A publication by Christian Mirkes, Researcher at Skope MR Technologies

30 Oct 2020

The data acquired from the field probes and that received by the MRI coils are typically subject to delays of various sources inherent to the acquisition process, e.g., signal propagation delays in cables, probe heads, matching, filtering, amplification and conversion circuitries, digital processing delays etc.

Although the absolute value of the delay is often not of core interest, the relative delay between the sensors and the MR data is crucial for correct application of the field probe information. For example, a k-space trajectory acquired by the field probes must be synchronous to the acquired coil information for proper reconstruction. For challenging applications, a synchronicity between coil data and measured magnetic field evolution in terms of group delay on the order of 10 ns-50 ns is required.

This equally applies to applications calibrating gradient and shim systems based on field probe data for feedforward and post-correction. Since in the final application, the timing with respect to the scanner and the acquired MR imaging data is relevant, the calibration must be referenced to the time frame of the MRI imaging system.

A default strategy used for data synchronization is to start the acquisitions on both systems upon a common trigger. In addition, the spectrometers can be synchronized by using a common reference clock. However, even in such cases the found differences in group delay can easily exceed the requirements for synchronicity. Therefore, additional methods for measuring, determining, calculating or post-correcting such relative delays are needed. Three different methods are outlined in the methods section.



A simple but relatively time-consuming method to obtain this delay, is by reconstructing the images using the acquired monitoring data and the MR raw data. e.g., multiple EPI images can be reconstructed assuming different relative delay values. Given that the employed field monitoring based trajectory accurately describes the encoding, the only unknown in the encoding is a delay value (that can be channel specific). This delay may then be found by minimizing the delay ghost artifact in EPI images. Similar methods may also work for spiral imaging where a delay primarily causes a rotation of the image that needs to be minimized.

A related method is to estimate the delay from the raw data itself rather than the image. This could be done, e.g., by matching the odd and even lines from EPI phase correction data. If this is done using measured or accurately described k-space trajectories, the only degree of freedom is a delay that needs to be determined. A similar approach should also be possible for other readout strategies that record closely neighboring k-space points multiple times.

A: k-space trajectory in the readout direction during the EPI phase navigator (blue); overlayed with the magnitude of object signal (red, yellow, purple). B: Phase of even and odd lines. The linear phase difference indicates a timing delay.

A third method that is recommended by Skope does not rely on the MR data of the object, and therefore works independently from the MR imaging sequence. Here, the delay from an artificial time varying RF signal is fed into the field probe and into the receive coil pathway via the field probe electronics. The relative delay between the signal acquired on the two spectrometers can then be found as the one that best aligns both data.

An option to generate these signals and transmit it via the field probe pathway is available with Skope’s acquisition system software.

Amplitude of an artificial MR signal as coupled into the receive coil pathway, acquired by the MR system spectrometer (A) and the same signal as coupled into the field probe pathway acquired by the field monitoring spectrometer (B). Note that the time axis is scaled differently for A and B.

Conclusion and Summary

We went through several possibilities to synchronize field monitored data with the MR raw data. After obtaining this delay value (which in general can be specific for each raw data channel), both data can be realigned. The time aligned data can be directly used in image reconstruction. This is particularly important for non-Cartesian imaging, where methods to mitigate gradient encoding errors, such as the EPI phase correction, do not exist.