Thank you Michael,

Now it's clearer to me.

I manually tell quasil the --tis values so the default should be overwritten. Let me know please if quasil is fixed to 7 phases only, so I can change acquisition parameters.

Smth more on conversion, is there any dicom tag (or any way of) telling which image is tag and which control?

Thanks again.

Dorian

TJU

2013/8/2 Michael Chappell <[log in to unmask]>

When I did a par/rec conversion I got various separate files that then needed to be combined together, quasil was not designed to directly take the Philips scanner output, but it was designed such to be relatively easy to get form the files output by dcm2nii that I was getting into the right input format. What you want is a single file with the tag-control pairs grouped as each set of TIs together and then each repeat of the TIs within each phase together, a diagram might be helpful:

  V1 V2 V3 V4 V5 V6 ….

| T1 C1 T2 C2 T3 C3 … | T1 C1 T2 C2 T3 C3 … | … || T1 C1 T2 C2 ... 

|         -- repeat 1 --          |         -- repeat 2 ---          | … ||   --repeat 1 ---

|                               --- phase 1 ----                                   ||        --- phase 2 ---

With V=volume number in the final single file, T=tag and C=control.

What might worry me is that you say you have 8 phases, but QAUSAR normally has 7 and this is what quasi will be expecting.

It is not impossible to do pair-subtraction and some form of analysis with oxford_asl, but you will need to choose that dat you use carefully and you would end up throwing away the advantages of having QUASAR data.

Michael

On 1 Aug 2013, at 17:59, Dorian P. <[log in to unmask]> wrote:

Thank you Micheal,

I tried dcm2nii on parrec and it produces 16 different files (for 8 phases). Quasil does not accept separate files, so it need to be a single 4D gile. How did you manage to get this?

I tried to swap the tag-control on my 4D file and got positive perfusion in occipital cortex as expected. However, the results are not exactly inverted, in fact they are weaker.

Therefore the first issue to resolve is how to convert dicom or par-rec in the order Quasil is expecting them? Has quasil been set up for Philips scanners output, or is there a convention on the order scanners output tag-control data?

P.s. I tried also to do pair subtraction with asl_file and analyze with oxford_asl but that way all the results are just noise.

Thank you.

2013/8/1 Michael Chappell <[log in to unmask]>

QUASIL assumes that the data is tag followed by control. I think that if the order were reversed in the data then you might not expect to see sensible activations, since a negative ASL signal is not expected - this might differ depending upon whether you are using the model based or model free analysis options though.

I have in the past converted from par/rec using dcm2nii with success.

Michael

On 31 Jul 2013, at 15:37, Dorian P. <[log in to unmask]> wrote:

> Dear FSL list,
>
> First thank you for creating the ASL and quasar analysis methods.
>
> I am currently using quasil to analyse quasar. It seems to work well (30 volumes on a checkerboard task) but I have the results inverted, such that the condition with activation is more negative then the rest condition (of course in occipital lobe).
>
> I suspect the tag and control images are in the opposite order of which the software thinks. So my question is in what order does quasil thinks the tag-control couples are in the 4D niftii?
>
> What software do you use to convert dicom or par/rec ASL data for analyze with quasil? I used MRIconvert with output 4D FSL niftii, but apparently the tag-control order is not good. DCM2NII did not even convert the whole brain correctly.
>
> Thank you for any help.
> Dorian
> TJU