Many thanks Donald,

Ok this makes sense now to me. However how about having the same voxel being significant in the linear term as well as the quadratic one even when looking at them individually (ie performinga t test at the linear regressor only and another one at the quadratic regressors only). This also even when the regressors are orthogonal ?     

 

Kind regards

Aser

On 30 Sep 2015, at 18:45, "MCLAREN, Donald" <[log in to unmask]> wrote:

If I follow you have the PPI term, you have the the linear PM PPI term, and you have a quadratic PM PPI term. You found that only the quadratic PM PPI was non-zero. This is certainly possible. It means that the connectivity doesn't change with the task at the mean PM value, but increases as the PM value gets lower or higher. This would be a U-shaped effect of connectivity. If you look at y=x^2, then you can think of y being connectivity and and X being your PM.

Best Regards, 

Donald McLaren, PhD

On Wed, Sep 30, 2015 at 1:36 PM, Aser A <[log in to unmask]> wrote:

Hi Donald,

I have another question related to the PM result. I sometime get activations in the PPI of the PM but these activations are not significant in the 0 order or the main effect result. Is this possible or usual ? How this is can be explained ?

Thanks

Aser

On Mon, Sep 28, 2015 at 11:19 AM, Aser A <[log in to unmask]> wrote:

Many thanks Donald,

I have to admit that I did not quite get the method of the plot. But many thanks for the other comments 

On Fri, Sep 25, 2015 at 1:37 AM, MCLAREN, Donald <[log in to unmask]> wrote:

See below.

Best Regards, 

Donald McLaren, PhD

On Thu, Sep 24, 2015 at 4:28 PM, Aser A <[log in to unmask]> wrote:

Many thanks Donald. I saw in the article you mentioned a diagram that I did not understood it well. The one that shows the 25 % 50 or 75% connectivity in circles. 

From this I have these  questions  :

- What does not mean and how it can be created ? 

I would suggest that you read my recent paper on the flexible modulation of connectivity in AD (McLaren et al. 2014 in NeuroImage), it has a better example. I've also sent my OHBM presentation on the topic in a separate email. Briefly,

(1) Select the top N% of voxels in the brain,

(2) Count the number of voxels selected from step (1) in each ROI parcellation (in the Swallowing paper, we used AAL regions);

(3) Divide by the total number of voxels in each ROI parcellation.

Plot the results using the radar plot function in Excel (or your favorite graphing program)

 

- Say that I found using gPPI, that more than one region are connected with region X. How can I tell or get indication of which of those two regions has stronger connectivity ? Can the T value be an indicator for this ? Or the number of voxels. 

The number of voxels tests the spatial distribution of the effect. The T-value can be used as an indicator about where the connectivity change is the most significant. The con_values tells you where the connectivity is greatest in amplitude. To directly compare the two regions, you should compare the mean con_ values in each region using a paired t-test. Generally, you can talk about the significance of the effect. The same rules apply when comparing the activity of two regions as comparing the connectivity of two target regions.

 

- In the PPI, is it possible to get the same region connected with itself ? I got a result where within the seed ROI there is significant activation ?

Yes. This is theoretically possible. The seed region is the mean or eigenvariate of the seed time course. If the seed region changes connectivity with itself over time, then you could get effects in the seed region. This could be interpreted as more synchrony in activity across the seed region. If you look closely, it's likely only to be a portion of the seed region and not the entire seed. For small seed region, this won't occur because there isn't enough variability in the activity across the seed for it to fluctuate over time. 

I should probably run some simulations to see what causes this to happen.

Hope this helps.

Many thanks

Aser

On Thu, Sep 24, 2015 at 12:44 AM, MCLAREN, Donald <[log in to unmask]> wrote:

This means that the connectivity of X and Y are parametrically modulated by your PM variable. For example, if swallow number is your PM and you have a positive value for the PPI PM, then you would interpret this as the connectivity is increasing with each swallow (see Humbert and McLaren 2014).

Hope this helps.

Best Regards, 

Donald McLaren, PhD

On Wed, Sep 23, 2015 at 5:58 PM, Aser A <[log in to unmask]> wrote:

Hi all,

I have a design (task) where I also included parametric modulation (PM) in. 

I think performed PPI in the main effect of the task as well as the PM. 

My question is more related to understanding the meaning of PPI in the PM. So say for example if region X was shown to be connected with region Y in the PPI of the PM. 

Does this indicate that both regions are functionally behaving in the same way and have similar relationship of the PM? 

Or in other words how would you explain this ? 

Many thanks 

Aser