CN106373116A - Two-photon image-based synapse detection method - Google Patents

Abstract

The invention discloses a two-photon image-based synapse detection method, and relates to the technical field of mode identification and neurology. The method comprises the steps of obtaining a two-photon image; dividing the two-photon image into an axon image and a dendrite image; performing 2D detection of axon noduli in the axon image; performing 2D detection of dendrite spines in the dendrite image; performing 2D synapse detection by using the detected axon noduli and dendrite spines; and performing 3D synapse detection based on a 2D synapse detection result. Through the method, the synapse detection efficiency and precision are improved; and the method has universality and is not sensitive to image types.

Description

Synapse detection method based on two-photon image

Technical field

The present invention relates to pattern recognition and neurological technical field, especially relate to a kind of synapse based on two-photon image Detection method.

Background technology

Numerous studies show, the synaptic plasticity of neuron includes function plasticity and structure plasticity, with study and note Recall closely related.Synaptic plasticity dysregulation is often thought of as cognition, the handicapped main cause of learning and memory.Permitted How spirit is closely related with the exception that synaptic plasticity is adjusted with cognitive disorder pathogenesis, the research that synaptic plasticity is adjusted Be conducive to disclosing intelligent development sluggishness, the pathomechanism of this kind of neuropsychiatric disease of cognitive dysfunction.

In recent years, using just putting Two Photon Fluorescence system and fluorescent probe labelling technique method always is the heat of research Point.Existing detection synapse method is it is simply that neurosurgeon passes through naked eyes searching synapse one by one.The method expends artificial in a large number Time, and precision is not high, situation about omitting easily.

In view of this, the special proposition present invention.

It is below prior art related to the present invention:

[1]n.dey,l.blanc-feraud,c.zimmer,“richardson–lucy algorithm with total variation regularization for 3d confocal microscope deconvolution,” microscopy research and technique,vol.69,no.4,pp.260-266,2006.

[2]j.e.ledoux,“emotion circuits in the brain,”annual review of neuroscience,vol.23,no.2,pp.155-184,2000.

[3]v.d.paola,a.holtmaat,g.knott,et al.,“cell type-specific structural plasticity of axonal branches and boutons in the adult neocortex,”neuron, vol.49,no.6,pp.861-875,2006.

[4]n.becker,c.j.wierenga,r.fonseca,et al.,“ltd induction causes morphological changes of presynaptic boutons and reduces their contacts with spines,”neuron,vol.60,no.4,pp.590-597,2008.

[5]f.jing,x.zhou,j.g.dy,et al.,“an automated pipeline for dendrite spine detection and tracking of 3d optical microscopy neuron images of in vivo mouse models,”neuroinformatics,vol.7,no.2,pp.113-130,2009.

[6]h.park and m.m.poo,“neurotrophin regulation of neural circuit development and function,”nature reviews neuroscience,vol.14,no.1,pp.7-23, 2013.

[7]f.w.grillo,s.song,l.m.teles-grilo,et al.,“increased axonal bouton dynamics in the aging mouse cortex,”proceedings of the national academy of Sciences of the united states of america, vol.110, no.16, pp.1514-1523,2013.

[8]y.yang,d.q.liu,w.huang,et al.,“selective synaptic remodeling of amygdalocortical connections associated with fear memory,”neuroscience,to be published.

Content of the invention

The embodiment of the present invention provides a kind of synapse detection method based on two-photon image, prominent so that how solution improves detection Tactile efficiency and the problem of precision.

To achieve these goals, provide technical scheme below:

A kind of synapse detection method based on two-photon image, methods described includes:

Obtain described two-photon image；

Described two-photon image is divided into aixs cylinder image and dendron image；

Carry out the 2d detection of aixs cylinder brief summary in described aixs cylinder image；

Carry out the 2d detection of dendritic spine in described dendron image；

Carry out 2d synapse detection with the aixs cylinder brief summary and dendritic spine that detect；

Based on 2d synapse testing result, carry out 3d synapse detection.

Further, the described 2d detection carrying out aixs cylinder brief summary in described aixs cylinder image, specifically includes:

The 3d deconvolution that described aixs cylinder image is iterated is processed, and obtains deconvolution image；

Determine the Local modulus maxima in described deconvolution image；

Based on described aixs cylinder image, determine and strengthen image；

Binary segmentation process is carried out to described enhancing image；

Based on described Local modulus maxima and binary segmentation result, remove in described binary segmentation result not Region including described Local modulus maxima.

Further, the described Local modulus maxima determining in described deconvolution image specifically includes:

Set first threshold；

Find out the point that pixel value in described deconvolution image is more than the doubtful local maximum of first threshold；

Judge that the described the pixel value whether point of described each doubtful local maximum meets in a predetermined neighborhood is maximum , if it is, determining that the point of described doubtful local maximum is described Local modulus maxima.

Further, described based on described aixs cylinder image, determine that strengthening image specifically includes:

On the basis of two-dimentional xy coordinate system, based on described aixs cylinder image, described enhancing image is determined according to below equation:

$g=-(\frac{f_{xx}\×f_1^2}{1+f_1^2}-2\frac{f_{xy}\×f_1}{1+f_1^2}+\frac{f_{yy}}{1+f_1^2})$

$f_1=\frac{2f_{xy}}{f_{xx}-f_{yy}+\mathrm{sgn}(f_{xx}+f_{yy})\sqrt{{(f_{xx}-f_{yy})}^2+4{f_{xy}}^2}}$

Wherein, described g represents described enhancing image；Described f_xRepresent described aixs cylinder image local derviation in the x direction；Described f_yRepresent described aixs cylinder image local derviation in y-direction；Described f_xxRepresent described f_xLocal derviation on described x direction；Described f_xy Represent described f_xLocal derviation in said y direction；Described f_yyRepresent described f_yLocal derviation in said y direction.

Further, methods described includes: the described 2d detection carrying out dendritic spine in described dendron image specifically includes:

Described dendron image is normalized and structure enhancement process；

Dendron in skeletonizing structure enhancement process result, obtains dendron skeleton binary map；

Based on described dendron skeleton binary map, determine the bifurcation on dendron skeleton；

Based on described dendron skeleton binary map, and according to described bifurcation, position described dendritic spine.

Further, described described dendron image is normalized and structure enhancement process specifically includes:

By described dendron image normalization between 0 and 1；

According to below equation, set dendritic spine INTERFACE MODEL:

$f(x,y)=\exp (-\frac{{(x\sin \mathrm{\θ}+ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2})$

Wherein, described θ represents the angle of described dendritic spine；Described σ represents the width of described dendritic spine；Described f (x, y) table Show described dendritic spine INTERFACE MODEL；Described x, described y represent the horizontal stroke of pixel, vertical coordinate in the image after normalization respectively；

Line structure according to strengthening image after normalization with drag:

$\mathrm{\λ}(x,y)=-\frac{1}{{\mathrm{\σ}}^2}\exp (-\frac{{(xsin\mathrm{\θ}+ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2}).$

Further, the dendron in described skeletonizing structure enhancement process result, obtains dendron skeleton binary map and specifically wraps Include:

Row threshold division is entered to structure enhancement process result, obtains dendron bianry image；

Travel through described dendron bianry image, pixel is deleted according to following condition, until described dendron bianry image is no longer Till change, obtain the first bianry image:

x_h(p)=1；2≤min{n₁(p),n₂(p)}≤3；

Wherein, described p represents pixel；Described x1, x2 ..., x8 is the institute starting from right side adjoint point by inverse time needle sort State 8 consecutive points pixels of p；

Travel through described first bianry image, pixel is deleted according to following condition, until described dendron bianry image is no longer Till change, obtain described dendron skeleton binary map:

x_h(q)=1；2≤min{n₁(q),n₂(q)}≤3；

Wherein, described q represents pixel；Described x₁,x₂,...,x₈It is the institute starting from right side adjoint point by inverse time needle sort State 8 consecutive points pixels of q.

Further, described based on described dendron skeleton binary map, determine that the bifurcation on dendron skeleton specifically includes:

Construction wave filter；

Described wave filter and described dendron skeleton binary map are carried out process of convolution, and Second Threshold will be more than in trellis diagram Point be defined as bifurcation.

Further, described described dendritic spine and according to described bifurcation, are positioned based on described dendron skeleton binary map, Specifically include:

The structural element of construction predetermined radii, and morphological erosion is carried out to described dendron skeleton binary map；

Deduct the skeleton drawing after corrosion with described dendron skeleton binary map, obtain dendritic spine alternate location；

Based on described dendron skeleton binary map, centered on described bifurcation, extract the region of the predetermined length of side；

Choose the skeleton part of described bifurcation region, and judge whether described dendritic spine alternate location and be somebody's turn to do Skeleton is partly connected；If there are it is determined that described dendritic spine alternate location is dendritic spine.

Further, the described aixs cylinder brief summary with detecting and dendritic spine carry out 2d synapse detection and specifically include:

Represent described aixs cylinder brief summary with the central point of each described aixs cylinder brief summary；

Represent described dendritic spine with the central point of described dendritic spine alternate location；

Calculate and record the minima of distance between described each aixs cylinder brief summary central point and all described dendritic spine central points；

If the minima of described distance is less than the 3rd threshold value, described aixs cylinder brief summary and described dendritic spine is defined as 2d and dashes forward Touch.

Further, described based on 2d synapse testing result, carry out 3d synapse detection specifically include:

Represented described prominent with the midpoint of aixs cylinder brief summary central point and dendritic spine alternate location central point in described 2d synapse Tactile position；

With the distance of aixs cylinder brief summary central point described in described 2d synapse and described dendritic spine alternate location central point come generation The verity of synapse described in table；

Coordinate in layer before and after the section of described two-photon image sequence is adjacent, in the neighborhood of described 2d synoptic sites The described synapse of interior lookup；

Judge whether in verity layer before and after the section of described two-photon image sequence is adjacent of each described synapse be High, if it is determined that described synapse is 3d synapse.

The embodiment of the present invention provides a kind of synapse detection method based on two-photon image.By two-photon image is divided into Aixs cylinder image and dendron image, detection aixs cylinder brief summary and dendritic spine, to find doubtful synapse for neurosurgeon's screening, are used respectively The aixs cylinder brief summary detecting and dendritic spine carry out 2d synapse detection；It is then based on 2d synapse testing result, carry out 3d synapse detection, So greatly increase efficiency and the precision of detection synapse, also there is universality, and insensitive to image type.

Brief description

Fig. 1 is the synapse detection method schematic flow sheet according to the embodiment of the present invention based on two-photon image；

Fig. 2 is based on dendron bianry image according to the embodiment of the present invention, deletes the schematic diagram of pixel；

Fig. 3 is the filter schematic according to the embodiment of the present invention；

Fig. 4 is the two-photon image sequence section interlayer relation schematic diagram according to the embodiment of the present invention.

Specific embodiment

For making the object, technical solutions and advantages of the present invention become more apparent, below in conjunction with the drawings and specific embodiments, The present invention is described in more detail.

The embodiment of the present invention provides a kind of synapse detection method based on two-photon image.As shown in figure 1, the method is permissible Including:

S100: obtain two-photon image.

For example: this image can be projection remote to Mus brain corpus amygdaloideum come aixs cylinder and auditory cortex dendron carry out glimmering The image of signal.

S110: two-photon image is divided into aixs cylinder image and dendron image.

Wherein, in two-photon image, aixs cylinder and dendron are to be shown with different colors.By extracting corresponding color Part can therefrom extract aixs cylinder or dendron image.

S120: carry out the 2d detection of aixs cylinder brief summary in aixs cylinder image.

This step may include that further

S121: the 3d deconvolution that aixs cylinder image is iterated is processed, and obtains deconvolution image.

In actual applications, this step can be by the plug-in unit diffraction psf 3d in imagej instrument and plug-in unit Iterative deconvolution 3d.class is realizing.Wherein, run plug-in unit diffraction psf 3d, input phase Related parameter, generates a convolution kernel.Preferably, |input paramete and selection range: medium refraction index can be come according to following configuration: Usual range (1.3-1.5)；Objective aperture: usual range (1.0-1.35)；Wavelength: usual range (460-500nm), most preferably Ground, is 488；Plane picture resolution: about 200nm；Z-axis resolution: about 700nm；Wide, high, the number of plies of image are followed successively by 512, 512,140；Longitudinal spherical aberration maximum diameter of hole: usual range (120-870), most preferably, is 500-700nm.Select former simultaneously Figure and the convolution kernel generating, and input deconvolution number of times (preferably 10 times).Then run plug-in unit iterative Deconvolution 3d.class calculates the image of deconvolution.

S122: determine the Local modulus maxima in deconvolution image.

Specifically, this step may include that

S1221: preset a first threshold (for example: 1000).

S1222: find out the point that pixel value in deconvolution image is more than the doubtful local maximum of first threshold.

S1223: judge that the pixel value whether point of each doubtful local maximum meets in a predetermined neighborhood is maximum, If it is, execution step s1224；Otherwise execution step s1225.

S1224: determine this doubtful local maximum point be Local modulus maxima.

S1225: the point determining this doubtful local maximum is not Local modulus maxima.

S123: based on aixs cylinder image, determine and strengthen image.

Wherein, this step may include that

On the basis of two-dimentional xy coordinate system, based on aixs cylinder image, determined according to below equation and strengthen image:

$g=-(\frac{f_{xx}\×f_1^2}{1+f_1^2}-2\frac{f_{xy}\×f_1}{1+f_1^2}+\frac{f_{yy}}{1+f_1^2})$

$f_1=\frac{2f_{xy}}{f_{xx}-f_{yy}+\mathrm{sgn}(f_{xx}+f_{yy})\sqrt{{(f_{xx}-f_{yy})}^2+4{f_{xy}}^2}}$

Wherein, g represents enhancing image；f_xRepresent aixs cylinder image local derviation in the x direction；f_yRepresent aixs cylinder image in y direction On local derviation；f_xxRepresent f_xLocal derviation in the x direction；f_xyRepresent f_xLocal derviation in y-direction；f_yyRepresent f_yIn y-direction Local derviation.

S124: carry out binary segmentation process to strengthening image.

S125: based on Local modulus maxima and binary segmentation result, remove in binary segmentation result and do not wrap Include the region of Local modulus maxima.

S130: carry out the 2d detection of dendritic spine in dendron image.

This step may include that

S131: dendron image is normalized and structure enhancement process.

Specifically, this step may include that

S1311: by dendron image normalization to [0,1].

S1312: according to below equation, set dendritic spine INTERFACE MODEL:

$f(x,y)=\exp (-\frac{{(x\sin \mathrm{\θ}+ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2})$

Wherein, θ represents the angle of dendritic spine；σ represents the width of dendritic spine；F (x, y) represents dendritic spine INTERFACE MODEL；x、y Represent the horizontal stroke of pixel, vertical coordinate in the image after normalization respectively.

S1313: the line structure according to strengthening image after normalization with drag:

$\mathrm{\λ}(x,y)=-\frac{1}{{\mathrm{\σ}}^2}\exp (-\frac{{(xsin\mathrm{\θ}+ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2})$

Wherein, the implication of the parameter being related in this step, referring to the relevant explanation of step s1312, will not be described here.

S132: the dendron in skeletonizing structure enhancement process result, obtains dendron skeleton binary map.

This step may include that

S1321: row threshold division is entered to structure enhancement process result, obtains dendron bianry image.

S1322: traversal dendron bianry image, deletes pixel according to following condition, until dendron bianry image no longer becomes Only turn to, obtain the first bianry image:

x_h(p)=1；2≤min{n₁(p),n₂(p)}≤3；

Wherein,

Wherein, p represents pixel；x₁,x₂,...,x₈Be start from right side adjoint point 8 of p by inverse time needle sort adjacent Point pixel.

In this step, x_hP () is one and discriminates whether to delete the condition of pixel p.b_iDetermine x_hThe value of (p).For example: If meeting x₁=0 and (x₂=1 or x₃=1), then b₁=1.And then draw x_hThe value of (p).

x₁,x₂,...,x₈It is 8 consecutive points pixels starting the p by inverse time needle sort from right side adjoint point, as Fig. 2 institute Show.

$n_1\left(p\right)=\underset{i=1}{\overset{4}{\σ}}{x}_{2k-1}\∪x_{2k},n_2\left(p\right)=\underset{i=1}{\overset{4}{\σ}}{x}_{2k}\∪x_{2k+1}.$

S1323: traversal the first bianry image, deletes pixel according to following condition, until dendron bianry image no longer becomes Only turn to, obtain dendron skeleton binary map:

x_h(q)=1；2≤min{n₁(q),n₂(q)}≤3；

Wherein,

Wherein, q represents pixel；x₁,x₂,...,x₈Be start from right side adjoint point 8 of q by inverse time needle sort adjacent Point pixel.

The explanation of the parameter about being related in this step may refer to the explanation of step s1322, will not be described here.

S133: based on dendron skeleton binary map, determine the bifurcation on dendron skeleton.

This step specifically may include that

S1331: construction wave filter.

For example: construction size is 3 × 3 wave filter, as shown in Figure 3.Wherein, 4 summit values are 0, and remaining puts value For 1.

S1332: wave filter and dendron skeleton binary map are carried out process of convolution, and Second Threshold will be more than in trellis diagram Point is defined as bifurcation.

For example, it is possible to the point that trellis diagram intermediate value is more than or equal to 4 is defined as bifurcation.

S134: based on dendron skeleton binary map, and according to bifurcation, position dendritic spine.

Specifically, this step may include that

S1341: the structural element of construction predetermined radii, and morphological erosion is carried out to dendron skeleton binary map.

In this step, predetermined radii can determine according to the length of dendritic spine (its account for several pixels, such as 3).

S1342: deduct the skeleton drawing after corrosion with dendron skeleton binary map, obtain dendritic spine alternate location.

S1343: based on dendron skeleton binary map, centered on bifurcation, extract the region of the predetermined length of side.

Wherein, the predetermined length of side can be determined according to the length of dendritic spine (it accounts for several pixels, example, 31) it is preferable that its 2 times of dendritic spine length can be more than.

S1344: choose the skeleton part of bifurcation region, and judge whether dendritic spine alternate location and this bone Frame is partly connected；If it has, then execution step s1345；Otherwise, execution step s1346.

S1345: determine that dendritic spine alternate location is dendritic spine.

S1346: determine that dendritic spine alternate location is not dendritic spine.

S140: carry out 2d synapse detection with the aixs cylinder brief summary and dendritic spine that detect.

This step may include that further

S141: represent this aixs cylinder brief summary with the central point of each aixs cylinder brief summary.

S142: represent dendritic spine with the central point of dendritic spine alternate location.

S143: calculate and record minima d of distance between each aixs cylinder brief summary central point and all dendritic spine central points.

S144: if this is less than the 3rd threshold value apart from minima d, aixs cylinder brief summary and dendritic spine are defined as 2d synapse.

If this is more than or equal to the 3rd threshold value apart from minima d, aixs cylinder brief summary and dendritic spine are not defined as 2d synapse.

S150: based on 2d synapse testing result, carry out 3d synapse detection.

This step may include that

S151: represent synapse with the midpoint of aixs cylinder brief summary central point in 2d synapse and dendritic spine alternate location central point Position.

S152: represented prominent with the distance of aixs cylinder brief summary central point and dendritic spine alternate location central point in above-mentioned 2d synapse Tactile verity.

Between aixs cylinder brief summary central point and all dendritic spine central points, the minima of distance is less, and verity is higher.

S153: in layer before and after the section of two-photon image sequence is adjacent, in the coordinate in the neighborhood of 2d synoptic sites Search synapse.

S154: judge whether in verity layer before and after the section of two-photon image sequence is adjacent of each synapse be High, if so, then execution step s155；Otherwise, execution step s156.

S155: determine that this synapse is 3d synapse.

S156: determine that this synapse is not 3d synapse.

Fig. 4 schematically illustrates two-photon image sequence section interlayer relation.Wherein, r represents two-photon image sequence Section；R-1, r+1 represent the adjacent layer in front and back of two-photon image sequence section respectively.

The section of two-photon image sequence carries out 2d segmentation, then 2d segmentation result is stringed together on 3d, find 3d Synapse.

Embodiment in above-described embodiment can be further combined or replace, and embodiment is only to the present invention's Preferred embodiment is described, and not the spirit and scope of the present invention is defined, without departing from design philosophy of the present invention Under the premise of, the various changes and modifications that professional and technical personnel in the art make to technical scheme, belong to this Bright protection domain.

Claims (11)

1. a kind of synapse detection method based on two-photon image is it is characterised in that methods described includes:

Obtain described two-photon image；

Described two-photon image is divided into aixs cylinder image and dendron image；

Carry out the 2d detection of aixs cylinder brief summary in described aixs cylinder image；

Carry out the 2d detection of dendritic spine in described dendron image；

Carry out 2d synapse detection with the aixs cylinder brief summary and dendritic spine that detect；

Based on 2d synapse testing result, carry out 3d synapse detection.

2. method according to claim 1 is it is characterised in that the described 2d carrying out aixs cylinder brief summary in described aixs cylinder image Detection, specifically includes:

The 3d deconvolution that described aixs cylinder image is iterated is processed, and obtains deconvolution image；

Determine the Local modulus maxima in described deconvolution image；

Based on described aixs cylinder image, determine and strengthen image；

Binary segmentation process is carried out to described enhancing image；

Based on described Local modulus maxima and binary segmentation result, remove in described binary segmentation result and do not include The region of described Local modulus maxima.

3. method according to claim 2 is it is characterised in that local maximum in the described deconvolution image of described determination Point specifically includes:

Set first threshold；

Find out the point that pixel value in described deconvolution image is more than the doubtful local maximum of first threshold；

Judge that the described the pixel value whether point of described each doubtful local maximum meets in a predetermined neighborhood is maximum, such as Fruit is it is determined that the point of described doubtful local maximum is described Local modulus maxima.

4. method according to claim 2 it is characterised in that described based on described aixs cylinder image, determine and strengthen image tool Body includes:

On the basis of two-dimentional xy coordinate system, based on described aixs cylinder image, described enhancing image is determined according to below equation:

$g=-(\frac{f_{xx}\×f_1^2}{1+f_1^2}-2\frac{f_{xy}\×f_1}{1+f_1^2}+\frac{f_{yy}}{1+f_1^2})$

$f_1=\frac{2f_{xy}}{f_{xx}-f_{yy}+\mathrm{sgn}\left(f_{xx}+f_{yy}\right)\sqrt{{\left(f_{xx}-f_{yy}\right)}^2+4{f_{xy}}^2}}$

Wherein, described g represents described enhancing image；Described f_xRepresent described aixs cylinder image local derviation in the x direction；Described f_yTable Show described aixs cylinder image local derviation in y-direction；Described f_xxRepresent described f_xLocal derviation on described x direction；Described f_xyRepresent Described f_xLocal derviation in said y direction；Described f_yyRepresent described f_yLocal derviation in said y direction.

5. method according to claim 1 is it is characterised in that methods described includes: described enters in described dendron image The 2d detection of row dendritic spine specifically includes:

Described dendron image is normalized and structure enhancement process；

Dendron in skeletonizing structure enhancement process result, obtains dendron skeleton binary map；

Based on described dendron skeleton binary map, determine the bifurcation on dendron skeleton；

Based on described dendron skeleton binary map, and according to described bifurcation, position described dendritic spine.

6. method according to claim 5 is it is characterised in that described be normalized to described dendron image and structure increasing Strength reason specifically includes:

By described dendron image normalization between 0 and 1；

According to below equation, set dendritic spine INTERFACE MODEL:

$f(x,y)=\exp (-\frac{{(xsin\mathrm{\θ}-ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2})$

Wherein, described θ represents the angle of described dendritic spine；Described σ represents the width of described dendritic spine；Described f (x, y) represents institute State dendritic spine INTERFACE MODEL；Described x, described y represent the horizontal stroke of pixel, vertical coordinate in the image after normalization respectively；

Line structure according to strengthening image after normalization with drag:

$\mathrm{\λ}(x,y)=-\frac{1}{{\mathrm{\σ}}^2}\exp (-\frac{{(xsin\mathrm{\θ}+ycos\mathrm{\θ})}^2}{2{\mathrm{\σ}}^2}).$

7. method according to claim 5 is it is characterised in that dendron in described skeletonizing structure enhancement process result, Obtain dendron skeleton binary map to specifically include:

Row threshold division is entered to structure enhancement process result, obtains dendron bianry image；

Travel through described dendron bianry image, pixel is deleted according to following condition, until described dendron bianry image no longer changes Till, obtain the first bianry image:

x_h(p)=1；2≤min{n₁(p),n₂(p)}≤3；

Wherein, described p represents pixel；Described x₁,x₂,...,x₈It is the described p starting from right side adjoint point by inverse time needle sort 8 consecutive points pixels；

Travel through described first bianry image, pixel is deleted according to following condition, until described dendron bianry image no longer changes Till, obtain described dendron skeleton binary map:

x_h(q)=1；2≤min{n₁(q),n₂(q)}≤3；

Wherein, described q represents pixel；Described x₁,x₂,...,x₈It is the described q starting from right side adjoint point by inverse time needle sort 8 consecutive points pixels.

8. method according to claim 5 it is characterised in that described based on described dendron skeleton binary map, determine dendron Bifurcation on skeleton specifically includes:

Construction wave filter；

Described wave filter and described dendron skeleton binary map are carried out process of convolution, and the point of Second Threshold will be more than in trellis diagram It is defined as bifurcation.

9. method according to claim 5 it is characterised in that described based on described dendron skeleton binary map, and according to institute State bifurcation, position described dendritic spine, specifically include:

The structural element of construction predetermined radii, and morphological erosion is carried out to described dendron skeleton binary map；

Deduct the skeleton drawing after corrosion with described dendron skeleton binary map, obtain dendritic spine alternate location；

Based on described dendron skeleton binary map, centered on described bifurcation, extract the region of the predetermined length of side；

Choose the skeleton part of described bifurcation region, and judge whether described dendritic spine alternate location and this skeleton Partly it is connected；If there are it is determined that described dendritic spine alternate location is dendritic spine.

10. method according to claim 9 is it is characterised in that the described aixs cylinder brief summary with detecting and dendritic spine are carried out 2d synapse detection specifically includes:

Represent described aixs cylinder brief summary with the central point of each described aixs cylinder brief summary；

Represent described dendritic spine with the central point of described dendritic spine alternate location；

Calculate and record the minima of distance between described each aixs cylinder brief summary central point and all described dendritic spine central points；

If the minima of described distance is less than the 3rd threshold value, described aixs cylinder brief summary and described dendritic spine are defined as 2d synapse.

11. methods according to claim 9 it is characterised in that described based on 2d synapse testing result, carry out 3d synapse inspection Survey specifically includes:

Represent described synapse with the midpoint of aixs cylinder brief summary central point and dendritic spine alternate location central point in described 2d synapse Position；

With the distance of aixs cylinder brief summary central point described in described 2d synapse and described dendritic spine alternate location central point to represent State the verity of synapse；

In layer before and after the section of described two-photon image sequence is adjacent, look in the coordinate in the neighborhood of described 2d synoptic sites Look for described synapse；

Judge whether in verity layer before and after the section of described two-photon image sequence is adjacent of each described synapse be highest, If it is determined that described synapse is 3d synapse.