I assume that you have a method for calculating the appropriate unwrapped phase differences from the multiple echo scans, as we do not have a tool in FSL to do that, but see below for some suggestions.
There are two main stages in using a fieldmap:
1) calculating the fieldmap in units of radians per second (rad/s)
2) applying the fieldmap to correct the (functional or diffusion) EPI
It is only for the second step that you require the echo spacing or dwell time. So the method that you used to acquire the fieldmap does not affect the second stage. That is, the echo spacing (or dwell time) is purely related to the functional/diffusion EPI scan. If you used acceleration for the EPI acquisition then you deal with this in exactly the same way as for any other study (independent of your fieldmap acquisition). That is, take the basic echo spacing and:
- divide it by in-plane acceleration factors (e.g. GRAPPA)
- do not do anything about partial Fourier
- do not do anything about multi-slice accelerations
To make the fieldmap, for step 1, you will potentially require different processing than what we can provide in FSL. You will need to derive a fieldmap in the correct units of radians per second (rad/s) from your data. One simple way to do this would be to unwrap the individual phase images, take the pairwise differences and divide each by the difference in echo times (not echo spacing, but the TE values associated with each image in the pair). This can be done for each pair and then the results averaged (possibly with outliers removed, to avoid situations where the phase unwrapping fails to get the right offset, which could easily happen with larger differences in echo times). Such a processing would be possible with the FSL tools, but it would not be optimal for this data. Ideally you would perform a regression on the wrapped phase values to extract the appropriate slope vs TE (which is then in rad/s) but we do not have such processing implemented. Maybe you already have that by some other means. Whatever you do though, once you have a fieldmap in rad/s then you can use it like any other fieldmap in FSL - at that point the way that the fieldmap was acquired will not affect the processing.
Finally, I would recommend using epi_reg rather than the basic fugue tools, as epi_reg generally does a much better job. The only additional thing that you need is a magnitude image that is already in the fieldmap space (ideally derived from the same acquisition/data that is used to get the rad/s fieldmap). Once you have the rad/s fieldmap and a magnitude image in the same space then you should be able to get good results from epi_reg.
I hope this helps.
All the best,
On 16 Sep 2014, at 18:30, Ajay Kurani <[log in to unmask]> wrote:
> Hi all,
> I found this link from a previous thread that explains several points of the calculation: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;92bd6f89.1403
> I think my main question remains if these equations can be extended if the B0 scan that the maps were derived from were acquired using multiple echos (ie. 4-6) as opposed to the standard 2.
> Additionally there are some parameters which take into account grappa and others do not from the post so any help would be much appreciated.
>> Dear FSL users,
>> I have a calculated unwrapped field map in rad/sec created from a non-standard sequence/processing which uses more than 2 echo times. The issue is that the following command is based on 2 echo times. I am trying to use FUGUE for non-siemens sequences (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE/Guide#Command-line_tools_.28advanced.29)
>> Based on the tutorial above I wanted to use the following command:
>> fugue -i epi --dwell=dwelltime --loadfmap=fieldmap -u result
>> From what I understand the dwell time is also known as echo spacing. When using multiple echos, sometimes echo train length is incorporated according to the following post:https://www.jiscmail.ac.uk/cgi-bin/webadmin?A3=ind1008&L=FSL&E=quoted-printable&P=606776&B=--_000_AAA761A2C05F5E4FAD4F0B6433BDD7110EC22179BDsdcmbx01excha_&T=text%2Fhtml;%20charset=iso-8859-1&pending=
>> However the equation there seems specific to a Philips system (I'm using a 3T Siemens Trio).
>> I wanted to see if I have several echos (ie. 4-6), what is the appropriate way to calculate the effective echo spacing to be used in the fugue command for the following scenarios:
>> a) effective echo spacing with no acceleration
>> b) effective echo spacing with acceleration used during acquisition
>> Ajay Kurani
>> University of Pittsburgh