Visualize Evoked data

In this tutorial we focus on plotting functions of :class:mne.Evoked.

First we read the evoked object from a file. Check out tut_epoching_and_averaging to get to this stage from raw data.

Notice that evoked is a list of :class:evoked <mne.Evoked> instances. You can read only one of the categories by passing the argument condition to :func:mne.read_evokeds. To make things more simple for this tutorial, we read each instance to a variable.

Let's start with a simple one. We plot event related potentials / fields (ERP/ERF). The bad channels are not plotted by default. Here we explicitly set the exclude parameter to show the bad channels in red. All plotting functions of MNE-python return a handle to the figure instance. When we have the handle, we can customise the plots to our liking.

All plotting functions of MNE-python return a handle to the figure instance. When we have the handle, we can customise the plots to our liking. For example, we can get rid of the empty space with a simple function call.

Now let's make it a bit fancier and only use MEG channels. Many of the MNE-functions include a picks parameter to include a selection of channels. picks is simply a list of channel indices that you can easily construct with :func:mne.pick_types. See also :func:mne.pick_channels and :func:mne.pick_channels_regexp. Using spatial_colors=True, the individual channel lines are color coded to show the sensor positions - specifically, the x, y, and z locations of the sensors are transformed into R, G and B values.

Notice the legend on the left. The colors would suggest that there may be two separate sources for the signals. This wasn't obvious from the first figure. Try painting the slopes with left mouse button. It should open a new window with topomaps (scalp plots) of the average over the painted area. There is also a function for drawing topomaps separately.

By default the topomaps are drawn from evenly spread out points of time over the evoked data. We can also define the times ourselves.

Or we can automatically select the peaks.

You can take a look at the documentation of :func:mne.Evoked.plot_topomap or simply write evoked_r_aud.plot_topomap? in your python console to see the different parameters you can pass to this function. Most of the plotting functions also accept axes parameter. With that, you can customise your plots even further. First we create a set of matplotlib axes in a single figure and plot all of our evoked categories next to each other.

Notice that we created five axes, but had only four categories. The fifth axes was used for drawing the colorbar. You must provide room for it when you create this kind of custom plots or turn the colorbar off with colorbar=False. That's what the warnings are trying to tell you. Also, we used show=False for the three first function calls. This prevents the showing of the figure prematurely. The behavior depends on the mode you are using for your python session. See http://matplotlib.org/users/shell.html for more information.

We can combine the two kinds of plots in one figure using the :func:mne.Evoked.plot_joint method of Evoked objects. Called as-is (evoked.plot_joint()), this function should give an informative display of spatio-temporal dynamics. You can directly style the time series part and the topomap part of the plot using the topomap_args and ts_args parameters. You can pass key-value pairs as a python dictionary. These are then passed as parameters to the topomaps (:func:mne.Evoked.plot_topomap) and time series (:func:mne.Evoked.plot) of the joint plot. For an example of specific styling using these topomap_args and ts_args arguments, here, topomaps at specific time points (90 and 200 ms) are shown, sensors are not plotted (via an argument forwarded to plot_topomap), and the Global Field Power is shown:

Sometimes, you may want to compare two or more conditions at a selection of sensors, or e.g. for the Global Field Power. For this, you can use the function :func:mne.viz.plot_compare_evokeds. The easiest way is to create a Python dictionary, where the keys are condition names and the values are :class:mne.Evoked objects. If you provide lists of :class:mne.Evoked objects, such as those for multiple subjects, the grand average is plotted, along with a confidence interval band - this can be used to contrast conditions for a whole experiment. First, we load in the evoked objects into a dictionary, setting the keys to '/'-separated tags (as we can do with event_ids for epochs). Then, we plot with :func:mne.viz.plot_compare_evokeds. The plot is styled with dictionary arguments, again using "/"-separated tags. We plot a MEG channel with a strong auditory response.

We can also plot the activations as images. The time runs along the x-axis and the channels along the y-axis. The amplitudes are color coded so that the amplitudes from negative to positive translates to shift from blue to red. White means zero amplitude. You can use the cmap parameter to define the color map yourself. The accepted values include all matplotlib colormaps.

Finally we plot the sensor data as a topographical view. In the simple case we plot only left auditory responses, and then we plot them all in the same figure for comparison. Click on the individual plots to open them bigger.

Visualizing field lines in 3D

We now compute the field maps to project MEG and EEG data to MEG helmet and scalp surface.

To do this we'll need coregistration information. See tut_forward for more details.

Here we just illustrate usage.