Print

Print
Dear Peter and Rita,

Thanks for the advice. I read the paper and there is enough evidence to consider model slice-timing a big issue in task-based fMRI studies (specially for distanced antero-posterior networks such as the DMN). I am working with data of rsfMRI on mice, with a relatively low TR (1.2s) running through 24 interleaved coronal slices (caudal-rostral direction of acquisition), without slice timing pre-processing steps. Even though this is a small TR, slice timing should be included. I will try with the TR/2 default and also with the anatomically informed delays and let you know if I get any differences. Any recommendations on how to do this systematically (as I have a 4-node network).

There is an article (Wu, et al., 2011) which suggests that rsfMRI signal fluctuation indices (functional conn., ALFFs, and fALFFs) are minimally affected by slice-timing (in the pre-processing). Please correct me if I am wrong, but I consider that when working with generative models of signal features (DCM of cross-spectra) and making inferences on effective connectivity, time-delays play a crucial role in causality. 

Rita, I would greatly appreciate the script and give it a try. Thanks a lot for your help, and hope to soon inform you on the findings.

Best regards,


Daniel Gutierrez Barragan, PhD Student
Center for Neuroscience and Cognitive Systems (CIMeC)
Italian Institute of Technology (IIT)
University of Trento
Rovereto (TN): Palazzo Fedrigotti - corso Bettini 31

2015-10-06 11:01 GMT+02:00 Zeidman, Peter <[log in to unmask]>:

Thanks Rita,

That’s very good advice. If you try this Daniel, do let us know if there’s much difference in your parameters / model evidence.

 

Best,

P

 

From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Rita Elena Loiotile
Sent: 05 October 2015 21:19
To: [log in to unmask]
Subject: Re: [SPM] DCM.delays inquiry

 

Hi Daniel,

Just to add to this.  There's a very good paper on this exact issue: http://www.ncbi.nlm.nih.gov/pubmed/17161624.

 

For interleaved acquisition, it's recommended that you do slice timing correction in preprocessing and then set DCM.delays to the TR of the reference slice.

 

For ascending or descending sequences, it's recommended that you NOT do slice timing correction in preprocessing, and instead set DCM.delays to the TR corresponding to when the center of each ROI was acquired.  There is an SPM script mentioned in that article that lets you input the regions and gives you the corresponding TRs.  (The trick is that the TR's have to be defined according to the pre-normalized T2* acquisition.)  Unfortunately, the script doesn't work if you've used the newer SPM12 normalization procedure.  I have a script that I made, if you are interested, but it is very ugly and calls several FSL functions.  

 

On Mon, Oct 5, 2015 at 4:52 AM, Daniel Gutierrez Barragan <[log in to unmask]> wrote:

Hello all,

 

I am working on performing DCM from script and was wondering if there are any specific recommendations regarding the DCM.delay information that DCM needs to be fed with. I know that these correspond to slice-timing temporal correction, but my question is if there is a preferred reference (slice with delay zero)? or if it is better to do it according to acquisition sequence used?

 

Best regards,

 


Daniel Gutierrez Barragan, PhD Student
Center for Neuroscience and Cognitive Systems (CIMeC)
Italian Institute of Technology (IIT)
University of Trento
Rovereto (TN): Palazzo Fedrigotti - corso Bettini 31