How to Incorporate Eddy Current Effects in Simulation?

[shenzhouxiaohao-cpu]

I am conducting simulations for ETL>1 FSE sequences and have validated them on a self-built MRI system. When using thicker slice thicknesses, the experimental results align well with simulations; however, with thinner slices, simulations still yield normal results while actual testing shows black stripes in the output‌.

The figure below shows a cross-sectional view of a water-filled test tube with a slice thickness of 0.5 mm, ETL=5, TE=15 ms, and TEeff=45 ms. As seen in the image, black stripes appear in the originally homogeneous circular foreground.

Due to the very small bore size of my system, high gradient strength and slew rates can be achieved, enabling slice selection at 0.5 mm thickness. In this case, reducing the ETL to 1 or enlarge slice thickness to 2mm would resolve the issue .

Therefore, I aim to verify whether this artifact is caused by eddy currents. I am unsure how to account for eddy current effects in pypulseq.

[schuenke]

Please read my comment in your Issue #317, which I just closed. Especially this part:

We also need to clarify that GitHub Issues are primarily for reporting bugs or discussing planned features. We expect users to research and attempt to resolve their implementation questions or problems independently before creating a new issue. Please ensure you have thoroughly investigated your solution before posting here or in the MRzero repository going forward.

I leave this Issue open for the moment to give you the chance to adjust it, but please keep in mind that everything related to MRI simulations is not part of (Py)Pulseq. Trying to reduce eddy currents could be an option. Did you try anything to do so?

[shenzhouxiaohao-cpu]

@schuenke I just want to know how to simulated the impact of gradient eddy current by using pypulseq.

[ShannonZ]

You need a arbitrary shaped gradient. But I don't find such a api in pypulseq. Maybe you could split the exponential-decayed eddy current induced grad into small trapezoid segments and then use extented_trapezoid

[btasdelen]

@ShannonZ Are you looking for this: make_arbitrary_grad().

@shenzhouxiaohao-cpu As @schuenke said, PyPulseq does not provide a way to simulate eddy currents, because PyPulseq is not a simulation framework. This question is not within the scope of this package.

Usual first steps to reduce eddy currents are reducing slew rate and playing with TR. Simulating or compensating eddy currents are a whole another world, and there may be techniques specific to your aims and sequences. These are some relevant works:
https://mriquestions.com/uploads/3/4/5/7/34572113/1991_ieee_tmi_eddy_current_00075610.pdf
https://mriquestions.com/uploads/3/4/5/7/34572113/eddy_currents_nihms312105.pdf

Moving this to the discussions as this is not an issue.

[shenzhouxiaohao-cpu]

make_arbitrary_grad works well for it. great thx!

[mzaiss]

Well, there is a simple work around to mimick eddy current effects that just act as a gradient delay, which is to delay e.g. the adc a bit:
adc = pp.make_adc(num_samples=Nread, duration=adc_raster_duration, phase_offset=0 * np.pi / 180, delay=gx.rise_time + eddy_currents_induced_delay, system=system)

so you can try to counteract existing gradients delays by matching this adc delay. But its a bit hacky.

[mzaiss]

For simulation with gradient imperfections based on pypulseq sequences you can check Daniel Wests Paper: https://onlinelibrary.wiley.com/doi/10.1002/mrm.70093?af=R and the corresponding github: https://github.com/mriphysics/Optimization-for-non-ideal-gradients

[btasdelen]

Very cool work!

[ShannonZ]

cooooooool ！

[shenzhouxiaohao-cpu]

What an interesting idea! Thank you very much.

[shenzhouxiaohao-cpu]

I've solved the eddy current simulation problem using make_arbitrary_grad. Thank you all very much.