Hi

I don't know if this was asked before, but I'm using SLOMOCO for my data, and I want to use ART Toolbox (artifact detection toolbox) for motion censoring (scrubbing).

ART needs the motion parameters in TXT file, and I was wondering which of the SLOMOCO's output should I input?
I'm leaning towards the "mocoafni.1D" file, but I want to be sure.

Thank you in advance

Eduardo Garza

Eduardo,

The best file to use for slicewise scrubbing is slomoco.TDzmetric.txt. This is a slicewise average voxel displacement motion metric (Jiang et al 1995 Human Brain Mapping) generated from the slicewise out-of-plane motion parameters, and are the most relevant for residual motion contamination after SLOMOCO has run. This will allow you to identify individual slices that are contaminated, although you can also take a max over each slice group (every zdim) to get a comparable volumetric metric (also saved as slomoco.volumetric.TDzmetric.txt) so you have one motion corruption value per volume. 
Erik

Thanks! I will try that

Eduardo

Hi Erick

In the same questioning line, for the Nuisance Regression, which would be the equivalent for motion parameters to input in the model? I'm using white matter, csf, global signal and I want to use motion parameters.

Is the volmotion.TDZmetric.txt enough or is there something similar to the volume motion correction at 6 degrees of freedom?
for example, mocoafni.txt seems like a good candidate, or would you suggest another file metrics?

Thank you

Eduardo

Eduardo,

The SLOMOCO companion script is actually also intended to be used with PESTICA to correct for head motion in parallel with physiologic noise. Basically, you run PESTICA, then you run SLOMOCO with the -r flag (and be sure to specify the same file you specified as input EPI data for PESTICA, NOT the corrected post-PESTICA file) and it will run a slicewise motion correction and at the end, it will regress out a slicewise motion model, PLUS the PESTICA physiologic noise signals in parallel. It doesn't support WM, CSF or global signal - there is some controversy whether WM, CSF or global signal are safe corrections so i have not supported additional regressors at this point.

Erik

Hi Erick

I see, so I was doing it all wrong. I was running SLOMOCO first, then PESTICA, it wasn't clear to me how to combine them.

This is great. I know it is controversial at the moment but it's very hard to publish if you don't do it. And I guess it will be fine to regress the signals after the slicewise and pestica regression.

Thanks a lot

Eduardo

Hi Erick,

Just to be sure, the file I end up with (*.slicemocoxy_afni.zalg_moco2_irfret+orig) is the PESTICA + SLOMOCO regressed (residual) EPI right?

One of my patients moved a lot at the end of the 10 min session and the consequent BOLD fluctuation is still a bit visible after Despike, PESTICA and SLOMOCO. You mention in the README that it's possible to motion sensor using slomoco.TDzmetric.txt. Is this so? I already have the outliers from FSL_motion_outliers that I can input for regression. Do you recommend using your SLOMOCO file or should I go ahead and scrub with the FSL file?

Thanks again

Eduardo

Hi Eduardo,

Yes, that is the correct file for the PESTICA+SLOMOCO regressed EPI. Ignore this if you've already done this, but if not, check the timeseries to make sure cardiac fluctuations experience more reduction than non-cardiac fluctuations. I do this by looking at the timeseries in an artery before and after correction in AFNI with the "Plot" viewer up and switching back and forth between raw and corrected - after correction, everything will have reduced variance and this is true even if you're regressing pure random noise, but if you're regressing cardiac, then the spikier stuff in arteries should experience more reduction than adjacent voxel timeseries (note, there will still be spikey timeseries stuff in arteries even after the best existing physio corrections).  After SLOMOCO, you shouldn't have to do any more regression for motion purposes.

If you have to do motion censoring, I do indeed recommend using the slomoco.TDzmetric.txt file instead of a volumetric file like the FSL_motion_outliers. One way you can do it is to take the maximum over each slice group (like volumetric_max_head_motion=squeeze(max(reshape(slomoco,[zdim tdim]))); - I might have the zdim and tdim backwards, I don't have a linux box with matlab available at the moment so I can't check) and then censor by volume with existing thresholds.
Erik

PS: will you be at ISMRM in Toronto in two weeks? I have a poster on SLOMOCO and I'd be happy to talk in more detail in person then.

Hi Erick

Sorry I didn't reply in time. I was not attending anyway, but I hope to see you in HBM or Brain Connectivity 2016 so we can talk more about this.

Thanks

Eduardo

Dear Erick

I was revising this yesterday as I'm very interested in using censoring with SLOMOCO.I'm having a hard time trying to use the SLOMOCO output in ART Toolbox, but now I don't think I need to.

Seems SLOMOCO already outputs the "slomoco.volumetric.TDzmetric.txt" where you get the MAX from the slice-wise TDz file as you mentioned.

It will take me a bit as I'm not an expert programmer, but it seems simple enough to implement a bash or matlab script that extracts and detects outliers based on the volumetric.TDz file, to be censored afterwards.

Thanks again

Eduardo