Bertram Wilm, PhD Skope

In the last decades, diffusion MRI has become an established technique in healthcare and neurosciences. Aside from being the standard method for the early detection of stroke, diffusion MRI and its structural information have also found growing interest in clinical practice and research. But like many other MR methods its successful development is challenged by several pitfalls.

Geometric inconsistency and low SNR in diffusion

Inaccurate gradient systems are one of the primary source of image artifacts. They induce eddy currents, which can affect the subsequent image encoding and lead to distortions and ghosting artifacts in the images. This is a major obstacle as it leads to geometric inconsistency among diffusion images as well as images acquired with different sequences. Another hurdle and major source of artifacts is patient motion such as breathing and cardiac pulsations. Compared to other MRI sequences, diffusion MRI is particularly prone to artifacts induced by such minute patient motion. The use of rapid single-shot acquisition strategies is therefore a popular approach in counteracting these artifacts. Finally, diffusion MRI faces the challenge that diffusion contrast always comes with lower image signal-to-noise ratio (SNR), which triggers the search for MR sequences with high SNR efficiency.

The search for SNR-efficient sequences

Even though promising results have been shown by using multi-shot DWI approaches, single-shot echo planar imaging (EPI) has proven to be the preferred strategy to tackle the aforementioned challenges. Due to its motion robustness and SNR efficiency, single-shot EPI is applied in the vast majority of diffusion scans. However, its popularity does not account for its susceptibility to several artifacts, including global and local image distortions as well as image ghosting. Those artifacts can impact the accuracy of quantitative diffusion parameters and impair the clinical value of acquired diffusion images – opening the question whether the most suitable sequence for accurate diffusion imaging has already been found.

Another acquisition strategy, which holds great potential is spiral k-space sampling. It offers distinct benefits over EPI. Most importantly it allows for a significantly shorter echo time, which increases the SNR. Spiral readouts are also attractive for single-shot acquisition and provide high efficiency for spatial encoding, which is most beneficial after a laborious signal preparation. Despite these advantages, the use of single-shot spiral strategies is rare and virtually absent from clinical practice today. Part of this reason might be their sensitivity to B0 off-resonance and gradient imperfections. However, it has been shown recently that single-shot spiral imaging can be utilized for diffusion images of high quality by an expanded and more accurate signal model.

The choice is yours

It remains to be seen which of the two strategies will get the competitive edge over the other. Also, exploring the possibilities that these new techniques offer for diffusion applications is an interesting field itself. Which one do you prefer: The classic Cartesian or the sinuous spiral strategy?