A Simulation Study On Methodological Feasibility Of Diffuse Raman Tomography

Raman imaging technique can provide specific spectral fingerprints and high spatial resolution images,and has been widely used in cancer and precancer detection and evaluation.However,the typically adopted raster-scanning way seriously hampers the measuring speed,imaging depth and detection area,which makes Raman imaging hard to meet the requirement of the in vivo three-dimensional(3D)bio-information extraction.Diffuse Raman Tomography(DRT)is an advanced label-free optical imaging technique,which combines the fingerprint characteristic of Raman spectroscopy and the high depth resolution of diffuse optical tomography.Consequently,it is capable of providing 3D molecular evaluation of in vivo tissues,effectively improving the topology-based scanning.The established DRT modalities have achieved 3D imaging of a specific biochemical component in the large-scale heterogeneous tissue at a single wavelength.For the skin and other superficial tissue model,however,it is essential to pay more attention to the multi-wavelength implementation and experimental conditions that enables DRT.This research presents a simulation study on methodological feasibility of DRT for the superficial optical model.Firstly,a graphics processing unit(GPU)-based reverse tracking Raman Monte Carlo(RMC)simulation is proposed to improve the efficiency of the DRT forward solution.The reverse tracking scheme is according to the Green reciprocity and further parallelized by GPU.The simulation results show that this new method effectively improve the DRT computational efficiency,which takes only 1/6×10~7 of the traditional model computational cost and 1/380 of the GPU-based traditional model computational cost(1?10~7 emitted photons).Next,by establishing a 3-layer heterogeneous model composed b y DNA,protein,lipid,carotene and calcium,the objective data of multi-band under different experimental conditions are obtained by RMC method.Then,in order to retrieve the spatial concentration distributions of various biochemical components,the“Four dimensions”data reconstructed by DRT is used to spectral extracting and fitting in voxel-by-voxel.It is proved,compared with the original model,that DRT method has effectively obtained a spatial density image with a 4 mm depth and is suitable for in vivo 3D functional reconstruction of a superficial tissue.Finally,a series of simulation experiments on different experimental conditions,such as incident light intensity,measuring time,and source-detector distribution are analyzed.Overall,this research provides a valuable feasibility study for the further development of DRT technique.