Effects of Turboprop diffusion tensor imaging acquisition parameters on the noise of fractional anisotropy

Introduction: Traditionally, DTI data is acquired with EPI because of its short imaging time. However, EPI-based DTI suffers from eddy-current and magnetic field inhomogeneity-related artifacts. Turboprop is characterized by low sensitivity to eddy-current and B0 effects, and is gaining popularity in DTI studies of the brain. In EPI-based DTI, it has been shown that, measuring diffusion weighted (DW) signals in multiple diffusion directions instead of acquiring multiple copies of few DW images, reduces the noise in fractional anisotropy (FA). In addition, for anisotropic diffusion, using acquisition schemes with less than approximately 30 diffusion directions causes the noise in FA to be dependent on the orientation of the primary eigenvector of the tensor. Due to the k-space sampling pattern used in Turboprop, where the central region of k-space is sampled by all blades, while the periphery is sampled by fewer or even one blade, the concept of acquiring a “single” or “multiple” copies of a DW image is not well-defined. One can produce a single Turboprop image using, for example, 16 or 32 blades. Also, the echo-train length (ETL) and turbo-factor affect the number of lines in a blade, the number of blades required to cover kspace, the signal decay in a blade, as well as the amount of data collected, and thereby the noise in FA. Therefore, the goal of this study was to investigate the effect of the number of blades, ETL, turbo-factor, and number of diffusion directions on the noise of FA in Turboprop-DTI.