We appreciate the comments from Dr. Frydrychowicz and colleagues. They pointed out that a relatively lower signal to noise ratio (SNR) of time-resolved three-dimensional (3D) phase-contrast images could mislead the flow pattern similar to that of Takayasu arteritis and that the flow visualization only at systole could overemphasize its usefulness in assessment of abnormal blood flow patterns in this disease.

Dr. Frydrychowicz has shown that the streamline close to the vessel wall can be eliminated by reducing the SNR. In our two control subjects, however, we did not observe such artificial streamline elimination on the velocity mapping generated by GTFlow software (GyroTools, Zurich). We deliberately positioned the emitter plane in the ascending aorta and refrained from using backward integration from a down-flow position. This resulted in a clear depiction of the high-velocity components of the flow field in the volume of interest. This effect is clearly visible in the images provided by Dr. Frydrychowicz in his letter. The additional streamlines in his images fill the edges of the vessel lumen, but mask the high velocity profile in the central regions of the vessel lumen simultaneously.

In patients with Takayasu arteritis, the absence of streamline was observed just along the thickened wall, and its “eccentricity” indicates that this finding should be pathologic, not be artificial. This finding was also supported by another observation that the flow pattern was partly improved by steroid, which paralleled with the decrease in arterial wall thickness. We gave some quantitative data in addition to 3D velocity mapping.

We observed 3D velocity mapping at all cardiac phases and with both 3D streamlines and path lines. However, we showed the images at systole in this paper, when the blood flow was fastest even close to the vessel wall. The property of streamlines to depict the velocity field independently from the temporal flow evolution guided our choice for streamlines and allowed to limit our comparisons to the systolic phase.

We certainly agree with their concluding remark pointing out that interpretation of 3D velocity mapping needs to be well-considered and thoughtful. Care should be taken about the imaging parameters of time-resolved 3D phase-contrast MRI. Probably because of these concerns, almost all researchers still use qualitative terms, such as “right-handed helical” and “turbulence,” rather than quantitative data in the assessment of 4D flow. Some quantitative analysis of the blood flow patterns may be preferable to the visualization-based analysis, but this would require consensus standards through experiences and evidences provided by researchers.

The authors thank Dr. Gerard Crelier (GyroTools and ETH Zurich) for his helpful advice and discussion about several technological topics.

Yasuo Amano MD*, Ryo Takagi MD*, Tetsuro Sekine MD*, * Department of Radiology, Nippon Medical School, Tokyo, Japan.