Measurement of diffusion fractional anisotropy with T2 contrast to reduce noise

The purpose is to improve the signal-to-noise ratio in fractional anisotropy of the diffusion tensor without the acquisition of the low-diffusion-weighted image. The fractional anisotropy (FA) was reconstructed from the six diffusion-weighted images, instead of the diffusion tensor elements. The contribution of the spin-spin relaxation and proton density was estimated and could be excluded. A linear relationship of signal attenuation from the diffusion process was further assumed. Monte Carlo simulations were performed using a data-set of 30 diffusion anisotropy levels and 100 proton density levels. Gaussian-distributed noises of 20 levels were added. Human images were acquired from 6 subjects. The original FA and the proposed indices were calculated. Our results show that the new indices increased linearly with the diffusion anisotropy. The indices had a comparable contrast with the original FA, but with improved noise level and reduced acquisition time. The FA map can be calculated without the estimation of the contribution from spin-spin relaxation and the proton density. The index can be calculated under the assumption of linear signal attenuation. The minimum number diffusion-weighted images for measuring FA is six, instead of seven.