WO2017004851A1 - 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法 - Google Patents

基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法 Download PDF

Info

Publication number
WO2017004851A1
WO2017004851A1 PCT/CN2015/084555 CN2015084555W WO2017004851A1 WO 2017004851 A1 WO2017004851 A1 WO 2017004851A1 CN 2015084555 W CN2015084555 W CN 2015084555W WO 2017004851 A1 WO2017004851 A1 WO 2017004851A1
Authority
WO
WIPO (PCT)
Prior art keywords
light source
boundary
equation
reconstruction
light
Prior art date
Application number
PCT/CN2015/084555
Other languages
English (en)
French (fr)
Inventor
冯金超
贾克斌
魏慧军
Original Assignee
北京工业大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 北京工业大学 filed Critical 北京工业大学
Priority to US15/108,394 priority Critical patent/US9672639B1/en
Publication of WO2017004851A1 publication Critical patent/WO2017004851A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
    • A61B5/0075—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by spectroscopy, i.e. measuring spectra, e.g. Raman spectroscopy, infrared absorption spectroscopy
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/003—Reconstruction from projections, e.g. tomography
    • G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
    • A61B5/0071—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by measuring fluorescence emission
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
    • A61B5/0073—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by tomography, i.e. reconstruction of 3D images from 2D projections
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/003—Reconstruction from projections, e.g. tomography
    • G06T11/006—Inverse problem, transformation from projection-space into object-space, e.g. transform methods, back-projection, algebraic methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/00—3D [Three Dimensional] image rendering
    • G06T15/50—Lighting effects
    • G06T15/506—Illumination models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B2503/00—Evaluating a particular growth phase or type of persons or animals
    • A61B2503/40—Animals
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B2576/00—Medical imaging apparatus involving image processing or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2210/00—Indexing scheme for image generation or computer graphics
    • G06T2210/41—Medical
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2211/00—Image generation
    • G06T2211/40—Computed tomography
    • G06T2211/432—Truncation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing

Definitions

  • the invention belongs to the field of medical image processing, and relates to a bioluminescence tomography reconstruction algorithm based on multi-task Bayesian compression sensing method.
  • Optical molecular imaging is a rapidly developing molecular imaging technique that combines optical processes with certain molecular properties to analyze and process bioluminescence or excitation fluorescence in a target, and to conduct qualitative and quantitative studies.
  • the most representative imaging methods in optical molecular imaging technology are fluorescence imaging (Fluorescence Imagi ng) and Bioluminescence Imaging (BLI). They are two-dimensional bioluminescence imaging technology. Although this imaging technology is convenient and simple to apply, this two-dimensional imaging method has limitations in the application process, especially for the limitation of imaging depth, two-dimensional fluorescence image.
  • the source depth information is not reflected and is difficult to quantify. They can only reflect the projection information of the fluorescent probe in the living body at a certain angle, and the projection information is a superposition of signals of multiple depths. Therefore, the two-dimensional imaging method has a very low resolution.
  • Bioluminescence Tomography which is based on two-dimensional bioluminescence imaging technology, has become an important branch of optical molecular imaging because it can reflect the information of signal depth.
  • Bioluminescence tomography does not require excitation from an external source, but instead emits light in vivo through a biochemical luminescence reaction.
  • the fluorescence generated in the body propagates in a certain pattern within the biological tissue and constantly interacts with the biological tissue and reaches the body surface.
  • the fluorescence image obtained by the high-sensitivity detector in the biological tissue surface can reconstruct the distribution of the fluorescent light source in the small animal, thereby revealing the activity law of the body molecule in essence.
  • the multi-spectral method increases the known amount of the equation to be solved, thus improving the results of the solution.
  • the multi-spectral based method is to keep the light source unchanged, and use multiple filters to acquire data of different bands to reconstruct the light source.
  • the measurement data between multiple spectra is not isolated, but related, and their common task is to reconstruct the fluorescent light source. Therefore, if the correlation between multiple spectra is taken into account in reconstruction, it will be possible to improve the BLT reconstructed image quality.
  • the present invention proposes a BLT reconstruction method based on multi-spectral intrinsic correlation.
  • the invention is based on a multi-task learning method, first adopts a high-order approximation model model to transmit light in a biological tissue, and based on a multi-task learning method to explore the intrinsic correlation between multiple spectra, and finally realizes a fluorescent light source based on the method. Three-dimensional reconstruction.
  • the present invention proposes a multi-task learning based method, which adopts a high-order approximation model model to transmit light in biological tissues, and explores based on multi-task learning methods.
  • the intrinsic correlation between multiple spectra, and finally the three-dimensional reconstruction of the fluorescent light source is realized.
  • the technical scheme adopted by the invention is as follows: firstly, the transmission law of light in biological tissue is adopted by a high-order approximation model model, and the intrinsic correlation between multiple spectra is explored based on the multi-task learning method, and the correlation between multiple spectra is taken as the first The information is integrated into the reconstruction algorithm to reduce the morbidity of BLT reconstruction. Finally, the three-dimensional reconstruction of the fluorescent light source is realized. Specifically, the following steps are included:
  • Step one problem definition and initialization - obtain the values of the P spectral segments at the M measurement points, and set the shape parameter a and the scale parameter b of the gamma prior distribution of the ⁇ 0 obey, wherein The variance of the Gaussian distribution obeyed by the measured value of the emitted light ⁇ ( ⁇ i );
  • the radiative transfer equation is a mathematical model that accurately describes the law of light propagation, but even under very simplified conditions, the solution of RTE is still very difficult and computationally intensive.
  • the diffusion equation is a low-order approximation of the radiative transfer equation, so its solution is simple, but the solution accuracy is low. Further, the diffusion equation must meet the conditions of strong scattering and low absorption of biological tissues when applied, and it is not suitable for the case where the light source is located at the boundary of the biological tissue, which limits the application of the diffusion equation in the biomedical field. Based on this, the present invention selects the Simplified Spherical Harmonics Approxima tion (SPN) equation.
  • SPN Simplified Spherical Harmonics Approxima tion
  • the model is able to more accurately describe the transmission of light in strongly absorbed low-scattering biological tissue, while maintaining a moderate computational complexity.
  • This model is used to replace the diffusion approximation transfer equation to accurately simulate the transmission law of light in biological tissues to improve the accuracy of forward solution.
  • the SP7 model has four equations when the band is ⁇ i and the position is r:
  • S represents a light source function
  • ⁇ a is a light absorption coefficient
  • ⁇ s is a light scattering coefficient
  • ⁇ ai ⁇ a + ⁇ s (1-g i )
  • g i is an anisotropy factor
  • a i , B i and C i are a series of constants, which are related to the angular distance of the boundary reflectivity.
  • the corresponding body surface emits light ⁇ (r, ⁇ i ):
  • Step 2 build a linear relationship——based on the finite element discrete SP3 diffusion equation method, build a relationship model between the boundary measured value and the unknown source for each wavelength;
  • the weak solution of the boundary condition is then incorporated into the weak solution of the SP3 equation using the Green's Formula.
  • W has the following form:
  • B is a matrix of size N ⁇ N, with Is two block sub-matrices, which can be expressed as:
  • the energy ratio ⁇ ( ⁇ i ) of the spectral segment is measured by prior spectral analysis, so that S represents the total energy of the light source of all spectral segments, ie p is the number of spectral segments. Integrate the source and boundary measurements of all spectral segments.
  • the relationship between the unknown fluorescent source distribution and the boundary exiting light in a multispectral case is as follows:
  • Step 3 sharing the prior estimate - inferring the multi-spectral related parameter ⁇ based on the empirical Bayesian maximum likelihood function
  • the measured value of the emitted light ⁇ ( ⁇ i ) usually contains noise. If the noise obeys the mean value, the variance is The Gaussian random distribution, then the maximum likelihood function of ⁇ ( ⁇ i ) with respect to the fluorescence source distribution S( ⁇ i ) and ⁇ 0 is expressed as:
  • ⁇ ( ⁇ 1 ,... ⁇ j ,... ⁇ N ) T is a hyperparameter used to represent a priori information. Then, in all the spectral segments, ⁇ j has the same value, that is, ⁇ is used to characterize the correlation between the respective spectral segments.
  • the empirical maximum Bayesian method is used to obtain the boundary maximum likelihood function of ⁇ , namely:
  • Step 4 Unknown Light Source Reconstruction - Estimation Based on Known Hyperparameters And the boundary measurement value ⁇ mul , using the maximum likelihood function to reconstruct the fluorescent light source.
  • the large likelihood function of the unknown source S can be expressed as:
  • the invention is based on a multi-task learning method, first adopting a high-order approximation model model for the transmission law of light in biological tissues, Based on the multi-task learning method, the intrinsic correlation between multiple spectra is explored. Finally, the three-dimensional reconstruction of the fluorescent light source is realized. From the results of reconstruction, the invention not only accurately reconstructs the fluorescent light source in the mouse, but also greatly improves the calculation efficiency.
  • Figure 1 is a structural diagram of a multitasking model framework
  • FIG. 2 is a schematic diagram of a digital mouse chest measurement value acquisition area
  • Fig. 3 is a reconstruction result of the light source in the lungs.
  • (a) shows the reconstructed position of the light source in the lung
  • (b)-(d) shows the coronal slice corresponding to the BLT reconstruction result of the XCT.
  • the position of the real light source is indicated by a circle mark;
  • FIG. 4 is a light source reconstruction result based on the L1 regularization reconstruction method (a)-(c) represents a coronal slice corresponding to the BLT reconstruction result of the XCT, and the position of the real light source is indicated by a circle mark;
  • Figure 5 is a diagram showing the results of light source reconstruction in the liver (a) and (b) are reconstructed slices using micro-CT.
  • (c)-(f) corresponds to the distribution of the light source reconstructed based on the algorithm of this chapter
  • (g)-(j) corresponds to the reconstruction result using the L1 norm regularization method.
  • the position of the real light source is marked with a circle in the figure.
  • Figure 6 shows the reconstruction results of the two algorithms.
  • (a) and (c) correspond to the results of the reconstruction of the algorithm
  • (b) and (d) correspond to the reconstruction results based on the L1 regularization algorithm.
  • the true light source position is marked with a circle.
  • Figure 7 is a reconstruction of the light source in the liver
  • (a) is a coronal slice of micro-CT data.
  • (b) and (d) are the results of IRIRM and L1-RMRM reconstruction, respectively.
  • (c) and (e) reconstruct the results on the XCT for both algorithms. Mark the actual light source position with a circle.
  • the reconstruction result of the light source of Fig. 8 is (a) the reconstruction result based on the Bayesian compression sensing method, and (b) the reconstruction result of the algorithm. Mark the actual light source position with a circle.
  • the present invention uses the finite element method to generate forward boundary measurement data.
  • the SP3 equation is used to generate forward data.
  • the forward data grid contains 42342 nodes and 208966 tetrahedral elements.
  • the required mesh for reconstruction contains 20196. Nodes and 108086 tetrahedrons.
  • the two spectral segments selected in the experiment were 620 nm and 700 nm, respectively.
  • the first group In the first group of experiments, a spherical light source with a radius of 1.5 mm was placed in the lungs of mice with a spherical center position of (14.5, 44.0, 12.8) mm and a distance of 4.8 mm from the surface of the living body. .
  • the reconstruction algorithm proposed by the present invention is used to reconstruct the fluorescence source, and the reconstruction result is analyzed.
  • the result is shown in FIG. 3. It can be seen that the fluorescent light source can be accurately reconstructed by the method proposed by the present invention.
  • the center position of the reconstructed light source is (14.69, 44.17, 12.91) mm, and the absolute distance from the real light source is 0.28 mm.
  • the reconstruction result based on the L1 regularization reconstruction method is shown in Fig. 4.
  • the center position of the light source reconstructed by the L1 regularization reconstruction method is (14.59, 43.63, 12.88) mm, and the distance from the center of the real light source is 0.39 mm.
  • the computation time for reconstruction is 1773.9 seconds, and the computation time of this chapter is only 0.67 seconds.
  • the second group In the second group of experiments, in the experiment, keeping the size of the light source unchanged, the light source was placed in the liver, the center position of the light source was (22.0, 51.0, 13.0) mm, and the surface of the living body was 7.06 mm.
  • the light source reconstruction was carried out by the method of the present invention, and the results are shown in Fig. 5. It can be seen that the fluorescence source can be reconstructed accurately by the method proposed in this chapter.
  • the center position of the reconstructed light source is (22.37, 50.68, 13.15) mm, and the absolute distance from the real light source is 0.5612 mm.
  • the center position of the light source reconstructed by the L1 regularization reconstruction method is (23.01, 50.06, 14.87) mm, and the distance from the center of the real light source is 2.3239 mm.
  • the reconstruction method based on L1 regularization has a poor reconstruction effect and a large error when the reconstructed fluorescent light source is located in a deep position in the biological tissue; and the method proposed in this chapter can accurately locate the fluorescent light source.
  • the calculation time of the L1 regularization method reconstruction is 1738.8 seconds, and the calculation time of the algorithm in this paper is only 0.936 seconds.
  • the first group In the first set of experiments, a spherical light source with a radius of 1.5 mm was placed in the liver and lungs, respectively, with center positions of (14, 51, 16) mm and (21, 51, 16) mm. The distance from the light source to the body surface is 5.56 mm and 5.82 mm, respectively.
  • the reconstruction is performed by the method of the present invention and the method based on L1 regularization, and the reconstruction result is shown in FIG. As can be seen from Figure 7, both algorithms can accurately reconstruct the fluorescent light source.
  • the second group In the second group of experiments, two light sources were placed in the liver, and the above two algorithms were used to reconstruct the light source.
  • the reconstruction results are shown in Fig. 8. It can be seen from Fig. 8 that the reconstruction method based on L1 regularization only reconstructs one light source, and the algorithm proposed by the invention can accurately reconstruct two light sources, which further proves the effectiveness of the method.
  • a mouse-shaped imitation is used, and a single light source is set in the lungs and the liver for simulation experiments, and the algorithm proposed by the present invention is compared with the reconstruction algorithm based on L1 regularization. Further, a dual-light source experiment was carried out, and the reconstruction method of the present invention and the reconstruction algorithm based on L1 regularization were respectively used to reconstruct the light source, and the experimental results were compared. The experimental results show that the proposed algorithm can not only accurately reconstruct the fluorescence source in mice, but also greatly improve the computational efficiency, and further validate the effectiveness of the algorithm.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Signal Processing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Physiology (AREA)
  • Mathematical Physics (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Computer Graphics (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Algebra (AREA)
  • Fuzzy Systems (AREA)
  • Evolutionary Computation (AREA)
  • Power Engineering (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Geometry (AREA)
  • Software Systems (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)

Abstract

一种基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法，属于医学图像处理领域，该方法首先采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，将多光谱之间的相关性作为先验信息融入重建算法中以降低BLT重建的病态性，最后在此基础上实现了荧光光源的三维重建。本发明方法与其他生物发光断层成像重建算法相比较，进一步地融入了多光谱相关性，降低了BLT重建病态性，不仅能对荧光光源进行准确重建定位，而且可以极大地提高了计算效率。

Description

基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法 技术领域
本发明属于医学图像处理领域，涉及一种基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法。
背景技术
光学分子影像是一种快速发展的分子影像技术，它通过将光学过程与一定的分子性质相结合，对目标体内的生物荧光或激发荧光进行分析和处理，并进行定性和定量研究。
光学分子影像技术中最具代表性的成像方式是激发荧光成像技术(Fluorescence Imagi ng)和生物发光成像技术(Bioluminescence Imaging,BLI)。它们均是二维的生物发光成像技术，虽然这种成像技术应用起来方便、简单，但这种二维成像方式在应用过程中存在局限性，尤其是对于成像深度的限制，二维的荧光图像不能反映光源深度信息和且难以定量化。它们仅可以反映生物体内的荧光探针在某一角度的投影信息，而这种投影信息是多个深度的信号叠加而成的。因此，二维成像方法具有很低的分辨率。
在二维生物发光成像技术基础上发展起来的生物发光断层成像技术(Bioluminescence Tomography,BLT)由于能够反映信号深度的信息，已经成为光学分子影像的一个重要分支。生物发光断层成像技术不需要外在光源的激发，而是通过一种生物化学发光反应在体内发光。体内产生的荧光在生物组织内部以某种规律传播并不断地与生物组织发生相互作用，并到达体表。最后，在生物组织体表利用高灵敏度的探测器获得的荧光图像就可以重建出荧光光源在小动物体内的分布情况，从而在本质上揭示在体分子的活动规律。
但光子在生物组织中不沿直线传输，而是经历了大量的散射过程，导致BLT逆问题在数学上是一个高度病态的问题，外界微小的测量扰动都会给重建结果带来很大的变化。国内外研究人员为降低其病态性做了很多工作，一般都是从不同角度为该问题提供各种不同的先验知识。
现有的降低其病态性的方法大多是基于多光谱信息和光源的稀疏特性展开BLT重建方法的研究，但是，这些方法并没有考虑多光谱之间的相关性。在数学上，多光谱方法增加了待求解方程的已知量，因此可以改善求解的结果。对于BLT而言，基于多光谱的方法是在光源保持不变，利用多个滤波片来获取不同波段的数据，从而进行光源的重建。实际上，多个光谱之间的测量数据不是孤立的，而是相关关联的，它们的共同任务是重建荧光光源。因此，如果在重建时考虑到多光谱之间的相关性，将有可能改善BLT重建图像质量。基于此，本发明提出了一种基于多光谱内在相关性的BLT重建方法。
本发明基于多任务学习方法，首先采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，最后在此基础上实现了荧光光源的三维重建。
发明内容
针对现有荧光断层成像重建算法中存在的上述问题，本发明提出了一种基于多任务学习的方法，它采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，最后在此基础上实现了荧光光源的三维重建。
本发明采用的技术方案为：首先采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，将多光谱之间的相关性作为先验信息融入重建算法中以降低BLT重建的病态性，最后在此基础上实现了荧光光源的三维重建。具体包括下述步骤：
步骤一，问题定义及初始化——获得P个谱段在M个测量点上的值，并设置α0服从的伽马先验分布的形状参数a和尺度参数b，其中
Figure PCTCN2015084555-appb-000001
为出射光测量值Φ(τi)服从的高斯分布的方差；
在模拟光在生物组织中传输规律时可以采用两种模型：辐射传输方程和扩散方程。辐射传输方程是一种精确描述光传播规律的数学模型，但是即使在非常简化的条件下RTE的求解仍然十分困难，且计算量非常巨大。扩散方程是辐射传输方程的低阶近似，因此其求解简单，但求解精度低。进一步，扩散方程在应用时必须满足生物组织强散射、低吸收的条件，而且不适用于光源位于生物组织边界时的情况，限制了扩散方程在生物医学领域中的应用。基于此，本发明选择高阶球谐近似(Simplified Spherical Harmonics Approxima tion，SPN)方程。该模型能够较为精确的描述光在强吸收低散射的生物组织中传输，同时，它还保持着适中的计算复杂度。用该模型替代扩散近似传输方程以准确地模拟光在生物组织中的传输规律，以提高前向求解的精度。
考虑到光谱影响，在波段为τi时，位置为r时，SP7模型共有四个方程：
Figure PCTCN2015084555-appb-000002
Figure PCTCN2015084555-appb-000003
Figure PCTCN2015084555-appb-000004
Figure PCTCN2015084555-appb-000005
其中，S代表光源函数；μa是光吸收系数；μs是光散射系数；μai＝μa+μs(1-gi)，gi为各向异性因子；
Figure PCTCN2015084555-appb-000006
代表辐射度的勒让德矩(Legendre Moments)φi的线性组合，可表示为：
Figure PCTCN2015084555-appb-000007
在SP7方程的基础上，令φ6＝φ4＝0，并保留公式(1)和(2)，即可得到SP3方程：
Figure PCTCN2015084555-appb-000008
Figure PCTCN2015084555-appb-000009
在SP7的边界条件的基础上，令φ6＝φ4＝0，得到SP3的边界条件：
Figure PCTCN2015084555-appb-000010
Figure PCTCN2015084555-appb-000011
其中，
Figure PCTCN2015084555-appb-000012
代表垂直边界向外的单位法向量；Ai、Bi和Ci为一系列常数，与边界反射率的角度距有关。对应的体表出射光Φ(r,τi)：
Figure PCTCN2015084555-appb-000013
其中，J0，…，J3为一系列常数。
步骤二，搭建线性关系——基于有限元离散的SP3扩散方程方法，针对每一个波长搭建边界测量值与未知光源的关系模型；
在公式(6)和(7)中，S(r,τi)表示的是荧光光源分布，是需要求解的未知量，为了使用有限元法对SP3方程及其边界条件进行求解，首先将SP3的两个方程写成对应的弱解形式：
Figure PCTCN2015084555-appb-000014
其中，Ψ(r,τi)是测试函数，公式(8)和(9)两个边界条件也要写为弱解形式：
Figure PCTCN2015084555-appb-000015
Figure PCTCN2015084555-appb-000016
然后利用格林公式(Green’s Formula)将边界条件的弱解形式融入到SP3方程的弱解形式中。
Figure PCTCN2015084555-appb-000017
Figure PCTCN2015084555-appb-000018
其中e1、e2、e3和e4分别为：
Figure PCTCN2015084555-appb-000019
Figure PCTCN2015084555-appb-000020
Figure PCTCN2015084555-appb-000021
Figure PCTCN2015084555-appb-000022
上式中，W具有如下形式：
Figure PCTCN2015084555-appb-000023
进行网格剖分，将光源分布S(r,τi)写成只与谱段相关的S(τi)，选择相应基函数，合成整体的刚度矩阵Mij，将SP3方程离散为如下形式的矩阵方程：
Figure PCTCN2015084555-appb-000024
其中B为N×N大小的矩阵，
Figure PCTCN2015084555-appb-000025
和
Figure PCTCN2015084555-appb-000026
是是两个分块子矩阵，可表示为：
Figure PCTCN2015084555-appb-000027
Figure PCTCN2015084555-appb-000028
这里需要建立出表面出射光Φ(τi)和未知荧光光源分布S(τi)之间的线性关系。在实验过程中，由于只能采集到生物体表面的光学信号，因此，只保留Φ(τi)中边界上节点值，去掉矩阵中不包含边界节点的行，得到方程：
Figure PCTCN2015084555-appb-000029
上述方程建立了未知荧光光源分布和边界出射光之间的关系，即：
Φ(τi)＝A(τi)S(τi)
在多光谱情况下，谱段的能量比ω(τi)通过预先的谱分析测得，令S代表所有谱段的光源的总能量，即
Figure PCTCN2015084555-appb-000030
p为谱段个数。将所有谱段的光源和边界测量值进行整合，多光谱情况下未知荧光光源分布和边界出射光之间的关系如下：
AmulS＝Φmul
此时，
Figure PCTCN2015084555-appb-000031
步骤三，共享先验估计——基于经验贝叶斯极大似然函数，推断表征多光谱相关的参数α；
在实际问题中，出射光测量值Φ(τi)通常含有噪声，若噪声服从均值为零，方差为
Figure PCTCN2015084555-appb-000032
的高斯随机分布，则Φ(τi)关于荧光光源分布S(τi)和α0的极大似然函数表示为：
Figure PCTCN2015084555-appb-000033
为了表示谱段间关系，引入一种多任务的思想，将整个光源作为一个整体任务T。其中，S(τi)是T中的第i个任务模型，用Sj(τi)表示为S(τi)的第j个组成部分。若S(τi)服从均值为零的高斯先验分布，参数α0服从Gamma先验分布，即有：
Figure PCTCN2015084555-appb-000034
p(α0|a,b)＝Ga(α0|a,b)
其中，α＝(α1,…αj,…αN)T是超参数，用来表示先验信息。那么，在所有的谱段中，αj的取值相同，即用α表征各个谱段之间的相关性。为了估计超参数α，使用经验贝叶斯的方法，得到α的边界最大似然函数，即有：
Figure PCTCN2015084555-appb-000035
其中，Bi＝I+A(τi)Λ-1A(τi)T，Λ-1＝diag(α1,α1,…,αN)。
步骤四，未知光源重建——根据已知的超参数的估计
Figure PCTCN2015084555-appb-000036
和边界测量值Φmul，利用极大似然函数重建荧光光源。
得到估计的
Figure PCTCN2015084555-appb-000037
后，再将所有谱段的光源和边界测量值进行整合，则未知光源S的大似然函数可表示为：
Figure PCTCN2015084555-appb-000038
这里，
Figure PCTCN2015084555-appb-000039
Figure PCTCN2015084555-appb-000040
式中，
Figure PCTCN2015084555-appb-000041
多任务模型框架如图1所示。
本发明基于多任务学习方法，首先采用高阶近似模型模型光在生物组织中的传输规律， 并基于多任务学习方法探索多光谱之间的内在相关性，最后在此基础上实现了荧光光源的三维重建。从重建结果来看，本发明不但将小鼠体内的荧光光源进行了准确重建定位，而且计算效率也得到了极大提高。
附图说明
图1为多任务模型框架结构图；
图2为数字鼠胸部测量值采集区示意图；
图3为肺部中光源的重建结果图(a)为肺部中光源的重建位置，(b)-(d)表示对应XCT的BLT重建结果的冠状切片。在(b)-(d)中，用圈标记示意真实光源的位置；
图4为基于L1正则化重建方法的光源重建结果(a)-(c)表示对应XCT的BLT重建结果的冠状切片，用圈标记示意真实光源的位置；
图5为肝脏中光源重建结果图(a)和(b)是利用micro-CT的重建切片。(c)-(f)对应基于本章算法重建的光源分布，(g)-(j)对应使用L1范数正则化方法的重建结果。(c)和(i)是y＝51mm时的冠状切片，(f)和(j)为z＝12.97mm的横截面。图中用圈标记真实光源位置。
图6为两种算法的重建结果。(a)和(c)对应本算法重建的结果，(b)和(d)对应基于L1正则化算法的重建结果。(a)和(b)为y＝51.05mm的冠状面，(c)和(d)为z＝15.97的横截面。真实光源位置用圈标记。
图7为肝脏中光源的重建结果(a)是micro-CT数据的冠状切片。(b)和(d)分别为IRIRM和L1-RMRM重建的结果。(c)和(e)为两种算法在XCT上重建结果。用圈标记真实光源位置。
图8光源的重建结果(a)为基于贝叶斯压缩感知方法的重建结果，(b)为本算法重建结果。用圈标记真实光源位置。
具体实施方式
下面结合附图和实施例对本发明做更详细的说明。
为了检验本发明提出方法的有效性，采用了非匀质的数字鼠仿体进行仿真实验，共进行了两项实验。在实验中，认为生物组织的光学特性参数已知，详细参数如表1所示。本发明采用有限元方法产生前向边界测量数据。为有效地避免“逆行为”，在产生前向数据时，采用SP3方程产生前向数据，前向数据的网格含有42342个节点和208966个四面体单元，重建时所需要的网格含有20196个节点和108086个四面体。考虑到光谱分布，实验中选择的两个谱段分别为620nm和700nm。为真实模拟荧光数据采集，仅使用了数字鼠胸部的测量数据(共1311个边界测量点)，如图2所示。
表1 生物组织光学特性参数
Figure PCTCN2015084555-appb-000042
Figure PCTCN2015084555-appb-000043
实验一：
第一组：第一组试验中，将半径大小为1.5mm的球形光源放置在小鼠的肺中，球心位置为(14.5,44.0,12.8)mm，光源位置距生物体表面距离为4.8mm。
使用本发明提出的重建算法进行荧光光源重建，并对重建结果进行分析，结果如图3所示。可以看到，利用本发明提出的方法可以准确的重建荧光光源，重建光源的中心位置为(14.69,44.17,12.91)mm，与真实光源的绝对距离为0.28mm。
进一步，为了验证本章算法的有效性，与基于L1正则化的重建方法进行了对比，基于L1正则化的重建方法重建结果如图4所示。利用L1正则化重建方法重建的光源中心位置为(14.59,43.63,12.88)mm，与真实光源中心距离为0.39mm。
另外，利用L1正则化重建方法，重建耗费的计算时间为1773.9秒，而利用本章算法重建，耗费的计算时间仅为0.67秒。
第二组：在第二组实验，在实验中，保持光源的大小不变，将光源放置于肝脏中，光源中心位置为(22.0,51.0,13.0)mm，距离生物体表面7.06mm。
然后，用本发明的方法进行光源重建，结果如图5所示。可以看到，利用本章提出的方法可以准确的重建荧光光源，重建光源的中心位置为(22.37,50.68,13.15)mm，与真实光源的绝对距离为0.5612mm。
为了验证算法有效性，依然与基于L1正则化的重建方法进行了对比，结果如图6所示。利用L1正则化重建方法重建的光源中心位置为(23.01,50.06,14.87)mm，与真实光源中心距离为2.3239mm。
因此可以看出，基于L1正则化的重建方法在重建的荧光光源位于生物组织中较深位置时，重建效果较差，误差较大；而本章提出的方法可以准确地定位荧光光源。同时，基于 L1正则化方法重建的计算时间为1738.8秒，而本文的算法的计算时间仅为0.936秒。
将以上两组实验进行总结性量化比较，将本章提出的基于多任务压缩感知方法的重建算法简写为IRIRM，将基于L1范数正则化的重建算法简写为L1-RMRM。对比结果如表2所示。
表2 两种算法的量化比较
Figure PCTCN2015084555-appb-000044
实验二：
为了进一步测试本算法的有效性，开展了第二项实验：双光源的实验。
第一组：在第一组实验中，首先在肝脏和肺中各放置一个半径为1.5mm的球形光源，其中心位置分别为(14,51,16)mm和(21,51,16)mm，光源到生物体体表的距离分别为5.56mm和5.82mm。利用本发明的方法和基于L1正则化的方法分别进行重建，重建结果如图7所示。从图7中可以看到，两种算法都可以准确的重建荧光光源。
第二组：在第二组实验中，将两个光源均放置于肝脏中，利用上述两种算法进行光源重建，重建结果如图8所示。从图8可以看到，基于L1正则化的重建方法仅重建出了一个光源，而本发明提出的算法可以准确的重建两个光源，这进一步证明了本方法的有效性。
将以上两组双光源的实验进行总结性量化比较，对比结果如表3所示。
表3 双光源两种算法的量化比较
Figure PCTCN2015084555-appb-000045
在本发明中，利用小鼠形状的仿体，先后在肺部和肝脏中设置了单光源进行仿真实验，并将本发明提出的算法与基于L1正则化的重建算法进行对比。进一步地，又开展了双光源实验，分别用本发明的重建方法和基于L1正则化的重建算法进行光源重建，并对比实验结果。实验结果表明，本算法不但可以在小鼠体内对荧光光源进行准确重建定位，而且可以极大地提高计算效率，进一步验证了本算法的有效性。

Claims (1)

  1. 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法，其特征在于：首先采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，将多光谱之间的相关性作为先验信息融入重建算法中以降低BLT重建的病态性，最后在此基础上实现了荧光光源的三维重建；具体包括下述步骤：
    步骤一，问题定义及初始化——获得P个谱段在M个测量点上的值，并设置α0服从的伽马先验分布的形状参数a和尺度参数b，其中
    Figure PCTCN2015084555-appb-100001
    为出射光测量值Φ(τi)服从的高斯分布的方差；
    在模拟光在生物组织中传输规律时可以采用两种模型：辐射传输方程和扩散方程；辐射传输方程是一种精确描述光传播规律的数学模型，但是即使在非常简化的条件下RTE的求解仍然十分困难，且计算量非常巨大；扩散方程是辐射传输方程的低阶近似，因此其求解简单，但求解精度低；进一步，扩散方程在应用时必须满足生物组织强散射、低吸收的条件，而且不适用于光源位于生物组织边界时的情况，限制了扩散方程在生物医学领域中的应用；基于此，本方法选择高阶球谐近似方程；该模型能够较为精确的描述光在强吸收低散射的生物组织中传输，同时，它还保持着适中的计算复杂度；用该模型替代扩散近似传输方程以准确地模拟光在生物组织中的传输规律，以提高前向求解的精度；
    考虑到光谱影响，在波段为τi时，位置为r时，SP7模型共有四个方程：
    Figure PCTCN2015084555-appb-100002
    Figure PCTCN2015084555-appb-100003
    其中，S代表光源函数；μa是光吸收系数；μs是光散射系数；μai＝μa+μs(1-gi)，gi为各向异性因子；
    Figure PCTCN2015084555-appb-100004
    代表辐射度的勒让德矩(Legendre Moments)φi的线性组合，可表示为：
    Figure PCTCN2015084555-appb-100005
    在SP7方程的基础上，令φ6＝φ4＝0，并保留公式(1)和(2)，即可得到SP3方程：
    Figure PCTCN2015084555-appb-100006
    在SP7的边界条件的基础上，令φ6＝φ4＝0，得到SP3的边界条件：
    Figure PCTCN2015084555-appb-100007
    其中，
    Figure PCTCN2015084555-appb-100008
    代表垂直边界向外的单位法向量；Ai、Bi和Ci为一系列常数，与边界反射率的 角度距有关；对应的体表出射光Φ(r,τi)：
    Figure PCTCN2015084555-appb-100009
    其中，J0，…，J3为一系列常数；
    步骤二，搭建线性关系——基于有限元离散的SP3扩散方程方法，针对每一个波长搭建边界测量值与未知光源的关系模型；
    在公式(6)和(7)中，S(r,τi)表示的是荧光光源分布，是需要求解的未知量，为了使用有限元法对SP3方程及其边界条件进行求解，首先将SP3的两个方程写成对应的弱解形式：
    Figure PCTCN2015084555-appb-100010
    其中，Ψ(r,τi)是测试函数，公式(8)和(9)两个边界条件也要写为弱解形式：
    Figure PCTCN2015084555-appb-100011
    然后利用格林公式(Green’s Formula)将边界条件的弱解形式融入到SP3方程的弱解形式中；
    Figure PCTCN2015084555-appb-100012
    其中e1、e2、e3和e4分别为：
    Figure PCTCN2015084555-appb-100013
    Figure PCTCN2015084555-appb-100014
    Figure PCTCN2015084555-appb-100015
    Figure PCTCN2015084555-appb-100016
    上式中，W具有如下形式：
    进行网格剖分，将光源分布S(r,τi)写成只与谱段相关的S(τi)，选择相应基函数，合成整体的刚度矩阵Mij，将SP3方程离散为如下形式的矩阵方程：
    Figure PCTCN2015084555-appb-100018
    其中B为N×N大小的矩阵，
    Figure PCTCN2015084555-appb-100019
    和
    Figure PCTCN2015084555-appb-100020
    是是两个分块子矩阵，可表示为：
    Figure PCTCN2015084555-appb-100021
    这里需要建立出表面出射光Φ(τi)和未知荧光光源分布S(τi)之间的线性关系；在实验过程中，由于只能采集到生物体表面的光学信号，因此，只保留Φ(τi)中边界上节点值，去掉矩阵中不包含边界节点的行，得到方程：
    Figure PCTCN2015084555-appb-100022
    上述方程建立了未知荧光光源分布和边界出射光之间的关系，即：
    Φ(τi)＝A(τi)S(τi)
    在多光谱情况下，谱段的能量比ω(τi)通过预先的谱分析测得，令S代表所有谱段的光源的总能量，即
    Figure PCTCN2015084555-appb-100023
    p为谱段个数；将所有谱段的光源和边界测量值进行整合，多光谱情况下未知荧光光源分布和边界出射光之间的关系如下：
    AmulS＝Φmul
    此时，
    Figure PCTCN2015084555-appb-100024
    步骤三，共享先验估计——基于经验贝叶斯极大似然函数，推断表征多光谱相关的参数α；
    在实际问题中，出射光测量值Φ(τi)通常含有噪声，若噪声服从均值为零，方差为
    Figure PCTCN2015084555-appb-100025
    的高斯随机分布，则Φ(τi)关于荧光光源分布S(τi)和α0的极大似然函数表示为：
    Figure PCTCN2015084555-appb-100026
    为了表示谱段间关系，引入一种多任务的思想，将整个光源作为一个整体任务T；其中，S(τi)是T中的第i个任务模型，用Sj(τi)表示为S(τi)的第j个组成部分；若S(τi)服从均值为零的高斯先验分布，参数α0服从Gamma先验分布，即有：
    Figure PCTCN2015084555-appb-100027
    p(α0|a,b)＝Ga(α0|a,b)
    其中，α＝(α1,…αj,…αN)T是超参数，用来表示先验信息；那么，在所有的谱段中，αj的取值相同，即用α表征各个谱段之间的相关性；为了估计超参数α，使用经验贝叶斯的方法，得到α的边界最大似然函数，即有：
    Figure PCTCN2015084555-appb-100028
    其中，Bi＝I+A(τi)Λ-1A(τi)T，Λ-1＝diag(α1,α1,…,αN)；
    步骤四，未知光源重建——根据已知的超参数的估计
    Figure PCTCN2015084555-appb-100029
    和边界测量值Φmul，利用极大似然函数重建荧光光源；
    得到估计的
    Figure PCTCN2015084555-appb-100030
    后，再将所有谱段的光源和边界测量值进行整合，则未知光源S的大似然函数可表示为：
    Figure PCTCN2015084555-appb-100031
    这里，
    Figure PCTCN2015084555-appb-100032
    Figure PCTCN2015084555-appb-100033
    式中，
    Figure PCTCN2015084555-appb-100034
    本方法基于多任务学习方法，首先采用高阶近似模型模型光在生物组织中的传输规律，并基于多任务学习方法探索多光谱之间的内在相关性，最后在此基础上实现了荧光光源的三维重建。
PCT/CN2015/084555 2015-07-08 2015-07-21 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法 WO2017004851A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/108,394 US9672639B1 (en) 2015-07-08 2015-07-21 Bioluminescence tomography reconstruction based on multitasking bayesian compressed sensing

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510397424.0A CN104921706B (zh) 2015-07-08 2015-07-08 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法
CN201510397424.0 2015-07-08

Publications (1)

Publication Number Publication Date
WO2017004851A1 true WO2017004851A1 (zh) 2017-01-12

Family

ID=54109393

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/084555 WO2017004851A1 (zh) 2015-07-08 2015-07-21 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法

Country Status (3)

Country Link
US (1) US9672639B1 (zh)
CN (1) CN104921706B (zh)
WO (1) WO2017004851A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111759335A (zh) * 2020-05-29 2020-10-13 东软医疗系统股份有限公司 多能谱成像数据的获取方法、装置、电子设备、存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107392977B (zh) * 2017-08-22 2021-04-13 西北大学 单视图切伦科夫发光断层成像重建方法
CN109222900A (zh) * 2018-08-15 2019-01-18 北京交通大学 激发荧光断层成像重建的方法
CN110457782B (zh) * 2019-07-24 2023-05-09 西安工程大学 基于Marshall-Olkin扩展指数分布的可靠性分析方法
CN112037297B (zh) * 2020-07-29 2023-09-22 西北大学 一种基于网格重组的混合光传输建模方法
CN112089434B (zh) * 2020-10-16 2024-05-03 陕西师范大学 一种多光谱生物发光断层成像方法和系统
US11759162B2 (en) * 2021-06-11 2023-09-19 Canon Medical Systems Corporation Total SPECT scatter estimation and correction using radiative transfer equation

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070244395A1 (en) * 2006-01-03 2007-10-18 Ge Wang Systems and methods for multi-spectral bioluminescence tomography
CN101342075A (zh) * 2008-07-18 2009-01-14 北京工业大学 基于单视图的多光谱自发荧光断层成像重建方法
CN103271723A (zh) * 2013-06-26 2013-09-04 西安电子科技大学 一种生物发光断层成像重建方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101221128B (zh) * 2007-04-18 2010-11-24 中国科学院自动化研究所 一种基于自适应有限元的多光谱重建方法
WO2011137247A2 (en) * 2010-04-30 2011-11-03 The Trustees Of Columbia University In The City Of New York System, method and computer-accessible medium for performing attenuation-corrected multispectral luminescence tomography of cerenkov and bioluminescent light sources
CN102741707B (zh) * 2010-12-23 2014-04-23 中国科学院自动化研究所 基于切伦科夫效应的断层成像方法
WO2014081945A2 (en) * 2012-11-21 2014-05-30 The Trustees Of Columbia University In The City Of New York Systems, methods, and devices for image reconstruction using combined pde-constrained and simplified spherical harmonics algorithm

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070244395A1 (en) * 2006-01-03 2007-10-18 Ge Wang Systems and methods for multi-spectral bioluminescence tomography
CN101342075A (zh) * 2008-07-18 2009-01-14 北京工业大学 基于单视图的多光谱自发荧光断层成像重建方法
CN103271723A (zh) * 2013-06-26 2013-09-04 西安电子科技大学 一种生物发光断层成像重建方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111759335A (zh) * 2020-05-29 2020-10-13 东软医疗系统股份有限公司 多能谱成像数据的获取方法、装置、电子设备、存储介质

Also Published As

Publication number Publication date
US20170148193A1 (en) 2017-05-25
CN104921706A (zh) 2015-09-23
US9672639B1 (en) 2017-06-06
CN104921706B (zh) 2018-02-09

Similar Documents

Publication Publication Date Title
WO2017004851A1 (zh) 基于多任务贝叶斯压缩感知方法的生物发光断层成像重建算法
Ren et al. Frequency domain optical tomography based on the equation of radiative transfer
EP1514093B1 (en) Imaging volumes with arbitrary geometries in non-contact tomography
JP3310782B2 (ja) 吸収物質濃度の空間分布画像化装置
CN101342075B (zh) 基于单视图的多光谱自发荧光断层成像重建方法
Gorpas et al. A three-dimensional finite elements approach for the coupled radiative transfer equation and diffusion approximation modeling in fluorescence imaging
CN111915733B (zh) 基于LeNet网络的三维锥束X射线发光断层成像方法
CN105581779B (zh) 一种直接融合结构成像的生物发光断层成像重建的方法
Zhu et al. A three-dimensional finite element model and image reconstruction algorithm for time-domain fluorescence imaging in highly scattering media
CN106097441A (zh) 基于l1范数与tv范数的复合正则化生物发光断层成像重建方法
CN105559750A (zh) 组织结构引导的复合正则化生物发光断层成像重建方法
CN109924949A (zh) 一种基于卷积神经网络的近红外光谱断层成像重建方法
Slavine et al. Iterative reconstruction method for light emitting sources based on the diffusion equation
Nishimura et al. Determination of optical properties in double integrating sphere measurement by artificial neural network based method
Wojtkiewicz et al. Parallel, multi-purpose Monte Carlo code for simulation of light propagation in segmented tissues
CN112089434B (zh) 一种多光谱生物发光断层成像方法和系统
Martelli et al. Retrieval procedure for time-resolved near-infrared tissue spectroscopy based on the optimal estimation method
CN107374588B (zh) 一种基于同步聚类的多光源荧光分子断层成像重建方法
Slavine et al. Semi-automated image processing for preclinical bioluminescent imaging
JP3730927B2 (ja) 吸収物質濃度空間分布画像化装置
JP3276947B2 (ja) 吸収物質濃度空間分布画像装置
Wang An EM‐like reconstruction method for diffuse optical tomography
CN114159021B (zh) 基于双输入-单输出深度学习的切伦可夫激发的荧光扫描断层成像重建方法
Chen et al. Cone Beam X‐Ray Luminescence Tomography Imaging Based on KA‐FEM Method for Small Animals
Guo et al. Iterative reconstruction for bioluminescence tomography based on an adaptive region shrinking strategy

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 15108394

Country of ref document: US

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15897507

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15897507

Country of ref document: EP

Kind code of ref document: A1