"longitudinal change" sign of results

[Daniel]

Hi there,

I have a longitudinal "decrease" result, but when I pull the mean values from the database they are mostly positive. Is there a better way to investigate the underlying data, and perhaps plot?

Thank you,

Daniel

[Frank Yeh]

If you select 'filter decrease' when creating the longitudinal database, then the positive value will be zeroed and negative values changed to positive.

This you may need to pull from no filtered database.

Hope this helps.

Best,

Frank

--
You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/f71e1d35-4014-4c2a-917a-e37a5b6006b1n%40googlegroups.com.

[Daniel]

I ran a longitudinal analysis which showed a "decrease" in tracts in [follow up-baseline]. My intuition is that the values of the "difference" database in these tracts should be negative.

However, applying the resulting tracts to the "positive" and "negative" filtered data helps with some insights. There is a lot of positive and negative change across the brain, and please correct me if I'm mistaken: correlational tractography tries to find the longest continuous tracts of either positive or negative change when doing "longitudinal change" as term of interest, permuting the sign of the data to determine a null distribution and compute FDR.

For display/reporting purposes, I'm hoping to plot something about the tracts that were identified that is more accessible to someone who is less familiar with the procedure. My first intuition was the mean QA of the tracts, but because there are areas of positive/negative change, the mean of the tracts is very close to 0 or even positive, which does not align with the intuition (and is not technically how the tract was determined). What is a good measure that describes the found tracts? Length? Number of pos/neg voxels?

Thanks for your help.

Daniel

[Daniel]

Is the suggestion to pull the filtered data using the result of regular database? Or re-calculate the result using the filtered dataset and apply it to the unfiltered? 

I don't see the utility in reporting the mean of only the negative-valued parts of the tracts - shouldn't the tracts be negative?

On Friday, June 14, 2024 at 3:40:00 PM UTC-4 Frank Yeh wrote:

[Daniel]

Okay, it seems doable. 

Yet, DSI studio can calculate the "decrease" just fine in correlational tractography - does it have to do with the calculating the length of tracts? mean of these "decrease" tracts are in fact positive? 

Just confirming that the "longitudinal change" is essentially a 1-sided t-test of the difference scores? or something else?

On Friday, June 14, 2024 at 3:40:00 PM UTC-4 Frank Yeh wrote:

[DSI Studio]

Sorry I am not very sure about the questions here.
Perhaps you may check out previous threads about "connectometry" and "filter" function in the updated DSI Studio: https://groups.google.com/g/dsi-studio/search?q=connectometry%20filter

Likely we are talking about different things due to version differences.

Best regards,

Frank