Perfusion MRI or perfusion-weighted imaging is perfusion scanning by the use of a particular MRI sequence. The acquired data are then post-processed to obtain perfusion maps with different parameters, such as BV, BF, MTT and TTP.
Arterial spin labelling : Magnetic labeling of arterial blood below the imaging slab, without the need of gadolinium contrast
It can also be argued that diffusion MRI models, such as intravoxel incoherent motion, also attempt to capture perfusion.
Dynamic susceptibility contrast
In Dynamic susceptibility contrast MR imaging, Gadolinium contrast agent is injected and a time series of fast T2*-weighted images is acquired. As Gadolinium passes through the tissues, it induces a reduction of T2* in the nearby water protons; the corresponding decrease in signal intensity observed depends on the local Gd concentration, which may be considered a proxy for perfusion. The acquired time series data are then postprocessed to obtain perfusion maps with different parameters, such as BV, BF, MTT and TTP.
Dynamic contrast-enhanced imaging
Dynamic contrast-enhanced imaging gives information about physiological tissue characteristics. For example, it enables analysis of blood vessels generated by a brain tumor. The contrast agent is blocked by the regular blood–brain barrier but not in the blood vessels generated by the tumor. The concentration of the contrast agent is measured as it passes from the blood vessels to the extracellular space of the tissue and as it goes back to the blood vessels. The contrast agents used for DCE-MRI are often gadolinium based. Interaction with the gadolinium contrast agent causes the relaxation time of water protons to decrease, and therefore images acquired after gadolinium injection display higher signal in T1-weighted images indicating the present of the agent. It is important to note that, unlike some techniques such as PET imaging, the contrast agent is not imaged directly, but by an indirect effect on water protons. The common procedure for a DCE-MRI exam is to acquire a regular T1-weighted MRI scan, and then gadolinium is injected before further T1-weighted scanning. DCE-MRI may be acquired with or without a pause for contrast injection and may have varying time resolution depending on preference – faster imaging allows pharmacokinetic modelling of contrast agent but can limit possible image resolution. Slower time resolution allows more detailed images, but may limit interpretation to only looking at signal intensity curve shape. In general, persistent increased signal intensity in a DCE-MRI image voxel indicates permeable blood vessels characteristic of tumor tissue, where Gd has leaked into the extravascular extracellular space. In tissues with healthy cells or a high cell density, gadolinium re-enters the vessels faster since it cannot pass the cell membranes. In damaged tissues or tissues with a lower cell density, the gadolinium stays in the extracellular space longer. Pharmacokinetic modelling of gadolinium in DCE-MRI is complex and requires choosing a model. There are a variety of models, which describe tissue structure differently, including size and structure of plasma fraction, extravascular extracellular space, and the resulting parameters relating to permeability, surface area, and transfer constants. DCE-MRI can also provide model-independent parameters, such as T1 and area under the gadolinium curve, which may be more reproducible. Accurate measurement of T1 is required for some pharmacokinetic models, which can be estimated from 2 pre-gadolinium images of varying excitation pulse flip angle, though this method is not intrinsically quantitative. Some models require knowledge of the arterial input function, which may be measured on a per patient basis or taken as a population function from literature, and can be an important variable for modelling.
Arterial spin labelling
has the advantage of not relying on an injected contrast agent, instead inferring perfusion from a drop in signal observed in the imaging slice arising from inflowing spins having been selectively saturated. A number of ASL schemes are possible, the simplest being flow alternating inversion recovery which requires two acquisitions of identical parameters with the exception of the out-of-slice saturation; the difference in the two images is theoretically only from inflowing spins, and may be considered a 'perfusion map'.
