Diffusion Tensor Imaging (DTI) of localized anatomical regions (i.e brainstem, cervical spinal cord, and optic nerve) is challenging because of the existence of significant susceptibility differences in the surrounding tissues, their high motion sensitivity and the need for high spatial resolution to resolve the underlying complex histoarchitecture. The aim of the present methodology is to achieve high resolution DTI with motion compensating capability in localized regions of the central nervous system. We accomplish this by implementing self-navigated multi-shot variable density spiral encoding with outer volume suppression. In vivo application of the technique on the human brainstem demonstrates a clear delineation of the multiple local neural tracts. We also investigate the partial volume effect on the extracted diffusion anisotropy metrics by varying the in-plane resolution while maintaining a constant signal-to-noise ratio.