Research On Hyperspectral CT Image Reconstruction And Material Identification Method

Medical imaging is a study on how to obtain anatomical and physiological information of human,which provides a basis for diagnosis and evaluation.The image model can be divided into two categories:one is to provide the structure of anatomical information imaging;the other is to provide functional imaging of physiological information.Some cases could not be diagnosed by a single imaging category.Different imaging methods are often able to complement information,which results in the recent rapid development of multi-mode medical imaging technology.Traditional CT is typical,structural imaging,and noninvasive lesion positioning must be combined with component information analysis.In this paper,based on the CT imaging,hyperspectral reconstruction and identification were introduced,which combined the advantages of high spatial resolution of CT and the ability of high spectral detection to realize the integration of structure and component analysis.At present,the mainstream dual energy,multi-energy CT was divided by the X-ray tube voltage.The broadband light of them has a large overlap,which will bring some interference in material identification.Based on the present research situation,this paper presented a Hyperspectral X-ray CT system based on photon counting principle,which can realize the identification of infinite species by hyperspectral coding and achieve the purpose of fine division of functional attributes.In this study,a sample containing ten kinds of plastic was carried out for recognition.First,the structural information of complex samples was obtained by CT reconstruction.Then the region of interest was identified from the structural information.Through the spectral reconstruction of the voxels in the region of interest,the absorption spectrum information is obtained,and the material classification is identified.The success rate of the material identification is high.After hyperspectral CT scan of the sample,six methods were used to obtain the absorption spectra of the region of interest,which included Filter Back Projection,Filter Back Projection based on Beam-hardening Correction,Algebraic Reconstruction Technique,Algebraic Reconstruction Technique based on Beam-hardening Correction,Simultaneous Algebraic Reconstruction Technique,Simultaneous Algebraic Reconstruction Technique based on Beam-hardening Correction.The results showed that the Algebraic Reconstruction Technique based on Beam-hardening Correction could achieve the highest recognition accuracy.Filtered Back Projection algorithm had the fastest computing speed but the worst identification accuracy.The reconstruction of X-ray spectra can be used for nondestructive analysis of structure and composition of samples,which has promising prospects in biomedical engineering research,clinical diagnosis,non-destructive testing and so on.At the same time,it will provide reference for the development of new radiation imaging equipment and new photon counting line/surface array detector.