We will use the equivalent current dipole (ECD) spatial model with the following dipoles and their prior locations:Īs we are modeling a visual response, use a prior onset of sensory input at 70ms with a standard deviation of 16ms determining its duration.Ĭlick the red arrow to continue.
Here, use linear detrending (1), no subsampling (1) and 8 eigenmodes.Ĭlick the red arrow to continue - this will activate the next part of the “DCM for M/EEG” window where you will be able to specify the electromagnetic model. Finally, you can model a limited number of spatiotemporal modes explaining the most variance in your data. Second, you can only use subsampled data (e.g., every 2nd or every 4th datapoint). First, you can remove a number of subsequent polynomial trends (linear, quadratic, cubic etc.) from the data. There are several options to reduce the data before modeling.
You can call this effect “faces” in the window on the left. ), and specify the effect in the window below as – this will model the modulatory effects of face presentation, with “scrambled faces” treated as a baseline. As we want to model the main effect of faces, under “between-trial effects” we can leave all three conditions (i.e. You can access this information by loading the data into SPM (in Matlab: “D = spm_eeg_load(filename)”) and inspecting the “D.conditions” array. In the data file, the conditions are ordered as follows: 1 – famous faces, 2 – unfamiliar faces, 3 – scrambled faces. This will force the signals to decay towards the window edges. We will model the first 400ms of the data (time window: 1 – 400ms) and use a Hanning window. We will want to use the standard neuronal model based on neural masses with three subpopulations per region, so in the second drop-down menu select “ERP” The new pop-up window shows (on the left) the event-related field averaged across trials for single channels (separate lines), and (on the right) the same data in columns.Īs we will model the evoked potentials (as opposed to e.g., induced responses or cross-spectral density), in the first drop-down menu please select “ERP” Select “MEG” as the modality to be modeled. A figure of this panel has been included at the end of the tutorial.Ĭlick “new data” and select the file containing low-pass filtered data (“fwmPapMcbdspmeeg_run_01_sss.mat”) In the Menu window press the DCM button – this will open a new window called “DCM for M/EEG”. Now you can specify one DCM model for this subject’s data: The dataset prepared for you has been low-pass filtered with a 48Hz cut-off. With DCM for evoked responses one typically models smooth ERP/ERF deflections. In this demo we will specify one subject’s dynamic causal model, compare several models across participants, and look at posterior estimates of parameters (connectivity weights).
0 kommentar(er)
