Dear João,
On Thu, 16 Jan 2020 19:02:55 +0000, João Duarte <[log in to unmask]> wrote:
>Dear SPMers,
>
>I am using CAT12 to analyze cortical thickness and gyrification data.
>After having a smp-T map I transform it to log-p and open it with CAT12
>"show surface results".
>I now am trying to extract the actual thickness and gyrification values for
>each image within each cluster in the spm-T map.
>What do the values in "y" represent when I call the function "plot mean
>data inside cluster" by clicking in a surface spm-T map?
>As far as I understand, the raw values are the actual measure values (e.g.
>cortical thickness values) inside that cluster for each image that was
>entered in the statistical model that resulted in the spm-T map.
>Am I right?
Yes, these are the raw cortical thickness values.
>
>Then, what does "adjusted" and "predicted" mean?
SPM is using the estimated beta values (and the resulting contrast values based on the given contrast) plus the error estimate to obtain the predicted values that are also due to mean correction in the model often below zero. If you also have any nuisance parameters the "adjusted" values additionally consider this.
I tried to extract
>"adjusted" and "predicted" mean data inside a cluster in a spm-T map of
>gyrification and I get low positive and negative values. And a bar plot of
>the selected contrast. Shouldn't the contrast be just one value? Why do I
>end up with one value per image? As far as I understand I can't have a
>contrast value per image, as the contrast tests a difference between 2
>groups of images. Am I missing something?
You are right that a difference contrast should have even one value based on the contrast image. However, it is better to use the effects of interest for plotting the parameter estimates, which now shows the beta estimates for each column of the model where the contrast is defined. This helps to identify the mean values for each group (which are always mean corrected due to the GLM model).
Best,
Christian
>
>Hope you can help me clarify this.
>
>Thanks in advance.
>
>Best,
>João Duarte
>
|