afni_proc.py anaticor

AFNI Message Board
1

Hello,

I have a question about the examples in afni_proc.py help where #9 uses anaticor and says: Note that -regress_anaticor implies options -mask_segment_anat and -mask_segment_erode, and #10 says: Like example #9, but also regress the eroded white matter averages. (and includes options mask_segment_anat yes -mask_segment_erode yes as well as -regress_ROI WMe). I would think the mask_segment_ant and erode options would take place of the anaticor option, however, when I run #10 as in the example, it doesn’t take a lot of time whereas running #9 does.

Should I still be including -regress_anaticor in example #10 even though the segment anat and segment erode options are specified? Below is the script including all of the regressors - it doesn’t give me an error

afni_proc.py -subj_id $subj
-script proc.$subj -scr_overwrite
-blocks despike tshift volreg mask regress
-copy_anat $top_dir/anat_strip_al2epi+orig
-anat_has_skull no
-tcat_remove_first_trs 0
-dsets
$top_dir/rest_01+orig.HEAD
$top_dir/rest_02+orig.HEAD
$top_dir/rest_03+orig.HEAD
$top_dir/rest_04+orig.HEAD
$top_dir/rest_05+orig.HEAD
$top_dir/rest_06+orig.HEAD
-volreg_align_to MIN_OUTLIER
-regress_censor_motion 0.2
-regress_censor_outliers 0.1
-regress_bandpass 0.01 0.1
-regress_anaticor
-regress_ROI WMe
-mask_segment_anat yes
-mask_segment_erode yes
-regress_apply_mot_types demean deriv
-regress_opts_3dD
-jobs 12
-regress_est_blur_errts
-regress_est_blur_epits

2

Hello,

I will update that help to explicitly state that Example 10
does not do anaticor (regress local white matter). Those
segment and erode options are used to make a WMe mask, but
unless -regress_anaticor (or -regress_anaticor_fast) is
specified, the ANANTICOR operation is not applied.

The ANATICOR part of Example 10 was removed because it was
strange to regress average white matter as well as local
white matter (though it was supposed to be a demonstration
of what one could do).

In any case, I will mention that Example 10 does not apply
the ANATICOR step.

Thanks,

  • rick

P.S. Yes, -regress_anaticor_fast is faster than -regress_anaticor.
It uses a slightly different (but much faster) method.

3

Thanks so much Rick!

4

Oh wait, Example 9 does not include ANATICOR, that
is Example 9b. I did not scroll up high enough to see
the numbers.

So maybe there is no change to make…

  • rick
5

I meant example 9b vs. example 10.

Example 10 says:
Like example #9, but also regress the eroded white matter averages.
The WMe mask come from the Classes dataset, created by 3dSeg via the
-mask_segment_anat and -mask_segment_erode options.

I just wasn’t sure if in example 10, the -mask_segment_anat and -mask_segment_erode options = -regress_anaticor in example 9b… because 9b says:
Like example #9, but also regress out the signal from locally
averaged white matter. The only change is adding the option
-regress_anaticor.

       [b]Note that -regress_anaticor implies options -mask_segment_anat and
       -mask_segment_erode.[/b]

In other words, does the -mask_segment options in #10 take place of the -regress_anaticor option in #9b. And when you use anaticor, doesn’t it use the WMe segment anyway -
(3dTproject -polort 0 -input pb04.babydoll.r01.scale+orig.HEAD pb04.babydoll.r02.scale+orig.HEAD pb04.babydoll.r03.scale+orig.HEAD pb04.babydoll.r04.scale+orig.HEAD pb04.babydoll.r05.scale+orig.HEAD pb04.babydoll.r06.scale+orig.HEAD -censor censor_babydoll_combined_2.1D -cenmode ZERO -dsort Local_WMe_rall+orig -ort X.nocensor.xmat.1D -prefix errts.babydoll.anaticor)

So, how are there options different besides the obvious align to MIN_OUTLIER ?

Thanks Rick, Sorry if that’s confusing.

6

Unless you provide a mask to use for -regress_anaticor,
it will imply -mask_segment_anat and -mask_segment_erode
(since an eroded WM mask is required). So -regress_anaticor
implies both -mask_segment_anat and -mask_segment_erode.
So Example 9b would be unchanged if those -mask options
were added. It is merely for convenience that using
-regress_anaticor implies those -mask options, too.

But the converse does not hold. Options -mask_segment_anat
and -mask_segment_erode do not imply anaticor.

The anaticor process is not run unless it is asked for.

Example 10 does not ask for anaticor. It uses those -mask
options to create a WMe mask, but that is actually applied
via “-regress_ROI WMe”. That is to say, in this example
the average white matter signal is regressed out, not a
local one (anaticor).

Does that seem clear?

  • rick
7

Yes! Thank You!