help
help > RE: How to use "age" as covariates?
Jun 18, 2014 03:06 AM | Alfonso Nieto-Castanon - Boston University
RE: How to use "age" as covariates?
Hi Yifei,
Some thoughts on your questions below.
Best
Alfonso
Originally posted by Yifei Zhang:
Not a big impact, the main issue will be that you might be introducing a little bit of "spillage" from nearby regions into your ROIs. This may affect the ability of the aCompCor method to correct for physiological and movement effects without introducing potential biases (because of the relative introduction of small gray-matter effects into the white/CSF masks), but given that these masks are being eroded this effect should be relatively minor.
In order to make sure I understand correctly, may you check if I use the calculator tool correctly? I am not sure about the 2nd case above..
1) to compare the association between performance and FC (predictor) within one group(e.g. HC), but control for age and gender:
Define contrast [1] for only HC in the secondly-level results and import values for new FC measures (e.g. 'conn between ROI01 and ROI02);
Select in the 'between-subject effects' list: HC, age_HC, gender_HC, new FC measure
Enter the between-subjects contrast:[0 0 0 1]
Select in the 'measures' list: performance_HC
Enter the between-measures contrast:[1]
2) to compare between the two groups(HC and AD) the association between performance and FC (predictor), but control for age and gender:
Define contrast [1] for HC and AD separately in the secondly-level results and import values for new FC measures, new_FC_measures_HC and new_FC_measures_AD (e.g. 'conn between ROI01 and ROI02);
Select in the 'between-subject effects' list: HC, AD, age, gender, new_FC_measure_HC, new_FC_measure_AD;
Enter the between-subjects contrast:[0 0 0 0 1 -1]
Select in the 'measures' list: performance_HC, performance_AD
Enter the between-measures contrast:[1 1]
Both of these look perfectly correct. The only simplification would be, in your second analysis, you could simply select in the 'measures' list the "performance" -both groups together- measure (and enter as contrast 1). This will result in exactly the same values as what you are observing with your original analyses, in any way.
And, could you please clarify what is the colour of ROIs stand for in ROI-to-ROI result explorer, axial display, when selecting multiply ROIs (see attached picture).
The color in these plots simply corresponds to the number of positive/negative connections to/from each ROI (summing the number of positive connections and subtracting the number of negative connections), so red indicates mostly positive connections to/from an ROI, and blue indicates mostly negative connections (the meaning of positive/negative depends of course on your particular second-level analysis design; if you are looking at simple connectivity effects they correspond to positive/negative connectivity, if you are looking at between-group differences they correspond to higher/lower connectivity values in the first group, etc.)
Hope this helps
Alfonso
Some thoughts on your questions below.
Best
Alfonso
Originally posted by Yifei Zhang:
I have checked the histogram displays shown in
the Preprocessing step and the ROI files, they both look correct.
However, for some reasons, we are using smoothed data(5mm FWHM) for
the ROI-to-ROI analysis, do you think it will have a big impact on
the result?
Not a big impact, the main issue will be that you might be introducing a little bit of "spillage" from nearby regions into your ROIs. This may affect the ability of the aCompCor method to correct for physiological and movement effects without introducing potential biases (because of the relative introduction of small gray-matter effects into the white/CSF masks), but given that these masks are being eroded this effect should be relatively minor.
In order to make sure I understand correctly, may you check if I use the calculator tool correctly? I am not sure about the 2nd case above..
1) to compare the association between performance and FC (predictor) within one group(e.g. HC), but control for age and gender:
Define contrast [1] for only HC in the secondly-level results and import values for new FC measures (e.g. 'conn between ROI01 and ROI02);
Select in the 'between-subject effects' list: HC, age_HC, gender_HC, new FC measure
Enter the between-subjects contrast:[0 0 0 1]
Select in the 'measures' list: performance_HC
Enter the between-measures contrast:[1]
2) to compare between the two groups(HC and AD) the association between performance and FC (predictor), but control for age and gender:
Define contrast [1] for HC and AD separately in the secondly-level results and import values for new FC measures, new_FC_measures_HC and new_FC_measures_AD (e.g. 'conn between ROI01 and ROI02);
Select in the 'between-subject effects' list: HC, AD, age, gender, new_FC_measure_HC, new_FC_measure_AD;
Enter the between-subjects contrast:[0 0 0 0 1 -1]
Select in the 'measures' list: performance_HC, performance_AD
Enter the between-measures contrast:[1 1]
Both of these look perfectly correct. The only simplification would be, in your second analysis, you could simply select in the 'measures' list the "performance" -both groups together- measure (and enter as contrast 1). This will result in exactly the same values as what you are observing with your original analyses, in any way.
And, could you please clarify what is the colour of ROIs stand for in ROI-to-ROI result explorer, axial display, when selecting multiply ROIs (see attached picture).
The color in these plots simply corresponds to the number of positive/negative connections to/from each ROI (summing the number of positive connections and subtracting the number of negative connections), so red indicates mostly positive connections to/from an ROI, and blue indicates mostly negative connections (the meaning of positive/negative depends of course on your particular second-level analysis design; if you are looking at simple connectivity effects they correspond to positive/negative connectivity, if you are looking at between-group differences they correspond to higher/lower connectivity values in the first group, etc.)
Hope this helps
Alfonso