecting plane
In the focusing of rotary shaft) (FP focal planes, rotary shaft)
Fig. 4 is another Technology Roadmap of tracking cell intelligent algorithm
Fig. 5 is that the top view of orthogonal optical projection chromatographic imaging system (shows orthogonal imaging system for intersecting plane
In the focusing of rotary shaft) (FP focal planes, rotary shaft)
Specific embodiment
In order that the technical problems to be solved by the invention, technical scheme and beneficial effect become more apparent, below tie
Drawings and Examples are closed, the present invention will be described in detail.It should be noted that specific embodiment described herein is only used
To explain the present invention, it is not intended to limit the present invention, can realizes that the product of said function belongs to equivalent and improvement, includes
Within protection scope of the present invention.Specific method is as follows：
Embodiment one：The concrete results of tracking cell intelligent algorithm are：
(1) propose a kind of instantaneous orthogonal optical projection tomography technology of high accuracy, sample-size and imaging precision it
Between weighed, improve resolution ratio, imaging efficiency and signal to noise ratio.
(2) the instantaneous orthogonal imaging three-dimensional tracking system of New Rotary is built, to the immunocyte of live body model animal
(such as：Neutrophil leucocyte) three-dimensional tracking is carried out, make great efforts to improve time resolution.Each time is done in relatively conventional three-dimensional tracking to sample
The three-dimensional imaging of point, the present invention realizes the minimum light radiation that three-dimensional tracking is carried out to sample using instantaneous orthogonal imaging system
(minimum photobleaching and phototoxicity), the injury to biological living is small, for biology, medical science, medical research provide new technology hand
Section, with important application value.
(3) realize that three-dimensional time postpones transgenic zebrafish embryo neutrality using twin shaft OPT (in rectangular projection angle)
Granulocyte mobile tracking.Research cell fine structure, detects intercellular multidate information, is that medical science and drug research field are badly in need of
The research direction of solution.
With the progress of science and technology, biomedical imaging technology achieves fast development.It is by means of certain energy
With the interaction of organism, the information of the form, structure and physiological function of biological in-vivo tissue or organ is extracted, be biological
Fabric study and clinical diagnosis provide image information.The wide range being related to, there is X ray computer Tomography (X-
Ray CT), Magnetic resonance imaging (MRI), ultrasonic imaging (UI), optical imagery etc..
It is imaged with the interaction of biological tissue using light, favor is enjoyed always, because optical imagery has to group
Knit interference it is small, sensitivity is high, abundant information (can multi-parameter combined measurement, these parameters include structure, frequency spectrum, polarization, quantum effect
Should be with the life-span etc.), up to subcellular fraction level level of resolution, it is suitable in vitro or live body real time imagery, it is easy to other technologies such as
Combination of ultrasound is used, and the advantages of possess the potential of direct treatment.Light enriches very much with the interaction mechanism of tissue, there is anti-
Penetrate, scatter, absorbing, fluorescence etc., these optical phenomenas are used equally to extract the relevant information in tissue, so as to produce based on them
Various imaging methods so that optical means turns into studies the most active, field with the fastest developing speed in imaging in biological tissues.
But, due to the inhomogeneities of biological tissue, there are absorption and strong scattering to light wave, therefore light wave is difficult deep tissue
It is internal, it is more difficult to clearly image is extracted from biological tissue.In order to solve this problem, people have been developed one and are in recent years
Row optical imaging method.For example：Copolymerization Jiao/multi-photon optical image technology, optical coherent chromatographic imaging, mating plate frequency microscope, sweep
Retouch laser optics tomoscan, optical projection tomography (OPT) etc..These medical imaging technique means have respective applicable model
Enclose and list the advantage and disadvantage of several conventional imaging techniques with respective limitation, table 1.
The current method for can be carried out to mm-cm magnitudes biological object three-dimensional optical imaging has：OPT、SLOT、LSM.
LSM is fast with respect to image taking speed, but each opposite sex in precision direction being imaged, and the image information of extraction is easily obscured.OPT and SLOT, knot
Structure is simply easily operated, the imaging precision all directions same sex.But OPT is wide field imaging mode, is excited by Whole-field illumination and is adopted with CCD
Collection projection imaging, picking rate is faster.SLOT is to be excited by single-point and detect and be imaged.
Biomedical three-dimensional imaging provides new technological means for biological and medical research.Study the dynamic of live body inner cell
State distribution (three-dimensional tracking) is one, biomedical imaging field significantly problem, to detect the change and research of disease
The mechanism of action of medicine provides new technological means.So far, cell three-dimensional tracking utilize microscope realization, it is necessary to
Three-D volumes information is obtained at each time point.Copolymerization Jiao/multi-photon scan laser microphotograph provides optical section, these optics
Section is conducive to the acquisition of fluorescence labeled cell 3-D view storehouse.This method is used for based on high-resolution research zebra fish
The neutrophil leucocyte for having inflammation zonule, such as：The model of wounded tail, about 100 μm of visual fields.However, in fact most
Number laser scanning microscope is restricted in penetration depth and visual field, and they substantially show the resolution ratio of anisotropic,
Jie that whole sample can be expanded to sees imaging, but needs exciting power and image acquisition time long high.We may be selected
, for the 3 Dimension Image Technique of mm-cm magnitudes biological tissue, these technologies have benefited from full field image collection, including optics is thrown for some
Shadow tomography (optical projection tomography) and mating plate microtechnic (light sheet
microscopy).These technologies can be directly extended to three-D volumes imaging, for example：Whole sample is obtained at each time point
Volume.Algorithm that is improved and realizing directly realizes that three-dimensional time postpones feature in being imaged using scan laser microphotograph
Tracking.Regrettably, volume imagery needs to obtain the full stack diagram picture on each time point for LSM or expiring for OPT
Sequential projection.Millimeter magnitude realizes that cell class resolution ratio requires more than 100 frames per volume.Which has limited time delay resolution ratio,
These features move effective distance in this segment body product acquisition time.And, obtain hundreds of images, knot at each time point
Fruit is the substantial amounts of light radiation in whole time course to sample, and this also leads to photobleaching or phototoxicity so that viviperception
Have a greatly reduced quality.
In sum, it is proposed that a new three-dimensional tracking, extends angular multiplexed OPT methods and intend with camera frame
Speed realizes three-dimensional tracking.Intend obtaining image simultaneously in different projection angles, using each time point triangulation, without
Whole sample volume is reconstructed, the positional information of three-dimensional feature is only estimated.So, to the collection of single image rather than whole volume
Collection, is equivalent to successive projection collection and the delay of each time point light radiation in three-dimensional feature tracking temporal resolution.It is this
Acquisition system can provide isotropic spatial resolution, and (focusing/multi-photon scan laser microphotograph or traditional mating plate show together
The three-dimensional tracking of microtechnology is different).Intend, with improved OPT acquisition schemes, tracking a disease model (such as：Zebrafish embryo)
Signaling is particularly convenient.
Embodiment two：(1) propose that a kind of instantaneous orthogonal imaging technique carries out the cell three-dimensional tracking of internal sparse distribution, intend
Time delay resolution ratio higher is obtained using optical projection tomography.When being needed to each with traditional three-dimensional tracking technique
Between point make three-dimensional space collection, the present invention intends reducing light radiation to sample.
(2) two width orthogonal images are gathered simultaneously in different projection angles using angular multiplexed OPT systems, intend using up to phase
The time of machine frame per second goes positioning and tracking characteristics target.
(3) by cell detection, identification and the invention of track algorithm, studying cell fine structure and morphological feature,
Intercellular multidate information is detected, is the research direction that solution is badly in need of in medical science and drug research field.
Mainly have：(1) instantaneous orthogonal optical projects building for chromatographic imaging system.
Sample is placed on turntable, two orthogonal An imaging arm of focusing to rotary shaft select suitable system unit parameter,
Laser, CCD, Electronic control shutter, optical filter, diaphragm, lens etc., build the instantaneous orthogonal OPT imaging systems of twin shaft.
(2) invention of cell detection and recognizer.
Image enhaucament, denoising and image segmentation, detection and recognizer of the research for cell object are carried out to cell.Profit
With cap transformation, watershed algorithm etc..
(3) invention of track algorithm.
Each target at any given time point target nearest with future time point is connected, by relatively more pre-
Location puts the geometric distance difference with physical location, judges whether to list in the cell trajectory lists, the piecewise linearity of definition it is thin
Born of the same parents track.
(4) measurement in cell morphology characteristic, movement velocity and direction
Behind the track of reconstruct tracking cell, a series of feature is can be evaluated whether：Total path length is that corresponding cell is total
Displacement；Along out-of-alignment movable length；Mobile total displacement is the distance of beginning and end；Different tachometric survey etc..
For example in an example for tracking characteristics, we calculate average tracking speed, and (total tracking time span is divided by the tracking time
Interval) and direction (the two average angle cosine of continuous increment).Completely unordered direction of motion value should be 0, close to folk prescription
The direction value of tropism is close to 1.
Embodiment three：Scheme of the invention
Cell detection and identification：(orthogonal CCD systems obtain a pair of figures to the initial data projected with orthogonal fluorescence
Picture) define the coordinate system tested, X, Y, Z axis.Z is common vertical coordinate, is measured since the bottom line of projected image, with
Rotary shaft is parallel.X is that abscissa is measured since projected image (one of camera imaging) left side, and Y is another orthorhombic phase
The abscissa of machine imaging.By after complete rotation projection acquisition, having k to projected image, by image segmentation and triangulation,
We can determine whether (X, Y, Z)kI.e. fluorescent target is (such as：Cell) in kth to the coordinate of image.
Preferable fluorescent characteristics, such as neutrophil leucocyte of expressing green fluorescent protein should have constant at different time points
Brightness.However, the fluorescence intensity in fact measuring the different anglecs of rotation be subject to light collection efficiency (such as：It is flat from Jiao
The diverse location in face), excitation spatial variations, exciting light and autofluorescence scattering and absorb influence.In addition, these
Neutrophil leucocyte cluster together, largely extends in moving process and changes their shape, and fluorescin
Expression is also change.Therefore, individual cells are detected in xz, yz projected image, the cell size in proper range
Under fluorescence intensity, it is critically important to select suitable image segmentation algorithm.We intend using size is adjustable, non-linear cap letter high
Number image processing algorithm interior identification target in estimated size range, all algorithms run under Matlab environment.By one group
Given pixel and the average gray ratio of its peripheral region pixel are compared, and this can be met using Local size selection threshold value
It is a little to require.Primary and secondary detection yardstick is set to 2 and 8 pixels accordingly.Smoothed by mathematical morphology and size selection
Image segmentation is carried out, the area that can set segmentation object is less than a certain particular value.Due to the motion of cell, iuntercellular is also present
The situation for contacting with each other and overlapping, can be an attempt to be carried out with watershed algorithm the segmentation of such situation.
After image segmentation, (X, Z) (Y, Z) coordinate that cell is identified in each pair projected image regards as the several of cut zone
What center.In projected image pair, if close Z coordinate is considered as same cell.The close standard of balance Z coordinate can
Set with according to experiment experience, such as：± 3 pixels.The Z coordinate of identified cell has the flat of Z coordinate in two width projected images
It is worth to.For the in-depth analysis of next step, the data information memory of each cell is got off.
Tracking cell
The first step of quantitative analysis is exactly the movement locus that reconstruct carrys out the segmentation cell of measurement projected coordinate system.Can be
The target that each target of any given frame is nearest with next frame is connected.Complicated method can use position, brightness, shape
Size, direction, speed and other cell characteristics calculate " similarity ".We can start with research from relatively simple algorithm, first
Consider geometric distance feature.
Measure coordinate of the target in time point k is Xk, Yk, Zk, wherein：Xk、Yk、ZkRespectively target is in time point k
In X-axis, in Y-axis, the measure coordinate on Z axis；Xk+1、Yk+1Respectively target in time point k+1 in X-axis, the survey in Y-axis
Position fixing value；
Prediction coordinate of the target in time point k is P (xk, yk, zk), wherein：P(xk)、P(yk)、P(zk) it is respectively target
In time point k in x-axis, in Y-axis, the measure coordinate value on Z axis；
Coordinate prediction is carried out using formula (1)；
And Pk(zk+1)=Zk+1Formula (1)
P in formula (1)k(zk+1)=Zk+1In Pk(zk+1) it is prediction coordinate of the target in time point k+1 on Z axis
Value；Zk+1It is measure coordinate value of the target in time point k+1 on Z axis；Pk(zk+1)=Zk+1Represent target in time point k+1
When prediction coordinate value on Z axis, it is equal with measure coordinate value of the target in time point k+1 on Z axis；
In formula (1) is dot product symbol
When θ in formula (1) is rotating acquisition in each time point k arbitrary neighborhood collection picture frame angle step,
XAOR, YAORBe the focal plane for projecting to instantaneous orthogonal optical imaging system two orthogonal images in rotary shaft horizontal position
Put；
Connected in each described target of any given frame target nearest with next frame；The institute of all identifications
State coordinate X of the target in k-th detection at time pointk, Yk, ZkList of coordinates transforms to+1 coordinate system P at time point of kth
(xk+1, yk+1, zk+1), P (xk+1, yk+1, zk+1) and in+1 coordinate X of time point detection of kthk+1, Yk+1, Zk+1It is compared, institute
The same target is considered when stating the space length of target in the range of some, each described target can only add one
Track；
P(xk+1, yk+1, zk+1) it is that target is being in k-th coordinate X at time point according to the target for recognizingk, Yk, ZkIn kth
+ 1 prediction coordinate at time point；
If corresponding target is not identified at+1 time point of kth, changed commanders+2 time points of kth using suitable change
The coordinate at+2 time points of prediction coordinate and kth be compared；If corresponding target is not still identified, target
Tracking terminates at k-th time point.
Example IV：First with algorithm for image enhancement, image segmentation algorithm, according to experiment experience (the size model of target
Enclose) target is identified, numbering is marked to multiple targets, the morphological feature of target is entered using feature extraction algorithm
Row is extracted, and three-dimensionalreconstruction is carried out to target trajectory using track algorithm, is finally completed the tracking of multiple target and is obtained target
Many reference amounts information.
Track algorithm, in initial time point k, defines initial cell aggregation, and ensuing track algorithm is applied to kth
Individual time point ,+1 time point of kth, and each cell in cell aggregation.To each cell in cell collection, it is calculated
In the k+1 prediction coordinate at time point.If in the vicinity of predicted position, finding a neighbouring identification cell (space length
Less than a certain particular value), track is added, algorithm continues next iteration.One cell is only possible in a track, is only listed in
Once.If in the m times iteration, listing track in and not succeeding, predict in the m+2 cell position at time point, it is therefore desirable to two
Secondary continuous prediction.As long as not succeeding twice, tracking is cut off, and begins setting up next tracking, until k-th institute at time point
There is initial cell to track to terminate.
Track algorithm flow chart is as shown in Figure 2.In order to gather tracking data as much as possible, algorithm continuously starts to use
One, the second grade frame is used as most initial frame.In next step analysis, calculate XY increments and distinguish measurement coordinate and prediction coordinate per frame
Difference.Increment is represented in sample coordinate system, cell track is finally defined.
Embodiment five：Design to optical projection tomography system (OPT) and preferable Research foundation, opposite are built
The treatment of thing sample image proposes full and accurate feasible study route and method with recognizer.
Laboratory has very complete experiment condition：It is 488nm lasers (200mw), 561nm lasers (200mw), many
Individual spectrometer, CCD, photomultiplier, detector, electricity driving displacement platform, controller, power meter, oscillograph, fluorescence microscope system
System, work station etc.；And preliminary OPT imaging systems have been built, this provides solid material base for work development in future
Plinth, has substantially ensured that smoothly completing for invention.
(1) propose to carry out fluorescent target using instantaneous orthogonal OPT imaging systems the new approaches of three-dimensional tracking, realize
The reconstruct of neutrophil leucocyte movement locus in complete live body model animal, for the quantitative description of cell movement provides one kind newly
Method.
(2) propose based on cell detection and recognizers such as cap algorithm high, watershed algorithms, to cell morphology characteristic etc.
The extraction of many reference amounts information, can accurately be described to the cell state under disease development and medicine effect.
(3) compound angle projection imaging method is proposed, the delay by successive projection image is realized rather than whole volume
The time delay resolution ratio that acquisition time is determined, substantially reduces the light radiation to biological living, is that biological intravital cell grinds
Study carefully there is provided possible ways.
A kind of cell image is built by the present invention and tracks intelligent algorithm, can be to the immunocyte in live body model animal
Motion carry out three-dimensional tracking and extract the various morphological features of cell, be that study of disease development, vital movement and medicine are made
With there is provided new technical method, it is with a wide range of applications.
Tracking cell intelligent algorithm has the beneficial effect that：By development of the invention, instantaneous orthogonal OPT imagings are built
System, realizes the three-dimensional tracking to neutrophil leucocyte in zebrafish embryo.Picking rate reaches cameras frame rate, to biological specimen
Light radiation is small.Various video procession algorithms are used for the detection of neutrophil leucocyte and identification in zebrafish embryo, are obtained
The change in location and morphological feature of cell.For next step data analysis lays the foundation.Analysis neutrophil leucocyte structure, morphology
Feature, movement locus and inflammation and pharmaceutically-active correlation, are that study of disease change and drug action mechanism establish experiment with reason
By basis.
The preferred embodiments of the invention is the foregoing is only, claims of the invention are not limited to.
Simultaneously it is described above, for those skilled in the technology concerned it would be appreciated that and implement, therefore other be based on institute of the present invention
The equivalent change that disclosure is completed, should be included in the covering scope of the claims.
Claims (1)
1. a kind of cell image tracks intelligent algorithm, it is characterised in that comprising following：
Step one, IMAQ：
Zebrafish embryo is placed and starts rotation on a spinstand, while rotation, by instantaneous orthogonal optical imaging system
IMAQ is carried out to cell object in the zebrafish embryo；By obtaining projected image after complete described image collection
It is right, to the projected image to carrying out image enhaucament, image denoising and image segmentation, sieved with size by the way that mathematical morphology is smooth
Choosing carries out described image segmentation, obtains multiple neutrophil leucocyte images, by tracking the multiple neutrophil leucocyte image in
The movement locus of all neutrophil leucocytes, the area of the multiple neutrophil leucocyte image of setting segmentation is less than a particular value；And
Carry out splitting the situation for contacting with each other and overlapping with watershed algorithm；
Gathering in the multiple neutrophil leucocyte image has the coordinate bit that record all neutrophil leucocytes described in current time
Put；It is separated by two field pictures shooting time when gathering the multiple neutrophil leucocyte image at intervals of a set time；
Step 2, track algorithm：
The multiple neutrophil leucocyte image is analyzed, and using the one of neutrophil leucocyte in the object for analyzing as mesh
Mark；
Any given time point k is time point k+1 by the future time point after the set time；The time point k+
1 is time point k+2 by the future time point after the set time；
Target during by the time point k is connected with the target nearest in the time point k+1, is led to
The geometric distance difference of comparison prediction position and physical location is crossed, judges whether to list the trajectory lists of the target in；
The measurement in step 3, cell morphology characteristic, movement velocity and direction：
Behind the track of the reconstruct tracking target, following characteristics are calculated：
The total length in the track of the target, is the total displacement of one neutrophil leucocyte；
The track total displacement of the target, is the distance of the beginning and end of one neutrophil leucocyte；
The average speed value and the direction of motion of the target；
Tracking time interval；
According to the object judgement direction of motion：From the target in the time point k, during the time point k+1, when described
Between point k+2 when three positions carry out line after, calculate the included angle cosine value of two neighboring displacement increment, obtain the direction of motion；
If 1 direction of motion direction value of the included angle cosine value is 1；The referred to as unidirectional direction of motion；The included angle cosine
If 0 direction of motion direction value of value is 0；The referred to as complete randomness direction of motion；
The target near apart from wound can be affected be with the unidirectional direction of motion, apart from wound it is remote described in
Target is unaffected being exactly the complete randomness direction of motion, for judging whether the target can be towards the zebra
The wound movement of fish embryo；
Step 4, cell detection and identification：
The X-axis of definition measurement projected coordinate system, Y-axis, Z axis；The Z axis are vertical coordinates；The X-axis is abscissa from described many
The left side of individual neutrophil leucocyte image starts measurement, and the Y-axis is another abscissa, orthogonal with the Y-axis；
The projected image in projected image and YZ axles that the projected image is divided on XZ axles with the Z axis；
The projected image in projected image and the YZ axles in each described set time on the XZ axles is respectively gathered
Piece image, to reduce light radiation when three-dimensional tracking is carried out to the zebrafish embryo, and between 0 to the time point k
Time period in obtain k projected image pair, by image segmentation and triangulation, the target is in the time point k
Measure coordinate is Xk, Yk, Zk, wherein：Xk、Yk、ZkRespectively described target in the time point k in the X-axis, the Y
On axle, the measure coordinate on the Z axis；Xk+1、Yk+1Respectively described target in the time point k+1 in the X-axis, institute
State the measure coordinate value in Y-axis；
Prediction coordinate of the target in the time point k is P (xk, yk, zk), wherein：P(xk)、P(yk)、P(zk) be respectively
The target in the time point k in the X-axis, in the Y-axis, the measure coordinate value on the Z axis；
Coordinate prediction is carried out using formula (1)；
And Pk(zk+1)=Zk+1Formula (1)
P in formula (1)k(zk+1)=Zk+1In Pk(zk+1) be the target in the time point k+1 on the Z axis
Prediction coordinate value；Zk+1It is measure coordinate value of the target in the time point k+1 on the Z axis；Pk(zk+1)=Zk+1
Prediction coordinate value of the target in the time point k+1 on the Z axis is represented, with the target in the time point k+
Measure coordinate value when 1 on the Z axis is equal；
In formula (1) is dot product symbol
When θ in formula (1) is rotating acquisition in each described time point k arbitrary neighborhood collection picture frame angle step,
XAOR, YAORIt is the coordinate position of XY axle of the rotary shaft in projected coordinate system；
Projected image in projected image and the YZ axles on the XZ axles detects the single target, using big I
Adjust, non-linear cap functional image Processing Algorithm high is identified the target；
The effective coverage of the single target is judged according to multiple target sizes empirical value segmentations, (fitting is exactly handle for fitting
A series of point in plane, is coupled together with a smooth curve) geometric center of multiple targets is calculated, sat in XZ
The coordinate for marking system is (x, z), is (y, z) in the coordinate of YZ coordinate systems；
When the target in the k to projected image pair is identified, if close Z coordinate value is then considered same
Target described in one；There is close Z coordinate value in the presence of 2 or more the targets in XZ coordinate systems and YZ coordinate systems, with institute
Turntable rotation is stated, if the different targets, the change of the Z coordinate value in ensuing image is unsatisfactory for the border of setting
Condition is so as to exclude；
The movement locus of the segmentation target of step 5, reconstruct from the measurement projected coordinate system：
Connected in each described target of any given frame target nearest with next frame；The mesh of all identifications
It is marked on the coordinate X of k-th detection at time pointk, Yk, ZkList of coordinates transforms to+1 coordinate system P (x at time point of kthk+1,
yk+1, zk+1), P (xk+1, yk+1, zk+1) and in+1 coordinate X of time point detection of kthk+1, Yk+1, Zk+1It is compared, the mesh
The same target is considered when target space length is in the range of some, each described target can only add a rail
Mark；
P(xk+1, yk+1, zk+1) it is that the target is being in k-th coordinate X at time point according to the target for recognizingk, Yk, Zk
In+1 prediction coordinate at time point of kth；
If the corresponding target is not identified at+1 time point of kth, changed commanders+2 time points of kth using suitable change
The coordinate at+2 time points of prediction coordinate and kth be compared；If the corresponding target is not still identified, institute
The tracking for stating target terminates at k-th time point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611186286.2A CN106846296A (en) | 2016-12-19 | 2016-12-19 | A kind of cell image tracks intelligent algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611186286.2A CN106846296A (en) | 2016-12-19 | 2016-12-19 | A kind of cell image tracks intelligent algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106846296A true CN106846296A (en) | 2017-06-13 |
Family
ID=59140822
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611186286.2A Pending CN106846296A (en) | 2016-12-19 | 2016-12-19 | A kind of cell image tracks intelligent algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106846296A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107944360A (en) * | 2017-11-13 | 2018-04-20 | 中国科学院深圳先进技术研究院 | A kind of induced multi-potent stem cell recognition methods, system and electronic equipment |
| CN111474149A (en) * | 2020-04-10 | 2020-07-31 | 复旦大学附属中山医院 | Dynamic assessment method for mitochondria |
| CN113192553A (en) * | 2020-01-14 | 2021-07-30 | 北京大学 | Method for predicting cell spatial relationship based on single cell transcriptome sequencing data |
| CN115018886A (en) * | 2022-08-09 | 2022-09-06 | 腾讯科技（深圳）有限公司 | Motion trajectory identification method, device, equipment and medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101144784A (en) * | 2007-09-04 | 2008-03-19 | 杭州电子科技大学 | Automatic tracking method for video frequency microscopic image cell |
| CN101294953A (en) * | 2008-06-05 | 2008-10-29 | 中国农业大学 | Motor cell real-time tracing system and method |
| CN106248643A (en) * | 2016-07-15 | 2016-12-21 | 陈玲玲 | A kind of system of the scanning laser optics chromatography imaging method of high spectral resolution |
-
2016
- 2016-12-19 CN CN201611186286.2A patent/CN106846296A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101144784A (en) * | 2007-09-04 | 2008-03-19 | 杭州电子科技大学 | Automatic tracking method for video frequency microscopic image cell |
| CN101294953A (en) * | 2008-06-05 | 2008-10-29 | 中国农业大学 | Motor cell real-time tracing system and method |
| CN106248643A (en) * | 2016-07-15 | 2016-12-21 | 陈玲玲 | A kind of system of the scanning laser optics chromatography imaging method of high spectral resolution |
Non-Patent Citations (1)
| Title |
|---|
| LINGLING CHEN, YURIY ALEXANDROV.ET.AL.: ""Mesoscopic in vivo 3-D tracking of sparse cell populations using angular multiplexed optical projection tomography"", 《BIOMEDICAL OPTICS EXPRESS》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107944360A (en) * | 2017-11-13 | 2018-04-20 | 中国科学院深圳先进技术研究院 | A kind of induced multi-potent stem cell recognition methods, system and electronic equipment |
| CN113192553A (en) * | 2020-01-14 | 2021-07-30 | 北京大学 | Method for predicting cell spatial relationship based on single cell transcriptome sequencing data |
| CN113192553B (en) * | 2020-01-14 | 2022-09-09 | 北京大学 | Method for predicting cell spatial relationship based on single cell transcriptome sequencing data |
| CN111474149A (en) * | 2020-04-10 | 2020-07-31 | 复旦大学附属中山医院 | Dynamic assessment method for mitochondria |
| CN111474149B (en) * | 2020-04-10 | 2023-08-08 | 复旦大学附属中山医院 | Dynamic evaluation method for mitochondria |
| CN115018886A (en) * | 2022-08-09 | 2022-09-06 | 腾讯科技（深圳）有限公司 | Motion trajectory identification method, device, equipment and medium |
| CN115018886B (en) * | 2022-08-09 | 2022-11-08 | 腾讯科技（深圳）有限公司 | Motion trajectory identification method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6732807B2 (en) | Biopsy specimen fluorescence imaging apparatus and method | |
| CN102488493B (en) | Small animal living body multi-mode molecule imaging system and imaging method | |
| CN106846296A (en) | A kind of cell image tracks intelligent algorithm | |
| JP2005539209A (en) | Optical projection imaging system and method for automatically detecting nucleated cells and cytoplasm density features involved in disease | |
| US20180042483A1 (en) | Systems and methods for hyperspectral imaging | |
| Yang et al. | 3-D reconstruction of microtubules from multi-angle total internal reflection fluorescence microscopy using Bayesian framework | |
| Zhao et al. | DSU-Net: Distraction-Sensitive U-Net for 3D lung tumor segmentation | |
| CN101947103A (en) | Optical bioluminescence tomography method | |
| CN106805944A (en) | A kind of instantaneous orthogonal optical projection chromatographic imaging system of high accuracy | |
| Cai et al. | One stage lesion detection based on 3D context convolutional neural networks | |
| Belykh et al. | Confocal laser endomicroscopy assessment of pituitary tumor microstructure: a feasibility study | |
| US11922623B2 (en) | Cellular diagnostic and analysis methods | |
| CN201641974U (en) | Multi-angle imaging device | |
| CN105662354B (en) | A kind of wide viewing angle optical molecular tomographic navigation system and method | |
| TW201819913A (en) | Magnetophorisis measuring system for determining moving status of object and quantifying amount of magnetic particles contained therein | |
| CN206603773U (en) | The burn degree and area detection system measured based on blood erythrocyte concentration | |
| CN201641971U (en) | Multi-angle imaging device | |
| CN114581553B (en) | Fluorescent molecular tomography reconstruction method based on magnetic particle imaging prior guidance | |
| CN209172280U (en) | Multi-modality imaging system | |
| US20230230398A1 (en) | Image processing device, image processing method, image processing program, and diagnosis support system | |
| Wang et al. | Improved Quantitative Analysis Method for Magnetic Particle Imaging Based on Deblurring and Region Scalable Fitting | |
| CN103750824B (en) | A kind of information extracting method for toy fluorescent chromatographic imaging system | |
| Bhaskar et al. | Live cell imaging: a computational perspective | |
| CN205795645U (en) | A kind of wide viewing angle optical molecular tomographic navigation system | |
| TWI838687B (en) | Method for evaluating immune status of tumor samples |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170613 |
|
| WD01 | Invention patent application deemed withdrawn after publication |