Skip to content

What is DCE-MRI

Dynamic contrast-enhanced MRI comprises a single 4D acquisition in which one T1-weighted image is repeatedly acquired before, during and after intravenous administration of a gadolinium-based contrast agent. A typical protocol acquires 5 to 100 frames at a temporal resolution of 2 to 60 seconds, covering five to fifteen minutes in total.

Gadolinium shortens the longitudinal relaxation time and therefore increase signal intensity on T1-weighted sequences. In DCE signal intensity therefore increases where contrast arrives and declines as it clears. Each voxel yields a signal-versus-time curve, and it is this curve, rather than any individual frame, that constitutes the measurement.

Determinants of the tissue curve

Three physiological factors govern the shape of a tissue curve.

Delivery determines how much contrast reaches the voxel and how rapidly, and reflects tissue perfusion. Permeability determines the rate at which contrast crosses the vessel wall into the surrounding tissue. In healthy brain this is negligible, in other tissues (muscle tumor, inflammatory lesion or infarct) contrast extravasates and the tissue curve rises higher and persists longer (delayed washout). Interstitial volume determines the extravascular extracellular space available to the extravasatedccontrast, and therefore how much may accumulate before tissue and plasma equilibrate.

A curve exhibiting rapid enhancement followed by rapid washout indicates high perfusion with limited retention. A curve enhancing slowly but monotonically indicates accumulation exceeding clearance. These patterns form the basis of qualitative interpretation; separating the underlying parameters is the object of pharmacokinetic modeling.

The arterial input function

A tissue curve cannot be interpreted in isolation, because its shape depends on the quantity of contrast delivered as well as on the properties of the tissue. Two voxels of identical physiology yield different curves if injection rate, cardiac output or dose per unit body mass differ between the subjects examined.

The arterial input function, the concentration-time curve measured in a feeding artery, quantifies this. Deconvolving the tissue response with respect to the input yields a result attributable to the tissue rather than to the administration. Error in the input function is the dominant source of systematic error in quantitative DCE-MRI, so extra care for its determination is warranted.

Conversion of signal intensity to concentration

Signal intensity is expressed in arbitrary units. It depends upon the scanner, the receive coil, the pulse sequence and the receiver gain, and additionally upon the native T1 of the tissue, which varies between tissues and between subjects.

Conversion to gadolinium concentration requires two further quantities:

  • the pre-contrast T1 of each voxel, typically derived from a variable flip angle series acquired before contrast administration. This is the output of parametric_scripts and of the T1 mapping stage of DCEPrep
  • the relaxivity of the contrast agent used

Physical interpretation of the derived parameters is valid only after this conversion. The preceding stages of the pipeline, namely motion correction, T1 mapping and input function determination, exist to render the conversion reliable.

Acquisition requirements

Quantitative analysis requires three acquisitions, which are those converted by dce2bids:

Acquisition Description Purpose
DCE The 4D dynamic series The measurement itself
VFA Multiple flip angles, pre-contrast Native T1 map for conversion to concentration
Structural Typically a T1 MPRAGE Anatomical reference for registration and region definition

Two acquisition parameters constrain all subsequent analysis. Temporal resolution determines which models may be fitted: the vascular contribution is carried by the initial rapid upslope, and a series sampled too coarsely to resolve it cannot support estimation of blood flow irrespective of the fitting procedure employed. Acquisition duration determines how well the slower extravasation component is constrained; a series terminated prematurely leaves the washout phase under-sampled.