Analytical Expressions for Noise Propagation in Diffusion Tensor Imaging
Shen Y, Pu I, Clark C
St. George's, University of London
Parameters derived from diffusion tensor imaging (DTI) are sensitive to errors arising from noise in diffusion-weighted imaging (DWI) data. We present an analytical determination of errors arising due to noise in the eigenvalues and eigenvectors and a range of tensor metrics calculated from these parameters. In particular, an appraisal of errors in the principal eigenvector has implications for tractography, which utilises this information to map the white matter pathways of the brain. The derived expressions for the errors in the tensor metrics can be used to calculate the errors encountered for any tensor and signal to noise ratio (SNR) conditions.