help > con_mat in nbs structure
Showing 1-2 of 2 posts
Feb 10, 2021 01:02 AM | asinha
con_mat in nbs structure
Dear Andrew,
I am using NBS to identify group differences in functional connectivity over several networks and do not have MNI coordinates for the ROIs, as they are based on a surface-based parcellation atlas. Below is my current setup for the statstical model:
Design Matrix:
1 0
1 0
1 0
1 0
...
0 1
0 1
...
Contrast = [-1 1] to test for group 2 > group 1
After running NBS and then accessing the nbs structure via global nbs, I get the following:
ans =
n: 1
con_mat: {[200x200 double]}
pval: 0.0012
test_stat: [200x200 double]
node_coor: [200x3 double]
node_label: {200x1 cell}
This means it found 1 significant network, correct? When I go into con-mat, I see a list of connections (i.e. (1, 7), ... (23, 143), etc.) To confirm, each of the connections in con_mat are the connections found to be significantly different between groups that are part of this 1 network, correct? Even though they span several brain networks?
I am a bit confused because even when I change the t-stat threshold, I still get con_mat with size 200x200, indicating that it found the same number of significant connections, right?
If you could provide some clarification on this, that would be greatly appreciated.
-Anita
I am using NBS to identify group differences in functional connectivity over several networks and do not have MNI coordinates for the ROIs, as they are based on a surface-based parcellation atlas. Below is my current setup for the statstical model:
Design Matrix:
1 0
1 0
1 0
1 0
...
0 1
0 1
...
Contrast = [-1 1] to test for group 2 > group 1
After running NBS and then accessing the nbs structure via global nbs, I get the following:
ans =
n: 1
con_mat: {[200x200 double]}
pval: 0.0012
test_stat: [200x200 double]
node_coor: [200x3 double]
node_label: {200x1 cell}
This means it found 1 significant network, correct? When I go into con-mat, I see a list of connections (i.e. (1, 7), ... (23, 143), etc.) To confirm, each of the connections in con_mat are the connections found to be significantly different between groups that are part of this 1 network, correct? Even though they span several brain networks?
I am a bit confused because even when I change the t-stat threshold, I still get con_mat with size 200x200, indicating that it found the same number of significant connections, right?
If you could provide some clarification on this, that would be greatly appreciated.
-Anita
Feb 10, 2021 05:02 AM | Andrew Zalesky
RE: con_mat in nbs structure
Hi Anita,
Yes - this indicates one significant subnetwork.
con_mat will always be dimensions of n x n where n is the number of regions in your connectivity matrix.
The subnetwork for which the null hypothesis can be rejected comprises the connections in this 200 x 200 matrix that are not zero.
While the network dimensions are always 200 x 200, the connections that are not zero can differ depending on the contrast and threshold.
Andrew
Originally posted by asinha:
Yes - this indicates one significant subnetwork.
con_mat will always be dimensions of n x n where n is the number of regions in your connectivity matrix.
The subnetwork for which the null hypothesis can be rejected comprises the connections in this 200 x 200 matrix that are not zero.
While the network dimensions are always 200 x 200, the connections that are not zero can differ depending on the contrast and threshold.
Andrew
Originally posted by asinha:
Dear Andrew,
I am using NBS to identify group differences in functional connectivity over several networks and do not have MNI coordinates for the ROIs, as they are based on a surface-based parcellation atlas. Below is my current setup for the statstical model:
Design Matrix:
1 0
1 0
1 0
1 0
...
0 1
0 1
...
Contrast = [-1 1] to test for group 2 > group 1
After running NBS and then accessing the nbs structure via global nbs, I get the following:
ans =
n: 1
con_mat: {[200x200 double]}
pval: 0.0012
test_stat: [200x200 double]
node_coor: [200x3 double]
node_label: {200x1 cell}
This means it found 1 significant network, correct? When I go into con-mat, I see a list of connections (i.e. (1, 7), ... (23, 143), etc.) To confirm, each of the connections in con_mat are the connections found to be significantly different between groups that are part of this 1 network, correct? Even though they span several brain networks?
I am a bit confused because even when I change the t-stat threshold, I still get con_mat with size 200x200, indicating that it found the same number of significant connections, right?
If you could provide some clarification on this, that would be greatly appreciated.
-Anita
I am using NBS to identify group differences in functional connectivity over several networks and do not have MNI coordinates for the ROIs, as they are based on a surface-based parcellation atlas. Below is my current setup for the statstical model:
Design Matrix:
1 0
1 0
1 0
1 0
...
0 1
0 1
...
Contrast = [-1 1] to test for group 2 > group 1
After running NBS and then accessing the nbs structure via global nbs, I get the following:
ans =
n: 1
con_mat: {[200x200 double]}
pval: 0.0012
test_stat: [200x200 double]
node_coor: [200x3 double]
node_label: {200x1 cell}
This means it found 1 significant network, correct? When I go into con-mat, I see a list of connections (i.e. (1, 7), ... (23, 143), etc.) To confirm, each of the connections in con_mat are the connections found to be significantly different between groups that are part of this 1 network, correct? Even though they span several brain networks?
I am a bit confused because even when I change the t-stat threshold, I still get con_mat with size 200x200, indicating that it found the same number of significant connections, right?
If you could provide some clarification on this, that would be greatly appreciated.
-Anita