Motion correction and registration of high b-value diffusion weighted images

It has been suggested that, high b-value diffusion weightedMRI improves the sensitivity and specificity of these imagesto tissue microstructure when compared with ‘‘clinical’’ b-value diffusion weighted MRI (b  1000 s/mm2). However, itsuffers from poor signal to noise ratio - leading to longer ac-quisition times and therefore more motion artifacts. Togetherwith the orientational sensitivity of the diffusion weighted MRIsignal, the contrast at different b-values and different gradientdirections is significantly different. These features of high b-value diffusion images preclude the ability to perform conven-tional image-registration-based motion/distortion correction.Here, we suggest a framework based on both experimentaldata (diffusion tensor MRI) and simulations (using the com-posite hindered and restricted model of diffusion framework)to correct the motion induced misalignments and artifacts ofhigh b- value diffusion weighted MRI. This approach was eval-uated using visual assessment of the registered diffusionweighted MRI and the composite hindered and restrictedmodel of diffusion analysis results, as well as residual analy-sis to assess the quality of the composite hindered and re-stricted model of diffusion fitting. Both qualitative andquantitative results demonstrate an improvement in fitting thedata to the composite hindered and restricted model of diffu-sion model following the suggested registration framework,thereby, addressing a long-standing problem and making thecorrection of motion/distortions in data collected at high b-values feasible for the first time. Magn Reson Med 67:1694–1702, 2012.VC2011 Wiley Periodicals, Inc.