Basic research scientists and clinicians who wish to learn about conventional and state-of-the-art pulse sequence options for DSC-MRI and DCE-MRI data acquisition.

At the end of this lecture participants should be able to:

1) Describe which pulse sequences and parameters are most appropriate for DSC- and DCE-MRI data acquisition, along with their strengths and limitations

2) Describe how pulse sequences can be designed to leverage the biophysical basis of DSC- or DCE-MRI contrast in order to assess additional physiological features.

Dynamic Susceptibility Contrast (DSC) and Dynamic Contrast Enhanced (DCE)-MRI methods rely upon the acquisition of dynamic MRI signals in order to track the dynamic passage of exogenously administered contrast agents. DSC-MRI methods relate dynamic T₂ and/or T₂* shortening to contrast agent (CA) concentration, enabling, most typically, the assessment of cerebral blood volume (CBV) and cerebral blood flow (CBF). DCE-MRI methods track changes in CA induced T₁ changes in order to assess kinetic features such as the volume transfer coefficient, Ktrans, and the volume fraction of the extravascular space (v_e). As one may expect, the reliability of the extracted kinetic features from DSC-MRI and DCE-MRI is heavily influenced by the pulse sequence type, input parameters and contrast agent injection protocol. Acquisition details should account for spatial, and temporal resolution and SNR requirements, and site-specific optimizations. Further, acquisition parameters can also alter the sensitivity of the MRI signals to the underlying contrast mechanisms. The goal of this lecture will be to describe the motivation and requirements for robust DSC/DCE-MRI data acquisition, current best practices and opportunities for development.

In order to estimate CBV and CBF, DSC-MRI methods rely upon the measure of signal changes during the first pass of a CA bolus. Given this, the first requirement for a DSC-MRI experiment is the rapid bolus injection of the CA, with injection rates of 3 – 5 mL/sec (total injection time ~ 4 sec). Though the bolus will experience some degree of dispersion, this rate of injection enables its first pass through a given voxel (e.g. in tissue or for assessing the arterial input function) to be adequately characterized. To characterize the rapid changes that are induced in the MRI signal, DSC-MRI data, irrespective of the type of pulse sequence, should be collected with a temporal resolution of at least 1.5 sec. Given these temporal demands and the expected brain coverage, the spatial resolution of DSC-MRI data is typically less than that used for anatomic scans or other functional imaging methods (e.g. DCE-MRI, DWI). Other imaging parameters (e.g. flip angle, echo time, slice thickness) are then selected to maximize SNR and minimize signal saturation in regions where the CA concentration induces very large signal drops (1,2). Given the increasing clinical use of DSC-MRI (as typically applied), there have been efforts to standardize these acquisition methods to improve repeatability and multi-site clinical trials (3).

Pulse sequence design may influence the quality and spatial and temporal resolution of DSC-MRI data. A gradient echo with an echo planar imaging readout is the most widely used DSC-MRI pulse sequence because it facilitates high temporal resolution and high contrast to noise ratios. To reduce EPI-related artifacts (e.g. signal dropout, geometric distortion) recent groups have explored the use of single-shot spiral-based readouts or single-line rapid acquisition sequences (e.g. 2D fast spoiled gradient recalled echo sequence). Further improvements in temporal and/or spatial resolution may also be achieved through the use of multiband excitation (4) and compressed sensing (5).

A unique feature of DSC-MRI is that one can optimize acquisition methods in order leverage its biophysical basis. It is well recognized that gradient and spin echo DSC-MRI data have unique sensitivities to vascular populations, where GE and SE derived perfusion maps reflect vessels of all sizes and the microcirculation, respectively (6). Furthermore, by simultaneously acquiring GE and SE data additional physiological features may be assessed, including the vessel size and architecture (7,8).

The choice of acquisition methods may also differ with respect to the pathology being investigated. For example, in tissues with a disrupted blood brain barrier (e.g. brain tumors), contrast agent leakage and distribution within the extravascular space leads to dynamic and simultaneous changes in T₂*, T₂ and T₁. These complex signal changes and their confounding effects on CBV have necessitated the development and optimization of brain tumor specific DSC-MRI protocols, such as multi-dose CA injections and multiple echo acquisitions (9,10). These advancements have enabled the assessment of new microstructural features (e.g. cellular atypia (11)) with DSC-MRI and simultaneous measures of DCE-MRI data (12).

In order to estimate K^trans and v_e (the parameters most commonly assessed using DCE-MRI), the CA-induced MRI signal changes in tissue need to be assessed. The signal evolution in tissues of interest is slower than the first-pass kinetics measured with DSC-MRI, and enables the use of acquisitions with lower temporal and higher spatial resolution. With lower temporal resolution requirements, DCE-MRI protocols do not require EPI and are typically acquired using a 3D fast spoiled gradient recalled echo sequence. Similar to DSC-MRI efforts, organizations (e.g. RSNA-QIBA) have released standardized, best practice protocols for DCE-MRI to improve multi-site consistency and facilitate comparisons (13). Since DCE-MRI can be used in most human tissues, organ specific pulse sequence optimizations are frequently developed (14-16).

Kinetic analysis of DCE-MRI data requires pre-contrast T₁ mapping in order to relate signal changes to CA concentration. While inversion recovery techniques are known to provide the most robust estimates of T₁, their scan times are prohibitive, which has led to the use of multi-flip angle (which is the recommended approach) and, to a lesser extent, Look-Locker techniques. For accurate kinetic analysis the arterial input function also needs to be quantified, thereby necessitating the use of higher temporal resolution (typically ~10 sec/image) and rapid bolus injection (3 – 4 mL/sec). To enable higher spatial resolution multi-dose strategies may be used, where a low dose injection and high-temporal resolution imaging is used initially for AIF quantification followed by a standard dose of CA and higher spatial resolution dynamic imaging (17,18). AIF quantification may also be improved though the use of multi-echo acquisitions, as they enable the removal of unwanted, competing T₂* contributions.

As with DSC-MRI, technological improvements continue to shape DCE-MRI acquisition options, offering improved spatiotemporal resolution. Numerous studies have leveraged vendor’s time-resolved MR angiography sequences, which are typically keyhole imaging methods, in order to achieve higher spatiotemporal DCE-MRI data in prostate and breast cancer (19-21). Further improvements have been gained through the application of compressed sensing (14,15,22-24).