Optical Imaging Based On The Broadband Time-resolved Diffuse Light For The Biological Tissue

With the progress of physics,scientific optimization algorithms,instruments and equipments in the recent decades,the exploration of disease diagnosis is investigated deeply,which promotes the development of the medical and industrial disciplines.Diffuse optical imaging employs visible and near-infrared light can be employed for non-invasive biomedical diagnosis of the biological tissue.analyzes the detected radiative signals by studying the interaction between the light and tissue,obtaining structural or functional information related to the internal optical parameters.Nowadays,the diffuse optical imaging is widely studied based on the diffusion approximation(DA)model.The reconstruction based on the complete radiative transfer model is complex.However,it has wider application to detect the cerebrospinal fluid,synovial fluid and other tissues that the DA model fails to take effect.In addition,the reconstruction based on the single mode causes limitations.Meanwhile,when it is extended to multi-spectral detection,the current spectroscopy research is mostly based on the time-resoved spectroscopy measurement for limited spectra,and there are few studies about the broadband time-resoved spectroscopy measurement.The topic of the thesis is aimed for optical parameter reconstruction of tissues.An accurate and effective reconstruction algorithm based on inverse algorithm and regularization method is proposed for the reconstruction employing radiative transfer model.The reconstruction is carried out in time-domain,frequencydomain and hybrid time-frequency domain.Broadband time-resoved spectroscopy measurements are carried out based on the phantoms,biological tissues.The main research is carried out in the following aspects:Firstly,the forward model and the inverse optimization algorithm based on the gradient are summarized.For the scattering-dominated media,the photon migration inside the tissue is described by DA model.The reconstrucion is carried out employing the two-layer TD-DA model and the SQP algorithm.The results show that the reflection signals are not sensitive to the scattering parameters of the bottom layer,thus it’s difficult to reconstruct this parameter.However,the reconstruction accuracy can be improved by selecting the appropriate objective function,using the whole time-varying signal waveform and adopting the multi-distance signals,which is helpful to the reconstruction of optical parameter.Secondly,the time-domain(TD)radiative transfer equation(RTE)is employed for the reconstrucion of the non-scattering dominated media,the media containing void-like region,which the DA model fails to work.The discrete coordinate method is employed to solve the TD-RTE and the sequence quadratic programming(SQP)is adopted for the inverse optimization method.For this simultaneous reconstruction of multiple optical parameter fields,the ill-posed problem is more serious and the reconstruction is more difficult.Thus,the adjoint equation method(AEM),finite difference method and complex variable difference gradient method are introduced to calculate the gradient of the objective function.Furthermore,three regularization terms are introduced based on the bayesian theory and maximum a posteriori probability theory as Tikhonov method,Generalized Gaussian Markov Random Field,total variation method for the optical parameter reconstruction.The results show that the gradient-based optimization algorithm combining the AEM and regularization method can reconstruct the optical parameter fields accurately and effectively.Considering that the geometrical shapes of tissues have obvious differences in practical diagnostic applications,the finite element method(FEM)is introduced to reconstruct the optical parameter fields of the medium with complex shape.The FEM combined with Open MP is used to solve the TD-RTE,and the optical parameter fields are reconstructed based on SQP and AEM.Then the diffusion synthesis acceleration(DSA)method is introduced to accelerate and improve the FEM calculation,and the optical parameter field is reconstructed based on the improved DSA model.The results show that the modified DSA acceleration model improves the reconstruction efficiency and is suitable for complex shape for near-infrared(NIRS)optical imaging.In order to improve the reconstruction efficiency and accuracy of radiative transfer model,a multi-mode optical parameter fields reconstruction method based on hybrid time-frequency domain is introduced.On the basis of frequency domain reconstruction,the optical parameter fields are further modified and updated based on threshold segmentation method and the time domain reconstruction.The results of the hybrid time-frequency model are compared with those of frequency domain and time domain.The results show that the hybrid model is more efficient and accurate than the original time-domain model,and the reconstructed image can show the internal structure of the medium clearly.Finally,the broadband time-resoved diffuse optical spectroscopy(DOS)platform is used to measure the scattering-dominated phantoms and tissues.In-vitro and in-vivo measurement are carried out at 600-1100 nm.The spectral characteristics and components are analyzed to obtain the content of oxygen saturation,moisture,lipids,collagen and other key components in the tissues,thus achieving bio-medical functional diagnosis.The broadband time-resoved spectroscopy of heterogeneous two-layer media is studied with different thicknesses.The validity and reliability of the proposed two-layer model are verified by the experimental studies.