help > Contrast define for longitudinal study
Showing 1-2 of 2 posts
Mar 7, 2017 07:03 AM | Geon-Ho Jahng
Contrast define for longitudinal study
Dear CONN Experts
We scanned a resting-state fMRI in three subject groups (Control, Patient1, Patient2), 10 subjects for each group.
Each subject has scanned three times, once a year (We use the condition to set this as: Year1, Year2, Year3). After pre- and post-processing, we want to know the following issues.
However, we are not clear how to set the contrast to test several interests.
First, for each group, we want to know how much functional networks (for example, DMN and language network) decrease during the three year period.
Second, we want to find a network which more decreases in the Patient2 group during the three year period compared with the Control.
I appreciate any comments.
Geon-Ho
We scanned a resting-state fMRI in three subject groups (Control, Patient1, Patient2), 10 subjects for each group.
Each subject has scanned three times, once a year (We use the condition to set this as: Year1, Year2, Year3). After pre- and post-processing, we want to know the following issues.
However, we are not clear how to set the contrast to test several interests.
First, for each group, we want to know how much functional networks (for example, DMN and language network) decrease during the three year period.
Second, we want to find a network which more decreases in the Patient2 group during the three year period compared with the Control.
I appreciate any comments.
Geon-Ho
Mar 16, 2017 07:03 PM | Alfonso Nieto-Castanon - Boston University
RE: Contrast define for longitudinal study
Dear Geon-Ho,
Simple-main effects: In order to look at decreases between year1 and year3 for control subjects, for example, simply select the 'Control' group in the subject-effects list, and then select 'Year1' and 'Year3' in the conditions list and enter a [1 -1] between-conditions contrast. This will effectively perform a paired t-test comparing the connectivity values between Time1 and Time3 for the control subjects only. If you want to extend these analyses to look at potential changes between year1 and year2 as well (which may or not be associated with changes between year1 and year3) you could also select all three conditions Year1, Year2, and Year3 and then enter a between-conditions contrast [1 -1 0; 1 0 -1].
Interaction effects: In order to look at temporal Time1-Time3 changes that may differ between the Patient2 and control groups, you may select both 'Controls' and 'Patients2' in the subject-effects list and enter a [1 -1] between-subjects contrast, and then select as before the Time1 and Time3 conditions and enter a [1 -1] between-conditions contrast. In the results explorer window remember to look at the two-sided results (unless you have an a priori expectation of the directionality of these differences between groups). These analyses will effectively perform a Time*Group interaction test in a repeated-measures anova with a within-subject factor (Time) and a betwen-subject factor (Groups).
Hope this helps
Alfonso
Originally posted by Geon-Ho Jahng:
Simple-main effects: In order to look at decreases between year1 and year3 for control subjects, for example, simply select the 'Control' group in the subject-effects list, and then select 'Year1' and 'Year3' in the conditions list and enter a [1 -1] between-conditions contrast. This will effectively perform a paired t-test comparing the connectivity values between Time1 and Time3 for the control subjects only. If you want to extend these analyses to look at potential changes between year1 and year2 as well (which may or not be associated with changes between year1 and year3) you could also select all three conditions Year1, Year2, and Year3 and then enter a between-conditions contrast [1 -1 0; 1 0 -1].
Interaction effects: In order to look at temporal Time1-Time3 changes that may differ between the Patient2 and control groups, you may select both 'Controls' and 'Patients2' in the subject-effects list and enter a [1 -1] between-subjects contrast, and then select as before the Time1 and Time3 conditions and enter a [1 -1] between-conditions contrast. In the results explorer window remember to look at the two-sided results (unless you have an a priori expectation of the directionality of these differences between groups). These analyses will effectively perform a Time*Group interaction test in a repeated-measures anova with a within-subject factor (Time) and a betwen-subject factor (Groups).
Hope this helps
Alfonso
Originally posted by Geon-Ho Jahng:
Dear CONN Experts
We scanned a resting-state fMRI in three subject groups (Control, Patient1, Patient2), 10 subjects for each group.
Each subject has scanned three times, once a year (We use the condition to set this as: Year1, Year2, Year3). After pre- and post-processing, we want to know the following issues.
However, we are not clear how to set the contrast to test several interests.
First, for each group, we want to know how much functional networks (for example, DMN and language network) decrease during the three year period.
Second, we want to find a network which more decreases in the Patient2 group during the three year period compared with the Control.
I appreciate any comments.
Geon-Ho
We scanned a resting-state fMRI in three subject groups (Control, Patient1, Patient2), 10 subjects for each group.
Each subject has scanned three times, once a year (We use the condition to set this as: Year1, Year2, Year3). After pre- and post-processing, we want to know the following issues.
However, we are not clear how to set the contrast to test several interests.
First, for each group, we want to know how much functional networks (for example, DMN and language network) decrease during the three year period.
Second, we want to find a network which more decreases in the Patient2 group during the three year period compared with the Control.
I appreciate any comments.
Geon-Ho