Fig. 1

Illustration of the proposed methodology. Top: The SNR was varied by adding complex white noise to the data. Magnitude images were passed to the tensor fitting algorithm. Middle: The diffusion encoding scheme allowed for the truncation of directions while maintaining an approximately uniform distribution. A dataset with the desired number of directions can be generated by selecting the first ND images. Bottom: The resolution was varied by either truncating the 3D k-space data (3D–DS), or by first applying slice-selection and then truncating the 2D k-space data (2D–DS)