Modified triexponential analysis of intravoxel incoherent motion for brain perfusion and diffusion

Abstract

Background To noninvasively obtain more detailed information on brain perfusion and diffusion using modified triexponential analysis. Methods On a 3.0 Tesla MRI, diffusion‐weighted imaging of the brain with multiple b‐values was performed in healthy volunteers (n = 12). We derived perfusion‐related, fast‐free, and slow‐restricted diffusion coefficients (Dp, Df, and Ds, respectively) and fractions (Fp, Ff, and Fs, respectively) in the frontal and occipital white matter, caudate nucleus, and putamen calculated from triexponential function by a two‐step approach. Ds was initially determined using monoexponential function in b‐values over 1000 s/mm2 and was applied to triexponential function. Additionally, the literature value of the diffusion coefficient of free water at 37 °C was assigned to Df. Finally, Dp and fractions were derived using all b‐values. Moreover, biexponential analysis was performed and compared with triexponential analysis. We also determined regional cerebral blood flow (rCBF) using arterial spin labeling and assessed its relation with each diffusion parameter. Results Significant positive correlations between Dp and rCBF were found in the caudate nucleus (R = 0.84; P = 0.01) and putamen (R = 0.86; P = 0.01), whereas no diffusion parameters were significantly correlated with rCBF on biexponential analysis (P > 0.05 for all). Conclusion Diffusion analysis with triexponential function enables noninvasive gathering of more detailed information on brain perfusion and diffusion. J. Magn. Reson. Imaging 2016;43:818–823

A robust deconvolution method to disentangle multiple water pools in diffusion MRI

“…14,23,24 Recent work 22 has shown that both the pseudo-diffusion and FW compartments should be considered when including the non-weighted image in the model fitting. This suggests the need for at least three compartments for an accurate disentanglement of partial volume effects in the brain, which is in line with other studies on brain 25 and non-brain [26][27][28] tissue. However, the appropriate fit of multiple exponentials with non-linear leastsquares (NLLS) approaches is a difficult and time-consuming numerical problem.…”

“…26,27,58 Furthermore, the diffusion coefficients of the three observed components were in agreement with previously reported literature values of brain tissue, FW and blood pseudo-diffusion. 3,8,25 Recent studies have suggested the existence of multiple components in the pseudo-diffusion part itself (i.e. micro-and macrovascular network).…”

“…22 The three-exponential model was fitted with NLLS and with a segmented NLLS fit, as suggested previously. 25,56 Then, the average signal fraction and diffusion coefficients were computed in GM and WM for comparison with P-L2-NNLS.…”

On the Field Strength Dependence of Bi‐ and Triexponential Intravoxel Incoherent Motion (IVIM) Parameters in the Liver

“…The decision for fitting a triexponential model was made because of recent reports indicating that a triexponential model might be more appropriate. 11,12,[17][18][19] The formula for the biexponential IVIM model reads 20 :…”

Reply to: On the perils of multiexponential fitting of diffusion MR data

“…We would like to thank the authors of the letter for their interest and comments on our recent article. 1 They have questioned our fitting procedure in which the fast-free diffusion coefficient (D f ) was fixed to the diffusion coefficient of free water at 378C (3.0 3 10 23 mm 2 /s). Moreover, they have interpreted that the fast and slow diffusion component fractions (F f and F s , respectively) correspond to the extracellular and intracellular water fractions, respectively, based on the results of previous studies [2][3][4][5] and showed that their fractions depended on whether the diffusion analysis was performed with constrained or unconstrained fits.…”