Units in sequence definitions

[bennyr...@mac.com]

Dear Tony et al.

It would be very helpful if the documentation for the sequence blocks included some specifications on the units used. For example, I am interested in looking at some slice selective pulses and trying to work out the values of RF bandwidth and gradient FlatTopArea to achieve a specific slice (or slab) thickness. Is RF bandwidth measured in kHz, for example? What about gradient amplitude, is this mT/m?

I modified the gre.xml sequence to use a sinc pulse with bandwidth=1 and a trapgradpulse of amplitude 1, this seems to produce a slice profile of FWHM approximately 6mm, whereas my calculations (assuming kHz and mT/m) suggest it should be about 24mm. Any chance you can shed some light on what is going on here?

Thanks a lot,

Ben

[Lars G. Hanson]

Hi Ben,
Units in Jemris are indeed tricky. I attach my notes on the subject, including a few comments from Tony, who may well have better advice to offer.
Kind regards,
Lars

--
You received this message because you are subscribed to the Google Groups "JEMRIS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jemris+un...@googlegroups.com.
To post to this group, send email to jem...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jemris/d2562db6-26d3-4b5b-b618-1d003f66cf91%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Tony Stoecker]

Hi Ben,

sorry, units are treated quite poorly in JEMRIS.  Thanks to Lars' notes, there is something for you to start with. As you can read at the end of this document, the general idea was to use gamma=1 (without units!), time in milliseconds, spatial position in millimeter, and frequencies as circular frequencies. Thus, rad/ms is the unit of temporal frequencies and rad/mm is the unit of k-space. Unfortunately, these definitions are not consistent across all objects. Sometimes natural frequencies are used, for example bandwidth is defined in kHz in some RF pulses. 

There is definitely more consistency and especially documentation needed! By then the best advice I can give is to try a factor of 2*PI, if results are not as expected ... :-(

Best,

Tony 

[Ben Rowland]

Thanks, Lars and Tony, some very helpful suggestions. I think that I have worked things out, by dividing the gradient by 2*Pi the profiles are now in line with what I was expecting. I also needed to add some negative ADCs to the sinc pulse to get it to behave properly - there are a few things that take a bit of figuring out to work out whether something is an artefact that goes away with more spins or more ADCs.

Best,

Ben

--
You received this message because you are subscribed to the Google Groups "JEMRIS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jemris+un...@googlegroups.com.
To post to this group, send email to jem...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/jemris/d88bae67-5bec-4b00-832f-f609f60f879d%40googlegroups.com.

[Tony Stoecker]

Hi Ben, 

This is strange: if standard pulse shapes are used, then the number of (negative) ADCs should not change the simulation i.e. the physics. Negative ADCs are only a feature for visibility of the sequence diagram. 

Would be interesting for me to see where you observed this behaviour. Can I easily reproduce your results?

Best,

Tony

[bennyr...@mac.com]

Hi Tony,

I will see if I can recreate the issue, but I made a lot of changes to my sequence while trying to get things working and it will be difficult to get it back to the weird times. It is quite possible that something else was causing this effect (like not enough spins per voxel) and I misdiagnosed the solution. Currently everything seems to be working for me so I am very happy for now.

Best,

Ben

[Tiago Fernandes]

Dear all,

Firstly, thank you all for the important input discussed in this topic. Sharing my experience on the topic, I would like to emphasize that when dealing with external gradient (in my case a spiral trajectory) one should be cautious. The units should be as Tony early expressed. However, in order to fit the units of jemris regarding k_space, i found that i had to multiple my gradients (before saving them into a '.h5' file) by a 2*pi factor. While doing so, i find my k_space in rad/mm and then it fits my simulations and my specified FOV.

Best,

Tiago

[Lars Hanson]

I sligthly updated my note on units in JEMRIS to ease reading. Attached. Same content, just formulated a little different. If there are recent relevant changes in JEMRIS, then they are not reflected.
Kind regards,
Lars

[Jörg Felder]

Dear all,

thanks for the info so far. Now, I am trying to look into RF pulses - so quite naturally the question arises how to scale these. Did anybody try this? Otherwise I think that I will try to simulate an analytic rect pulse with a given duration and try to figure in the issues of 2*pi and gamma to get the B1 magnitude in something like uT ...

Best regards,

Jörg

[joerg....@googlemail.com]

Ok - done. A rect pulse of 1 ms duration  and amplitude pi creates a 180 degree flip angle. Sounds reasonable :-)