TY  - GEN
AV  - public
A1  - Ma, Bo
TI  - Effect of Gradient Vectors Scheme and Noise Correction on Fractional Anisotropy in Diffusion Tensor Imaging of the Peripheral Nervous System
N2  - Diffusion Tensor Imaging (DTI) is a method widely used in research and clinic, especially for imaging and connectivity analysis of the white brain matter. Despite the many possibilities offered by DTI, this method suffers from an inherently low signal-to-noise ratio (SNR), since both the long echo time and the diffusion gradients weaken the signal. The SNR is particularly low at high spatial resolution, e.g. in the DTI of nerves. A low SNR leads to systematic and statistical errors in parameters calculated from the DTI, e.g. fractional anisotropy (FA). A low SNR can be partially compensated by increasing the number of diffusion directions or using methods for a posteriori noise correction. The most robust method for anatomical structures with unknown orientation is to distribute the diffusion gradients evenly in space. However, if the preferred direction of the anatomical structure is known in advance, it may be advantageous to limit the diffusion gradients to a cone centered on the axis of the structure. The aim of this work was to develop a DTI method with high accuracy and reliability for application in peripheral nerves. Two methods to reduce image noise were investigated: (1) A newly developed scheme of diffusion gradient vectors (DGV), where the vectors are restricted to a cone with an aperture angle Theta around the axis of the nerve and (2) different methods for a posteriori noise correction. 
For this purpose, Monte Carlo simulations were performed based on realistic values for diffusivity, FA and noise obtained from clinical investigations and studies. Furthermore, the methods were tested in a specially designed phantom simulating diffusion in peripheral nerves (FA = 0.65). These investigations were performed on a 3 Tesla whole-body magnetic resonance (MR) scanner. To determine the accuracy and reliability of the DTI using the appropriate measurement or correction procedures, systematic deviations of FA from baseline and the statistical error of FA were measured. The newly developed DGV scheme with limited space coverage was compared with gradient schemes with uniform space coverage (Jones, Downhill Simplex Method (DSM), gradient scheme of the manufacturer) based on their condition number (CN).
The study showed that with the newly developed DGV scheme FA can be measured with high accuracy when the angle Theta is at least 45° or 60°. The minimum Theta depends on the number of gradient directions and on FA. Basically, the higher the FA value and the greater the number of gradients, the better the accuracy of the DGV scheme. For N = 30, the DGV allowed an exact determination of FA for the entire FA range (0.4 - 0.8) investigated in this study, if Theta ?45° was. It could be shown that when using the new DGV scheme, a slight inclination of the investigated structure (?30°) does not affect the accuracy of FA. CN of the developed DGV-scheme was higher than CN of the Jones-scheme and the DSM-scheme for N = 6; for N?10 CN of the new DSM-scheme was lower than that of the Jones-scheme. However, it is also not to be expected that a method that concentrates the gradient vectors on a limited segment of space is as insensitive to interference as schemes with uniform gradient distribution. Nevertheless, the CN of the new DGV method was in the same order of magnitude as that of the other methods.
A comparison of the different a posteriori correction methods showed that the power image method is the most effective and robust method and compensates for both the systematic and statistical errors of FA. The efficiency of the power image method is independent of the number of diffusion gradients used. In addition, the method works reliably - regardless of the method used for the coil combination (square sum versus adaptive combination). In contrast, both correction factor methods used in this study were less efficient in terms of noise correction; furthermore, the correction efficiency depended on the coil combination method.
In conclusion, a combination of the newly developed DGV scheme with the power image method for a posteriori correction allows DTI of peripheral nerves with high SNR, high accuracy and reliability of the calculated parameters (e.g. FA) without the need for additional acquisition time.  So far, however, these newly developed and tested methods have not yet been applied in studies or clinical trials.
CY  - Heidelberg
Y1  - 2020///
ID  - heidok28484
UR  - https://archiv.ub.uni-heidelberg.de/volltextserver/28484/
ER  -