Perfusion magnetic resonance imaging of the liver

Table 1 Illustrative example of a perfusion MRI sequence performed on a 1.5 T MR platform

MRI platform	Avanto (Siemens, Erlangen, Germany)
Type of pulse sequence	3D FLASH
TR	2.72 ms
TE	1 ms
Partition thickness	8 mm
Slices per slab	10
Matrix	256 × 159
Phase encode direction	Anterior to posterior
Number of averages	1
Sensitivity encoding factor	2
Flip angle before contrast	2º and 14º
Flip angle after contrast	14º
Bandwidth	490 Hz
RF spoiling	Yes
Temporal resolution	1.98 s per slab of 10 slices.
Precontrast scans	10 measurements of each flip angle averaged for calculation of native T1
Gadolinium injection	0.2 mmol/kg at 3 mL/s followed by 20 mL flush
Patient respiration	Quiet breathing
Post contrast scans	A total of 180 consecutive measurements. Inject contrast only when the 20th measurement has been completed
Scan sections to use for processing	Center 6 image sections only

MRI: Magnetic resonance imaging; FLASH: Fast low-angle shot; TR: repetition time; TE: Echo time.

Table 2 Examples of the types of tracer kinetic models that have been applied for perfusion MRI of the liver

Study	Diseases	Comment
Single-input, single compartment, CC model
Scharf et al[48]	Preclinical study in pigs	Experimental model of normal liver in pigs. Only arterial input from hepatic artery taken into account. Such a model may lack physiological realism, especially when there is substantial vascular input contribution from the portal vein
Single-input, dual-compartment, DP model
Sahani et al[30]	HCC	Single input assumed because majority of vascular input to HCC is derived from hepatic artery. Dual-compartment model used to probe interstitial space and PS, which can be substantial in tumors. DP model implemented as standard on General Electric (GE) perfusion software 2.0 used for analysis
Dual-input, single-compartment, CC model
Materne et al[45-47]	Normal and cirrhotic livers	Assumption of single compartment based on understanding that the fenestra in the sinusoids of liver are extremely porous and allows free exchange of low-molecular-weight contrast tracers between the vascular and the sinusoidal interstitial space. To simplify calculations, assumption was made that there was instantaneous mixing of contrast medium from the dual input[6,7,27,45-47,49] within the single compartment. In this way, quantitative parameters such as arterial perfusion, portal venous perfusion, MTT and volume of distribution (Ve) could be derived. Cuenod et al[27,49] applied a deconvolution technique to fit these parameters, and variants of such a model were also used by Funabasam et al[50] and Miyazaki et al[51]
Cuenod et al[27,49]	Metastatic disease
Dual-input, dual-compartment, DP model
Koh et al[4,20]	Metastases, HCC and cirrhosis	The DP model applies a concentration gradient within the vascular space. Parameters derived include, arterial flow, portal venous flow, fractional arterial flow, permeability, fractional intravascular space, fractional interstitial space, MTT, contrast arrival time. A dual-input dual-compartment approximation of the DP model is used commercially (CT Perfusion 3.0; General Electric, Milwaukee, USA) and was also adopted by Chen et al[42,52,53] in perfusion studies of the liver

CC: Conventional compartment; HCC: Hepatocellular carcinoma; DP: Distributed parameter; MTT: Mean transit time; CT: Computed tomography.