A THEORETICAL-STUDY OF THE EFFECT OF EXPERIMENTAL NOISE ON THE MEASUREMENT OF ANISOTROPY IN-DIFFUSION IMAGING

Diffusion tensor imaging (DTT) is a modality known to be highly sensit ive to the detrimental effects of experimental noise. Here, using Mont e Carlo simulations, we compare and contrast how noise complicates the measurement of diffusion anisotropy in diffusion tensor and conventio nal diffusion-weighted imaging (DWI). As the signal-to-noise ratio (SN R) decreases below a value of approximately 20, the eigenvalues (lambd a(i)) of the diffusion tensor D are found to diverge rapidly from thei r true values, with the result that the measured anisotropy can be sig nificantly in error and isotropic structures falsely assigned a high l evel of anisotropy, The effect of noise on the rotationally variant in dices, calculated from a conventional diffusion-weighted imaging exper iment, is found to be much less insidious, because the apparent diffus ion coefficients (ADCs) diverge only slowly as the signal-to-noise dec reases. Thus, although rotationally variant indices almost always unde restimate the true diffusion anisotropy, they show only a small suscep tibility to experimental noise and hence, are preferred to their rotat ionally invariant counterparts when the signal-to-noise ratio is small . (C) 1998 Elsevier Science Inc.