Synopsis

Keywords: Neuro: Cerebrovascular, Contrast mechanisms: Perfusion, Image acquisition: Image processing

Dynamic susceptibility contrast-weighted MRI (DSC-MRI) is highly sensitive in detecting disturbed hemodynamics. However, many techniques exist for calculating perfusion status, and there are multiple parameters that can be measured. We will discuss technical considerations and potential pitfalls in calculating and interpreting DSC-MRI-derived maps.

Syllabus

Dynamic susceptibility contrast-weighted MRI (DSC-MRI) is highly sensitive in detecting disturbed hemodynamics. Several parametric maps can be derived from the sampled concentration of contrast-agent over time curves, C(t), measured using bolus-tracking DSC-MRI. We will discuss the pros and cons of different techniques for quantifying intra- and inter-subject DSC-MRI differences, ranging from approaches that characterize the shape of the bolus curve to methods utilizing tracer kinetic analyses.

Parameters characterizing the shape of C(t), e.g., bolus arrival time or time-to-peak, are indirect measurements of perfusion status that have been used as surrogates for mean-transit time (MTT) and cerebral blood flow (CBF). Although easy to calculate, these metrics are often considered less accurate than techniques based on tracer kinetic theory. Therefore, methods relying on deconvolution with an arterial input function (AIF) are usually preferred. The choice of deconvolution techniques greatly affects the estimates of CBF and MTT. A clinically prevalent method using truncated singular value decomposition (SVD) for deconvolution makes no assumptions regarding the form of the residue function R(t). Furthermore, SVD has been shown to perform well in the presence of noise. A major limitation of the SVD technique is its sensitivity to delay and dispersion of the bolus from the site where the AIF is measured to the origin of the capillary bed of the tissue under investigation that will introduce errors in the quantification process and is a contributing factor to the sensitivity of CBF and MTT calculation to AIF selection. The impact of delay and dispersion can be compensated with advanced deconvolution techniques. There are now several open-source digital reference objects that vendors of software packages can use to evaluate the quality of their calculated parametric maps against known values to allow optimization of their algorithms.

Even with optimal methods for calculating DSC-MRI parametric maps, the interpretation of perfusion maps may be confounded by several factors. Various factors, such as the number of baseline points, acquisition duration, TE, TR or temporal resolution, peak signal drop, and contrast agent concentration, affect the accuracy of the calculated perfusion metrics. One source of error is insufficient contrast-agent concentration to induce enough signal change to estimate CBF reliably. Another potential confound is patient motion leading to signal dephasing in voxels that may be erroneously construed as resulting from high contrast agent concentrations. Another possible source of error is the truncation of C(t) curves from acquiring insufficient data points, which can lead to inaccurate determination of perfusion parameters. By examining the acquired unprocessed DSC-MRI data, one can detect these sources of errors and mitigate misinterpretation.

Conclusion

DSC-MRI is an important clinical tool due to its many desirable characteristics, including acquisition speed and high sensitivity for identifying hemodynamically disturbed tissue. It is important to consider the potential limitations and confounds of DSC-MRI while interpreting perfusion maps for the diagnosis and prognosis of patients. When in doubt, examining the raw data may provide clarification. Ongoing research improving the accuracy and reproducibility of perfusion estimates has the potential to expand the role of DSC-MRI in patient management. There are initiatives underway to standardize data acquisition and analysis to improve the reproducibility of results across centers, thus accelerating the adoption of DSC-MRI parametric maps as treatable biomarkers of disease.

Acknowledgements

No acknowledgement found.