Diffusion spectrum imaging(DSI),a newly developed MRI technique,affords the capacity to map complex fiber architectures in tissues with sufficient angular resolution by imaging the spectra of tissue water diffusion.By contrast,diffusion tensor imaging(DTI),the currently widely used technique based on the 2nd order tensor model,obtains an approximation of the complex diffusion,and provides only one global maximal direction corresponding to the primary eigenvector for each voxel.As a generalized model-free diffusion imaging technique,firstly,DSI employs the probability density function to describe the diffusion process in each voxel;secondly,a sufficient dense signal sample derived from repeated applications of diffusion-weighed gradients ensures its capability to resolve the diffusion probability density function;thirdly,specific computer visualization techniques are used to extract the diffusion information and reconstruct the geometrical properties of tissue microstructure.The capacity to unravel complex tissue architecture,recent improvements in hardware and ongoing optimization of sequence design and algorithm enable a rapid growth of DSI for research use and future incorporation into clinical protocols.This paper introduces the basic principles of DSI and then compares the characteristics of DSI and DTI schemes.Finally,the typical applications of DSI to date are reviewed.Abstract:SUMM ARY D iffusion spectrum imaging(DSI),a newly developed MR I technique,affords the capacity to map complex fiber architectures in tissues with sufficient angular resolution by imaging the spectra of tissue water d iffusion.By contrast,d iffusion tensor imaging(DTI),the currently widely used technique based on the 2nd order tensormodel,obtains an approximation of the complex d iffusion,and provides on-ly one globalmaximal d irection correspond ing to the primary eigenvector for each voxel.As a generalized model-free d iffusion imaging technique,firstly,DSI employs the probability density function to describe the d iffusion process
