CN104706320A - Fluorescent diffusion optical cross-sectional image reestablishing method based on dfMC model - Google Patents

Abstract

The invention discloses a fluorescent diffusion optical cross-sectional image reestablishing method based on a decoupling fluorescence Monte Carlo (dfMC) model, and belongs to the technical field of biomedical engineering. The method includes the steps of firstly, determining a detecting area, selecting a plurality of scanning points in the detecting area, and obtaining fluorescent intensity distribution on a detector; secondly, establishing a three-dimensional digital model for depicting a tissue optical parameter space structure, conducting forward-direction white Monte Carlo simulation of stimulating photons according to the scanning positions and directions of a light source, tracking the stimulating photons, and recording the corresponding physical quantities of the photons reaching the detector on a path; thirdly, calculating the weight of fluorescent photons through a dfMC method, and calculating a fluorescent Jacobi matrix; fourthly, calculating the positions and absorbing coefficients of fluorophores in tissue through iterative reconstruction of GPU clusters. The method has the advantage of providing an accurate and rapid reestablishing method for a three-dimensional fluorescence tomography system through the high-precision dfMC model on the basis of the accelerated iterative reconstruction process of the GPU clusters.

Description

A kind of fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model

Technical field

The invention belongs to biomedical engineering technology field, particularly relating to a kind of fluorescence Diffuse Optical Tomography image rebuilding method based on separating coupling fluorescence Monte Carlo dfMC model.

Background technology

In recent years, fluorescence Diffuse Optical Tomography imaging technique has developed in order to a kind of important imaging tool ^[1], and be widely used in the visual research of cancer diagnosis, medicament research and development and gene expression.The accurate reconstruction of fluorescent target thing will depend on the inverse model (Reverse Problem) setting up photon transport model (i.e. so-called forward problem) and a photon propagation model accurately.Forward problem is mainly based on the diffusion approximation model under the Boltzmann radiation transport frame of preservation of energy ^[2]with MonteCarlo (MC) model ^[3].Because diffusion approximation model is low for image reconstruction times cost, be widely adopted; But existing diffusion model is only only applicable to high scattering material, suitable with scattering coefficient at absorptance, or when source spy distance is suitable with average transmission free path, model is just no longer applicable.Monte-Carlo model adapts to the biological tissue of any optical parametric, and arbitrarily complicated geometry object, but due to calculation cost very huge, computational efficiency is low.Generally as " goldstandard " of forward model ^[3].

Existingly be suitable for the forward direction fluorescence MC model that Reverse Problem solves and mainly contain adjoint method model (adjoint MC) ^[4]with perturbation method model (perturbation MC) ^[5].Adjoint method supposition source and detector are tradable, namely have equity, and this equity is set up for optical fiber type detection system, but for the situation that free space detects, the complex characteristics of light source and CCD makes this equity be difficult to be guaranteed.Be simultaneously low-down for the reconstruction efficiency based on free space detection situation, the detector of a large amount of CCD Pixel Dimensions sizes makes Monte Carlo simulation number of times increase severely.And perturbation method is mainly by coming calculating detector to receive to obtain fluorescence intensity in in-house MC path disturbance to exciting light.Perturbation method it by the restriction of imaging system, there is the advantage of speed and precision.Become the focus of research.But perturbation method ignores the impact of fluorogen on exciting light, have ignored the difference of exciting light and fluorescence optical parametric in organizing, this many hypothesis makes perturbation method be that one has inclined MC to simulate, its fluorescence statistic and the true fluorescent value of reality have deviation, and this deviation can change along with the change of optical parametric and number of photons, in most cases, error can be very large, and due to photon path information huge, will make to rebuild inefficiency under voxel model situation, whole process is consuming time very large.

List of references

[1]V.Ntziachristos，“Going deeper than microscopy：the optical imaging frontier inbiology，”Nat.Meth.7，603-614(2010).

[2]X.Song，D.Wang，N.Chen，J.Bai，and H.Wang，“Reconstruction for free-spacefluorescence tomography using a novel hybrid adaptive finite element algorithm，”Opt.Express 15，18300-18317(2007).

[3]J.Swartling，A.Pifferi，A.M.K.Enejder，and S.Andersson-Engels，“AcceleratedMonte Carlo models to simulate fluorescence spectra from layered tissues，”J.Opt.Soc.Am.A 20，714-727(2003).

[4]G.Ma，J.F.Delorme，P.Gallant，and D.A.Boas，“Comparison of simplified MonteCarlo simulation and diffusion approximation for the fluorescence signal fromphantoms with typical mouse tissue optical properties，”Appl.Opt.46，1686-1692(2007).

[5]Sassaroli，“Fast perturbation Monte Carlo method for photon migration inheterogeneous turbid media，”Opt.Lett.36，2095-2097(2011).

Summary of the invention

In order to overcome above-mentioned the deficiencies in the prior art, the invention provides a kind of fluorescence Diffuse Optical Tomography image rebuilding method based on separating coupling fluorescence Monte Carlo (dfMC) model, the object of the invention is in the precision keeping MC simulation " goldstandard ", improve the time consuming nature of MC simulation simultaneously, realize high accuracy rapid fluorescence tomographic image reconstructing.

The technical solution adopted in the present invention is:

Based on a fluorescence Diffuse Optical Tomography image rebuilding method for dfMC model, it is characterized in that, comprise the following steps:

(1) determine search coverage, select light source scanning element, and under different scanning position, obtain the fluorescence diffusion light distribution on detector;

(2) digitized object to be detected, sets up voxel model, carries out white Monte Carlo simulation according to scanning light source position and direction, and record is detected the weight of the excitation photon that device is collected and routing information in each voxel and corresponding physical quantity;

(3) according to dfMC model, fluorescent photon weight is calculated;

(4) fluorescence Jacobian matrix fluorescence Jacobi is set up by excitation photon routing information and corresponding physical quantity;

(5) according to weight and the Fluorescence Fluorescence Jacobian matrix of the fluorescent photon exported, iterative approximation calculates in-house fluorogen position and absorptance thereof.

In step (2), adopt GPU cluster to accelerate whole white Monte Carlo process, record arrives the related physical quantity on the excitation photon path of detector, the path related physical quantity L in voxel v _1v, L _2vbe respectively:

$L_{1v}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{P}_I({\hat{s}}_i\·{\hat{s}}_{i+1})/P_A({\hat{s}}_i\·{\hat{s}}_{i+1})$

$L_{2v}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j=1}{\overset{i}{\mathrm{\Σ}}}{l}_j{P}_I({\hat{s}}_i\·{\hat{s}}_{i+1})/P_A({\hat{s}}_i\·{\hat{s}}_{i+1})$

In formula: n is the scattering imaging in voxel v, the unit scattering direction vector of i-th scattering in voxel v, P _iisotropism directivity function, P _aanisotropic orientation function, l _jfor time scattering path length of the jth in voxel v.

In step (2), GPU cluster acceleration whole white Monte Carlo process is specific as follows:

2.1 on cluster, according to number of times number of light sources and each node of clustered node number assignment being carried out MC simulation, and corresponding light source position and directional information is transferred on node;

2.2 on each node, according to the photon number storage allocation that will simulate, can by organize models's information distribution to the texture depositor in GPU, optical parametric information distribution improves reading speed to literal register, and determines the number of parallel block and thread in optimum CUDA operation according to the technical information of GPU;

2.3 on each thread, simulates a photon bag, and in the process of scatter within tissue and absorption after following the tracks of photo emissions, and the introductory path information of record photon in each voxel and physical quantity, photon stops photonic analogy after overflowing tissue.

Step (3), calculates fluorescence weight w ^emp () computing formula is as follows:

$w_m^{\mathrm{em}}\left(p\right)=w_m^{\mathrm{ex}}\left(p\right)\underset{v=1}{\overset{k}{\mathrm{\Σ}}}\frac{\mathrm{\η}{\mathrm{\μ}}_{\mathrm{afv}}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}(L_{1v}-{\mathrm{\μ}}_{\mathrm{afv}}{L}_{2v})\underset{j=i}{\overset{m}{\mathrm{\Π}}}\left(\frac{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{em}})}{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{ex}})}\right)\mathrm{\Γ}(p_i\→p_e,{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{em}}-{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{wx}})$

In formula: p is the six-vector in position and unit direction, for unit direction vector, k by photon in in-house path the voxel number of process, m be photon at in-house scattering imaging, μ _afvfluorescence coefficient in voxel v, be the exciting light scattering coefficient in voxel v, η is quantum yield, w ^exfor excitation photon weight, g _v ^emfor the anisotropy factor of fluorescence in voxel v, g _v ^exfor the anisotropy factor of exciting light in voxel v, μ _av ^emfor the absorptance of fluorescence in voxel v, μ _av ^exfor the absorptance of exciting light in voxel v, and $\mathrm{\Γ}(p_i\→p_e,\mathrm{\μ})=\exp (\mathrm{\μ}\·l_{p_i\→p_e}).$

In step (4), the computing formula of fluorescence Jacobian matrix is as follows:

$\begin{array}{c}{J}_{\mathrm{jd}}^v=\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}\frac{{\∂w}_m^{\mathrm{em}}\left(p\right){|}_{\det \mathrm{e\; ctor\; d}}}{\∂{\mathrm{\μ}}_{\mathrm{afv}}}\\ =\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}{w}_m^{\mathrm{ex}}\left(p\right){|}_{\det \mathrm{ector\; d}}(\frac{\mathrm{\η}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{1v}-\frac{\mathrm{\η}{\mathrm{\μ}}_{\mathrm{afv}}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{2v})\underset{j=i}{\overset{m}{\mathrm{\Π}}}\left(\frac{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{em}})}{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{ex}})}\right)\mathrm{\Γ}(p_i\→p_e,{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{em}}-{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{ex}})\end{array}$

In formula: p is the six-vector in position and unit direction, for unit direction vector, m be photon at in-house scattering imaging, μ _afvfluorescence coefficient in voxel v, be the exciting light scattering coefficient in voxel v, η is quantum yield, w ^exfor excitation photon weight, g _v ^emfor the anisotropy factor of fluorescence in voxel v, g _v ^exfor the anisotropy factor of exciting light in voxel v, μ _av ^emfor the absorptance of fluorescence in voxel v, μ _av ^exfor the absorptance of exciting light in voxel v, and j be in node i light source numbering, d be detector numbering, v be need rebuild model voxel numbering, m be photon at in-house scattering imaging, J is fluorescence Jacobian matrix, for the fluorescence Jacobean matrix array element of corresponding voxel v, light source j, detector d.

The whole fluorescence process of reconstruction of step (5) is all accelerated by GPU cluster, the fluorescence Jacobian matrix block that each node calculates, and computing formula is as follows:

D _i＝[D ₁₁…D _1t；…；D _s1…D _st] ^T

In formula: s is the number of light sources in node i, t is detector sum, and n is the model body prime number order needing to rebuild, D _ithe fluorescence intensity distribution detected for the detector under light sources all in node i and dfMC simulate the difference of the fluorescence distribution obtained, D _stthe fluorescence intensity distribution detected for t probe unit in the detector under s light source in node i and corresponding dfMC simulate the difference of the fluorescence distribution obtained.

Step (5), intermediate quantity J _ij _i ^tx, J _id _ibe transferred on main joint and carry out conjugate gradient iterative fluorescence distribution again, the memory space that these intermediate quantities take than fluorescence Jacobian matrix block is few, and iterative computation formula is as follows:

$\underset{X}{\min }f\left(X\right)=\left|\right|J^T\mathrm{JX}-J^{T}D\left|\right|=\left|\right|\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(J_i{J_i}^{T}X-J_i{D}_i)\left|\right|$

In formula: i is node serial number, N is total interstitial content, and X is fluorogen absorptance distribution matrix J _i ^tfor matrix J _itransposed matrix, f (X) is object function.

Compared with prior art, beneficial effect of the present invention is: the reconstruction that can realize distributed in three dimensions to fluorogen in biological tissue and fluorescence coefficient, has reconstruction precision high, fireballing advantage.

Accompanying drawing explanation

Fig. 1 is white Monte Carlo simulation flow chart under GPU parallel accelerate of the present invention;

Fig. 2 is that under GPU cluster parallel accelerate of the present invention, dfMC three-dimensional fluorescence rebuilds flow chart;

Fig. 3 a, Fig. 3 b are respectively rebuilding body mould experimental result picture of the present invention.

Detailed description of the invention

Below in conjunction with accompanying drawing, the present invention is further described.

Implementation step of the present invention is as follows:

1, determine search coverage, select light source scanning element, and under different scanning position, obtain the fluorescence diffuse-reflectance light distribution D on detector, solve for fluorescence inverting;

2, digitized detection tissue, sets up organize models, according to scanning light source position and direction carry out GPU accelerates under white Monte Carlo simulation, acquisition is detected routing information in each voxel of excitation photon that device collects and related physical quantity L _1vand L _2v, as shown in Figure 1, step is as follows for the white Monte Carlo process under GPU cluster acceleration:

2.1 on cluster, according to number of times number of light sources and each node of clustered node number assignment being carried out MC simulation, and corresponding light source position and directional information is transferred on node;

2.2 on each node, according to the photon number storage allocation that will simulate, can by organize models's information distribution to the texture depositor in GPU, optical parametric information distribution improves reading speed to literal register, and determines the number of parallel block and thread in optimum CUDA operation according to the technical information of GPU;

2.3 on each thread, simulates a photon bag, and in the process of scatter within tissue and absorption after following the tracks of photo emissions, and the introductory path information of record photon in each voxel and physical quantity, photon stops photonic analogy after overflowing tissue.

3, carry out under GPU cluster, simulate based on fluorescence intensity distribution D and MC detected under each light source a large amount of physical messages along photon path record obtained and carry out fluorescence distribution inverting and solve, concrete steps as shown in Figure 2:

3.1 on cluster, distribute according to the cluster that cluster assignment information during white Monte Carlo simulation is carried out in process of reconstruction, make the corresponding physical quantity of excitation photon along paths record of some light source corresponding on each node, if total total n light source, that so each node has a 1/n light source MC simulation with information that is path-dependent;

3.2 on each node, is loaded into the along-path information of corresponding excitation photon, and calculate the Fluorescence Fluorescence Jacobian matrix block J of corresponding node i according to dfMC method _i, computing formula is as follows:

$\begin{array}{c}{J}_{\mathrm{jd}}^v=\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}\frac{{\∂w}_m^{\mathrm{em}}\left(p\right){|}_{\det \mathrm{e\; ctor\; d}}}{\∂{\mathrm{\μ}}_{\mathrm{afv}}}\\ =\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}{w}_m^{\mathrm{ex}}\left(p\right){|}_{\det \mathrm{ector\; d}}(\frac{\mathrm{\η}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{1v}-\frac{\mathrm{\η}{\mathrm{\μ}}_{\mathrm{afv}}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{2v})\underset{j=i}{\overset{m}{\mathrm{\Π}}}\left(\frac{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{em}})}{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{ex}})}\right)\mathrm{\Γ}(p_i\→p_e,{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{em}}-{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{ex}})\end{array}$

In formula: p is the six-vector in position and unit direction, for unit direction vector, m be photon at in-house scattering imaging, μ _afvfluorescence coefficient in voxel v, be the exciting light scattering coefficient in voxel v, η is quantum yield, w ^exfor excitation photon weight, g _v ^emfor the anisotropy factor of fluorescence in voxel v, g _v ^exfor the anisotropy factor of exciting light in voxel v, μ _av ^emfor the absorptance of fluorescence in voxel v, μ _av ^exfor the absorptance of exciting light in voxel v, and j be in node i light source numbering, d be detector numbering, v be need rebuild model voxel numbering, m be photon at in-house scattering imaging, J is fluorescence Jacobian matrix, for the fluorescence Jacobean matrix array element of corresponding voxel v, light source j, detector d.

3.3 can obtain according to the light source number in node and detector number the fluorescence Jacobian matrix block that node calculates, and its computing formula is as follows:

D _i＝[D ₁₁…D _1t；…；D _s1…D _st] ^T

In formula: s is the number of light sources in node i, t is detector sum, and n is the model body prime number order needing to rebuild, D _ithe fluorescence intensity distribution detected for the detector under light sources all in node i and pfMC simulate the difference of the fluorescence distribution obtained, D _stthe fluorescence intensity distribution detected for t probe unit in the detector under s light source in node i and corresponding dfMC simulate the difference of the fluorescence distribution obtained.

3.4 according to the weight of fluorescent photon exported and fluorescence Jacobian matrix block, and iterative computation bright dipping cumularsharolith is put and absorptance, and its computing formula is as follows:

$\underset{X}{\min }f\left(X\right)=\left|\right|J^T\mathrm{JX}-J^{T}D\left|\right|=\left|\right|\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(J_i{J_i}^{T}X-J_i{D}_i)\left|\right|$

In formula: i is node serial number, N is total interstitial content, and X is fluorogen absorptance distribution matrix J _i ^tfor matrix J _itransposed matrix, f (X) is object function.

The feasibility of the fluorescence Diffuse Optical Tomography formation method based on solution coupling fluorescence monte-Carlo model that the embodiment of the present invention provides is verified below with a simple fluorophor mould experiment, described below:

Make one and fill with fat emulsion solution glass (internal diameter 30mm, high 40mm) in be inserted with the body mould of two thin rods (internal diameter 2mm) of the fluorescent dye glass filling DIA-BOA, the optical parametric of Intralipid solution and the fluorescence coefficient of glo-stick is measured, fatty aqueous solution μ by Lamda 950 _s=214.7cm ^-1and μ _a=0.01cm ^-1, the fluorescence coefficient in two glass glo-sticks is respectively μ _af=0.6,0.8cm ^-1.Carry out fluorescence reconstruction by dfMC method to this model, as shown in Figure 3 a, 3 b, the glo-stick position rebuild in Fig. 3 b conforms to glo-stick position true in Fig. 3 a result.The experiment of body mould shows that the method precision is high, simultaneously by GPU cluster energy accelerated reconstruction.

Claims (7)

1., based on a fluorescence Diffuse Optical Tomography image rebuilding method for dfMC model, it is characterized in that, comprise the following steps:

(1) determine search coverage, select light source scanning element, and under different scanning position, obtain the fluorescence diffusion light distribution on detector;

(2) digitized object to be detected, sets up voxel model, carries out white Monte Carlo simulation according to scanning light source position and direction, and record is detected the weight of the excitation photon that device is collected and routing information in each voxel and corresponding physical quantity;

(3) according to dfMC model, fluorescent photon weight is calculated;

(4) fluorescence Jacobian matrix fluorescence Jacobi is set up by excitation photon routing information and corresponding physical quantity;

(5) according to weight and the Fluorescence Fluorescence Jacobian matrix of the fluorescent photon exported, iterative approximation calculates in-house fluorogen position and absorptance thereof.

2. the fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 1, it is characterized in that in step (2), GPU cluster is adopted to accelerate whole white Monte Carlo process, record arrives the related physical quantity on the excitation photon path of detector, the path related physical quantity L in voxel v _1v, L _2vbe respectively:

$L_{1v}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{P}_I({\hat{s}}_i\·{\hat{s}}_{i+1})/P_A({\hat{s}}_i\·{\hat{s}}_{i+1})$

$L_{2v}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j=1}{\overset{i}{\mathrm{\Σ}}}{l}_j{P}_I({\hat{s}}_i\·{\hat{s}}_{i+1})/P_A({\hat{s}}_i\·{\hat{s}}_{i+1})$

In formula: n is the scattering imaging in voxel v, the unit scattering direction vector of i-th scattering in voxel v, P _iisotropism directivity function, P _aanisotropic orientation function, l _jfor the jth time scattering path length of photon in voxel v.

3. the fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 1, is characterized in that: in step (2), GPU cluster acceleration whole white Monte Carlo process is specific as follows:

2.1 on cluster, according to number of times number of light sources and each node of clustered node number assignment being carried out MC simulation, and corresponding light source position and directional information is transferred on node;

2.2 on each node, according to the photon number storage allocation that will simulate, can by organize models's information distribution to the texture depositor in GPU, optical parametric information distribution improves reading speed to literal register, and determines the number of parallel block and thread in optimum CUDA operation according to the technical information of GPU;

2.3 on each thread, simulates a photon bag, and in the process of scatter within tissue and absorption after following the tracks of photo emissions, and the introductory path information of record photon in each voxel and physical quantity, photon stops photonic analogy after overflowing tissue.

4. the fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 1, is characterized in that step (3), calculates fluorescence weight w ^emp () computing formula is as follows:

$w_m^{\mathrm{em}}\left(p\right)=w_m^{\mathrm{ex}}\left(p\right)\underset{v=1}{\overset{k}{\mathrm{\Σ}}}\frac{\mathrm{\η}{\mathrm{\μ}}_{\mathrm{afv}}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}(L_{1v}-{\mathrm{\μ}}_{\mathrm{afv}}{L}_{2v})\underset{j=i}{\overset{m}{\mathrm{\Π}}}\left(\frac{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{em}})}{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{ex}})}\right)\mathrm{\Γ}(p_i\→p_e,{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{em}}-{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{ex}})$

In formula: p is the six-vector in position and unit direction, for unit direction vector, k by photon in in-house path the voxel number of process, m be photon at in-house scattering imaging, μ _afvfluorescence coefficient in voxel v, be the exciting light scattering coefficient in voxel v, η is quantum yield, w ^exfor excitation photon weight, g _v ^emfor the anisotropy factor of fluorescence in voxel v, g _v ^exfor the anisotropy factor of exciting light in voxel v, μ _av ^emfor the absorptance of fluorescence in voxel v, μ _av ^exfor the absorptance of exciting light in voxel v, and

5. the fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 1, is characterized in that in step (4), the computing formula of fluorescence Jacobian matrix is as follows:

$\begin{array}{c}{J}_{\mathrm{jd}}^v=\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}\frac{{\∂w}_m^{\mathrm{em}}\left(p\right){|}_{\det \mathrm{ectord}}}{{\∂\mathrm{\μ}}_{\mathrm{afv}}}\\ =\underset{m=0}{\overset{\∞}{\mathrm{\Σ}}}{w}_m^{\mathrm{ex}}\left(p\right){|}_{\det \mathrm{ectord}}(\frac{\mathrm{\η}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{1v}-\frac{{\mathrm{\η\μ}}_{\mathrm{afv}}}{{\mathrm{\μ}}_{\mathrm{sv}}^{\mathrm{ex}}}{L}_{2v})\underset{j=i}{\overset{m}{\mathrm{\Π}}}\left(\frac{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{em}})}{P_A({\hat{s}}_j\·{\hat{s}}_{j+1},g_v^{\mathrm{ex}})}\right)\mathrm{\Γ}(p_i\→p_e,{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{em}}-{\mathrm{\μ}}_{\mathrm{av}}^{\mathrm{ex}})\end{array}$

In formula: p is the six-vector in position and unit direction, for unit direction vector, m be photon at in-house scattering imaging, μ _afvfluorescence coefficient in voxel v, the exciting light scattering coefficient in voxel v, _ηquantum yield, w ^exfor excitation photon weight, g _v ^emfor the anisotropy factor of fluorescence in voxel v, g _v ^exfor the anisotropy factor of exciting light in voxel v, μ _av ^emfor the absorptance of fluorescence in voxel v, μ _av ^exfor the absorptance of exciting light in voxel v, and j be in node i light source numbering, d be detector numbering, v be need rebuild model voxel numbering, m be photon at in-house scattering imaging, J is fluorescence Jacobian matrix, for the fluorescence Jacobean matrix array element of corresponding voxel v, light source j, detector d.

6. the rapid fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 1, it is characterized in that the whole fluorescence process of reconstruction of step (5) is all accelerated by GPU cluster, the fluorescence Jacobian matrix block that each node calculates, computing formula is as follows:

D _i＝[D ₁₁…D _1t；…；D _s1…D _st] ^T

In formula: s is the number of light sources in node i, t is detector sum, and n is the model body prime number order needing to rebuild, D _ithe fluorescence intensity distribution detected for the detector under light sources all in node i and dfMC simulate the difference of the fluorescence distribution obtained, D _stthe fluorescence intensity distribution detected for t probe unit in the detector under s light source in node i and corresponding dfMC simulate the difference of the fluorescence distribution obtained.

7. the rapid fluorescence Diffuse Optical Tomography image rebuilding method based on dfMC model according to claim 6, is characterized in that step (5), intermediate quantity j _id _ibe transferred on main joint and carry out conjugate gradient iterative fluorescence distribution again, the memory space that these intermediate quantities take than fluorescence Jacobian matrix block is few, and iterative computation formula is as follows:

$\underset{X}{\min }f\left(X\right)=\left|\right|J^T\mathrm{JX}-J^{T}D\left|\right|=\left|\right|\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(J_i{J_i}^{T}X-J_i{D}_i)\left|\right|$

In formula: i is node serial number, N is total interstitial content, and X is fluorogen absorptance distribution matrix J _i ^tfor matrix J _itransposed matrix, f (X) is object function.