[color=#000000]Hi Toshiki,[/color]
[color=#000000]In general MVPA analyses look at the entire voxel-to-voxel correlation matrix (between all [i]source [/i]voxels and all [i]target [/i]voxels). When entering a masks, the mask limits the range of [i]target [/i]voxels considered, but the analyses are still performed across all [i]source [/i]voxels. If you find a significant cluster (anywhere in the brain, inside or outside the original mask), that indicates that there is a significant group-by-MRS interaction in the connectivity between that cluster (e.g. in your case somewhere around the brainstem) and at least some portion of the mask (e.g. in your case some unspecified cortical area; you may perform post hoc seed-to-voxel analyses using that brainstem cluster as a seed in order to better characterize "which" cortical areas in particular)[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Toshiki Iwabuchi:[/i][quote]Hello all,
To begin with, I would like to apologize about my previous post because I had overlooked FAQ regarding the issue of categorical-by-continuous interaction.
According to the FAQ instruction, I performed a group MVPA analysis, but it raised another question.
At the first-level MVPA, I checked the "Masked analyses" option and selected a gray matter mask image because I guessed that this option allowed me to restrict voxels to be tested. Then, at the 2nd-level, I selected the 2nd-level variables shown below and set the between-subjects contrast as [0 0 1 -1] to test the interaction between groups (Patient, Control) and MRS data.
Patient = [1 1 1 1 1 0 0 0 0 0]
Control = [0 0 0 0 0 1 1 1 1 1]
MRS_patient = [2.1 1.6 2.4 2.0 1.4 0 0 0 0 0]
MRS_control = [0 0 0 0 0 2.5 1.9 2.7 2.1 3.2]
(N is set to 10 for simplicity.)
The analysis was done with MATLAB 2019b and the newest version of CONN (v.20.b), and I found several significant clusters.
However, some clusters contained (or even fell into) voxels outside the gray matter mask that I specified (cf. the attached image).
Is it possible that a cluster is found outside the specified mask in group MVPA (or does it seem something went wrong in the analysis)?
If significant voxels can be observed outside gray matter, how should I interpret those results?
I would really appreciate any help.
Sincerely,
Toshiki[/quote]
[color=#000000]In general MVPA analyses look at the entire voxel-to-voxel correlation matrix (between all [i]source [/i]voxels and all [i]target [/i]voxels). When entering a masks, the mask limits the range of [i]target [/i]voxels considered, but the analyses are still performed across all [i]source [/i]voxels. If you find a significant cluster (anywhere in the brain, inside or outside the original mask), that indicates that there is a significant group-by-MRS interaction in the connectivity between that cluster (e.g. in your case somewhere around the brainstem) and at least some portion of the mask (e.g. in your case some unspecified cortical area; you may perform post hoc seed-to-voxel analyses using that brainstem cluster as a seed in order to better characterize "which" cortical areas in particular)[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Toshiki Iwabuchi:[/i][quote]Hello all,
To begin with, I would like to apologize about my previous post because I had overlooked FAQ regarding the issue of categorical-by-continuous interaction.
According to the FAQ instruction, I performed a group MVPA analysis, but it raised another question.
At the first-level MVPA, I checked the "Masked analyses" option and selected a gray matter mask image because I guessed that this option allowed me to restrict voxels to be tested. Then, at the 2nd-level, I selected the 2nd-level variables shown below and set the between-subjects contrast as [0 0 1 -1] to test the interaction between groups (Patient, Control) and MRS data.
Patient = [1 1 1 1 1 0 0 0 0 0]
Control = [0 0 0 0 0 1 1 1 1 1]
MRS_patient = [2.1 1.6 2.4 2.0 1.4 0 0 0 0 0]
MRS_control = [0 0 0 0 0 2.5 1.9 2.7 2.1 3.2]
(N is set to 10 for simplicity.)
The analysis was done with MATLAB 2019b and the newest version of CONN (v.20.b), and I found several significant clusters.
However, some clusters contained (or even fell into) voxels outside the gray matter mask that I specified (cf. the attached image).
Is it possible that a cluster is found outside the specified mask in group MVPA (or does it seem something went wrong in the analysis)?
If significant voxels can be observed outside gray matter, how should I interpret those results?
I would really appreciate any help.
Sincerely,
Toshiki[/quote]