Flow‐compensated intravoxel incoherent motion diffusion imaging

<jats:sec><jats:title>Purpose</jats:title><jats:p>The pseudo‐diffusion coefficient <jats:italic>D<jats:sup>*</jats:sup></jats:italic> in intravoxel incoherent motion (IVIM) imaging was found difficult to seize. Flow‐compensated diffusion gradients were used to test the validity of the commonly assumed biexponential limit and to determine not only <jats:italic>D<jats:sup>*</jats:sup></jats:italic>, but also characteristic timescale <jats:italic>τ</jats:italic> and velocity <jats:italic>v</jats:italic> of the incoherent motion.</jats:p></jats:sec><jats:sec><jats:title>Theory and Methods</jats:title><jats:p>Bipolar and flow‐compensated diffusion gradients were inserted into a flow‐compensated single‐shot EPI sequence. Images were obtained from a pipe‐shaped flow phantom and from healthy volunteers. To calculate the IVIM signal outside the biexponential limit, a formalism based on normalized phase distributions was developed.</jats:p></jats:sec><jats:sec><jats:title>Results</jats:title><jats:p>The flow‐compensated diffusion gradients caused less signal attenuation than the bipolar ones. A signal dependence on the duration of the flow‐compensated gradients was found at low b‐values in the volunteer datasets. The characteristic IVIM parameters were estimated to be <jats:italic>v</jats:italic> = 4.60 ± 0.34 mm/s and <jats:italic>τ</jats:italic> = 144 ± 10 ms for liver and <jats:italic>v</jats:italic> = 3.91 ± 0.54 mm/s and <jats:italic>τ</jats:italic> = 224 ± 47 ms for pancreas.</jats:p></jats:sec><jats:sec><jats:title>Conclusion</jats:title><jats:p>Our results strongly indicate that the biexponential limit does not adequately model the diffusion signal in liver and pancreas. By using both bipolar and flow‐compensated diffusion gradients of different duration, the characteristic timescale and velocity of the incoherent motion can be determined. Magn Reson Med 74:410–419, 2015. © 2014 Wiley Periodicals, Inc.</jats:p></jats:sec>