Toward optimized intravoxel incoherent motion (IVIM) and compartmental T2 mapping in abdominal organs

Purpose: To quantitatively assess the bias in the intravoxel incoherent motions (IVIM)-derived pseudo-diffusion volume fraction (f) caused by the differences in relaxation times between the tissue and fluid compartments, and to develop a 2D fitting approach and an optimal acquisition protocol for the relaxation compensated T2-IVIM imaging in the liver and kidney. Methods: Numerical simulations were conducted to investigate the TR- and TE-dependent bias in f when using the conventional IVIM model, and to evaluate the applicability of the extended 2D T2-IVIM model for reducing this bias. The in silico findings were then validated using the in vivo IVIM data from healthy volunteers on a clinical 3-Tesla MRI scanner. Finally, a numerical framework for optimizing the T2-IVIM protocol for relaxation-compensated f parameter estimation was proposed and tested using in vivo data. Results: When using the traditional IVIM model, a trend toward higher f with increasing TE was found in the liver (R = 0.42, P = 0.043), but not in the kidney cortex (R = -0.067, P = 0.76) and medulla (R = 0.039, P = 0.86). The 2D T2-IVIM modeling yielded lower f and reduced the intra-subject variability in the liver. Our results also suggest that a b-TE protocol with six b-values and three different TE values (50, 55, and 100 ms) might be optimal for liver T2-IVIM. Conclusion: The extended 2D T2-IVIM model combined effectively minimizes the TE-dependent bias in f and allows simultaneous estimation of the IVIM parameter and compartmental T2 values in the liver and kidney.