Tutorial 4: Exploring the recordings

This tutorial is based on the TutorialCTF protocol, and will show you how to explore efficiently the sensory response to an electric stimulation of the fingers. The two previous tutorials explain how to create this protocol and describe the data that are used in it.

Display modes

Several display modes are available for recordings. They are all accessible with a right-click.The first three menus (MEG, Stimulation input, Video Time), represent all the different types of channels that were found in the channel file. You can check that with Channel Edit (right-click on channel file > Edit).

Now repeat this operation several times to display all the available modes: Display time series, 3D sensor cap, 2D sensor cap, 2D disc and 2D Layout. You should now see all the following figures. Keep them opened for the next few paragraphs, you will learn how to manipulate multiple views of the same data.

Display time series

Values over the time for all the sensors.

• Mouse wheel: Zoom horizontal

• CTRL+mouse wheel: Zoom vertical

• Right click + move: Move in zoomed figure

• Left click + move: Time selection (described later)

• Right-click: Popup menu (described later)

• Left-click: Set current time (described later)

3D Sensor cap / 2D Sensor cap / 2D Disc

The recordings at a given time instant and interpolated over a 2D or 3D surface.

• Mouse wheel: Zoom / unzoom

2D figures: Zoom with mouse wheel, move with left+right or middle mouse click+move
3D figures: regular mouse operations as introduced in previous tutorials.

• 2D layout: The time course of each channel is drown at the actual position of the electrode, projected in 2D the same way as 2D sensor
  
  

  • Mouse wheel: Zoom / unzoom

  • Left+right click + move: Move into the window

  • Shift+mouse wheel: Increase/decrease the gain of the time series (ie. the sharpness)

  • Click on a line: Select a sensor

  • Shift + click on a line: Display the time-frequency decomposition for the selected sensor, useful for later

  • Control + E: Display the channels names

• If you are experiencing any kind of graphic bug, you should try to disable the OpenGL renderer. In Brainstorm main window, select menu File > Set preferences, and select this option. All the 3D views would be very slow and there would be no transparency, but it may solve all the problems. This is more likely to happen on 64bit machines or when working on a remote system (X-server, VirtualBox, Windows remote desktop connection...).

Time exploration

The time window and the current time instant are centralized and managed by the Time window panel in Brainstorm window. You can use the text box and the slider to change the current time value. All the figures always stay synchronized, you cannot display the 2D disc topography at t=10ms and the 3D sensor cap topography at t=50ms at once. It might look too strict at the beginning, but the contrary would have made the interface much too complex; we wanted to keep it as simple as possible.

The time window is not the best way to browse your data. There are many other shortcuts that are much more intuitive to navigate through the time.

• Click on time series figure:
  Left-click in any white area of the figure and it will move the time cursor at this point, and update all the other figure to match this new value. Don't click on a black line, it would select the corresponding channel instead of moving the time cursor.

• Press the keyboard arrows:
  Click on any data figure and try to press several time the left and right arrows, and also the PageUp and PageDown keys. This feature is really useful when you want to quickly see how the recordings are changing in the time. It might be a bit slow because you have many figures to refresh at each time change. It is much faster if you have only one 2D figure to refresh.

To remember: An efficient and quick way to review your recordings just after importation is to:

1. Open a Time series window (double click on recordings file)
2. Click on the time of interest (eg. first response peak)

3. Open a topography view of your choice (2D disc, 2D sensor cap, or 3D sensor cap)
   You can do this by right-clicking on the time series figure > MEG Topography (or Ctrl+T keyboard shortcut).

4. Use the left/right arrows, and PageUp/PageDown keys

Sensors selection

You can select some channels, and then display them separately or mark them as good or bad.

1. Now close all the figures except the Time series and the 2D Sensor cap topography (if you don't remember which one it was, close everything and open again these two figures following the 4 steps in previous section).

2. On 2D Sensor cap figure: right-click anywhere on the window > Channels > Display sensors. You'll see white dots representing the center of each MEG coil. Note that there is keyboard shortcut indicated for this menu: Ctrl + E.

3. Press Ctrl + E several times and see what happens. Come back to the display with only the sensors markers (no labels).

4. Click on some white dots.
   • They turn red and the corresponding lines in the time series window also turn red.
   • Left-click on the lines in the time series window: it also selects the sensors.
   • Right-click on a red line in time series figure: you'll have the name and indice of the selected channel.
   • Click on a selected (red) point or line: the sensor is deselected.

   • To deselect all the sensors at once, press Escape key, or right-click > Channels > Reset selection

5. Now select three sensors.
   
   

6. Right-click > Channels > View selected (Shortcut = enter). A new window is created, with only those three sensors, whose names are indicated in a legend box (you can move this legend if needed).
   
   

Time series in columns

There are two display modes for the time series. Open a time series figure. To switch from the current mode to the "column" mode, click on in the toolbar.

• You will first see a figure completely useless like the following one:
  
  

• This requires a little configuration: you need to select a subset of the channels and/or to increas the gain of the channels.
• Start by maximizing the size of the figure

• Channel gain: Hold the SHIFT key down, and use the mouse wheel to increase/decrease the amplitude of the waves.

• Vertical zoom: Hold the CTRL key down, and use the mouse wheel to zoom in or out.

• Horizontal zoom: Use the mouse wheel to zoom in or out.

• Move in the zoomed figure: Hold down left and right buttons of the mouse, and move the mouse.

• You can reach some display results that are already a little more interesting:
  
  

• Even better: you can create and save subsets of channels, or used pre-defined sets. Right-click on the figure > Display setup > ... Try several predefined sets (ex. CTF LF = Group of left frontal sensors)
  
  

• Note the keyboard shortcuts: SHIFT+letter. It makes it really easy to switch from a group of sensors to another (ex: Shift+A = all sensors, Shift+B = Left-frontal, Shift+C=Left-parietal...)

• To edit/create your own selections, Right-click on the figure > Display setup > Edit selections... Click on the first button to create a new selection Test1, select a few channels in the list.
  

• Click on Save. Back to the time series figure. Look in the popup menu again, you can see your new selection in the menu: Test1. You can now use the time series view with those pre-defined selections of sensors.

Time selection

Go back to the previous view mode. Left-click somewhere on the white part of the time series figure, hold the mouse button, and drag your mouse left or right: A transparent blue rectangle appears to represent the time selection.

• Now right-click on the figure to see the functions you can apply on this time window.
  
  
  

• Set selection manually: Select a time window by giving the bounds in ms.

• Compute mean/time: Average of the selected time window. Result is saved as a new file in database.

• Export to database: Extract the time series from this time window and save them in a new file in database.

• Export to file: Same, but in a user-defined file (not in the database)

• Export to Matlab: Same, but export as a Matlab variable in the current workspace.

Good / bad channels

If you find out that a sensor has unexpected values, that are not coherent with the other surrounding sensors, you can choose to ignore it in the displays and in the source estimation process. In the channel file, each channel has a flag which indicates if is good or bad. You may also import recordings files that already have some bad channels defined.

• Mark channels as bad:
  You should still have three sensors selected in both figures. Right-click > Channels > Set selected as bad, or press Delete key. The sensors should disappear in all figures, and the topography view (2D sensors cap) is updated so that the interpolation now ignores the bad channels.
  
  

• Get the channels back: two options

  • Right-click on figure > Channels > Mark all channels as good
    

  • Right-click on figure > Channels > Edit good/bad channels... : this menu open a window very similar to the Channel Editor window introduced in previous tutorials, but without the annoying location and orientation values, and with green and red dots. Click on the dots to mark a channel as good or bad.
    
    

  • Note that if you click on a row in this window, it will select the corresponding channel in the time series and topography figures.
  • Close this window to save the changes.

• Batching this from the database explorer:

  • You will find a "Channels" menu you right-click on any node in the tree that contains recordings.
  • If you select on the of the item in this menu, the operation will be applied recursively to all the recordings contained in the node. This way, you can quickly set that the channel 63 is always bad, or that electrode 43 is not working for subject #12, without having to visualize all the recordings one after the other.

  • The View all bad channels command displays the list of all the bad channels in all the files in the Messages tab, in main Brainstorm window.
    
    

• Important notes:

  • The good/bad channel flags are stored in the recordings files, not in the channel files. So if you marked some channels as bad in the ERP data file, there are still considered as good in the Std data file.

Colormap configuration

We will now describe how to manipulate the colormap in the topography windows. Open a 2D sensor cap view, and play with colors.

• Brainstorm keeps track of many user-defined colormaps: Anatomy, EEG, MEG, Sources, Stat, Time and Time-frequency. You can go to the Colormaps menu in the Brainstorm main window to see this list. Usually, you will use only popup menus on figures to edit the colormaps.
• If you modify a colormap, the changes will be saved in your user preferences and available the next time you start Brainstorm.
• For the moment, you are going to play only with the MEG colormap, but the others work exactly the same way.

• Right-click on you 2D sensor cap figure > Colormap.
  
  

• Colormap: You can change the colors that are used to represent the recorded MEG values. You can create your own colormaps with the Matlab colormap editor, clicking on the last option Custom.

• Absolute values: Displays the absolute values of the recordings, instead of the original values.

  • This is the default for Anatomy, Sources and Stat colormaps, but it is not very useful for recordings: for EEG and MEG, the sign of the values is very important.
    
    
    Fig.1: Relative values; Fig.2: Absolute values

• Normalize for each display:

  • If not selected: The bounds of the colormap are set to the maximum value across the whole time window. Eg. if you use the rbw colormap and the min and max values are [-80ft, +130ft], the colors will be mapped in the following way: -130ft is blue, +130ft is red, 0ft is white. And it will be like this for each time sample.

  • If selected: It uses the local min and max values at the current time frame, instead of the global min and max. Eg. at t=0ms, the extrema values are roughly [-17ft, +17ft]. So the colors will be mapped in order to have: -17ft = blue, and +17 = red.
    If you uncheck this option and go to t=0ms, the 2D topography figure will turn almost white.
    
    
    Fig.1: Local maximum (normalized); Fig.2: Global maximum (not normalized)

  • You can usually keep this Normalized option when looking at recordings, it makes things nicer. But keep in mind that it is not because you see flashy colors that you have a strong effects. It's always a matter of colormap configuration.

• Set colorbar max value: disabled when Normalized option is selected, because it wouldn't have any effect. This menu allows you to fix the maximum bounds of the colormap, instead of using the automatic ones.

  • Click on it, it will ask you for the maximum value you want.

  • Set it first to 5, then do it again and then to 500; and observe that it modifies the values displayed on the right side of the color bar, in the 2D sensor cap figure.
    
    
    Fig.1: Maximum = 5 fT; Fig.2: Maximum = 500 fT

  • Select one last time this menu, and follow the instruction to reset it: enter 0. It will go back again to the automatic bounds.
  • Remember that when you use this option, it is saved in your user preferences, so it can cause some trouble: You close Brainstorm, you start it again one week later, you totally forget that you had set this value to a very high level, you try to import and review new recordings, all you see is white, you spend two hours to understand why, you install a new version of Brainstorm and it stays desperately white... Just think about checking this option.

• Contrast and Brightness:

  • Brightness moves the center of the colormap up and down.

  • Contrast saturate/desaturate the colors.
    
    
    Fig.1: High brighness; Fig.2: High contrast

  • The words brightness/contrast may not be adapted for colormaps such as rbw, jet or hsv; but try selecting the grey colormap, it suddenly makes more sense.

  • On some combinations of Matlab version / operating systems, those sliders may not work (it depends on the embedded Java virtual machine version).

  • You can also modify those values by clicking directly on the color bar in the figures. Hold the mouse button, and move up/down to change the brightness and left/right to change the contrast. Even if the sliders do not work, you can modify the colormaps.

• Display colorbar: Just in case you want to hide the color bar...

• Restore defaults: Click on it now so all your experimentations will be discarded.

  • You can also reset the colormap by double-clicking on the color bar.

Manipulating multiple windows

One the interesting points in Brainstorm interface is that you can display easily different conditions or different subjects at the same time. If you want to open at the same time many subjects in many different experimental conditions, you will quickly have a lot of small windows everywhere. The buttons on the right part of the toolbar in the main Brainstorm window may help to organize them. From left to right:

• Set Brainstorm preferences: edit some general options.

• Display mode for time series: Already introduced in this tutorial.

• Average reference: If checked, display the average reference of EEG time series instead of the real values: at each time instant, substract the average of all the channels. For EEG only...

• Uniformize time series scales: If checked, all the time series figures with similar units have the same y-axis scale, so that you can compare visually the amplitudes between two datasets.

  1. Keep the two figures you've just opened, and click on this button several times.

  2. You'll see that the scale of the amplitude axis will change in the Right window, and fit alternatively its own maximum or the Left one.
     
     Uniformize on:
     
     
     Uniformize off:
     

• Tiled layout: If checked, the figures are reorganized automatically each time their number change.

  1. Uncheck this button.

  2. Double-click successively on Right / ERP and Left / ERP to open them. The two figures appear at the same place exactly, in the middle of the screen. This is the default Matlab positionning.

  3. Now selected again this toggle button, the two figures will be re-arranged.
     
     ---

  4. Unchecking this option can be useful if the auto-arrangement does not work well on your Linux system for some reason, or if you want to organize your windows by yourself.

• Show all figures: If you have many figures and they are all hidden because of some other fullscreen window (Matlab to run Brainstorm, Firefox to read this tutorial, etc.), you don't have click on all of them in the taskbar to get them back. Just make the Brainstorm window visible and click on this button, and it will show all the data figures.

• Close all windows: Close everthing and free most of the allocated memory.

Database navigator

The Navigator menu can help you to go quickly from a dataset to another. It can be almost indispensable when your are reviewing 200 trials of the same MEG response.

• You can access it from anywhere: Brainstorm main window, and all the figures that display functional data.
  
  

• As you can see in the menu, you can also use the keyboard shortcuts: F1, F2, F3, together with Shift key to go backwards.

• Now, close all the figures (use the Close all figures button)

• For Right / ERP, display three views :

  • Time series (double click)

  • 2D sensor cap (Ctrl+T)

  • 2D Layout (right click on ERP file > Display > 2D Layout)

• Press F3 once: It updates all the figures to display now Right / Std, and it also selects the Std file in the Database explorer.

• Press F3 again: Nothing happens, but a message is displayed in Messages tab: Navigator: already at last data file. You are already at the last dataset for this subject / condition.

• Press Shift + F3: And you'll be back to ERP file.

• Press F2: It does exactly the same thing as F3, but jumps from a condition to another, within the same subject.

• If you had many subjects you could also use F1 / Shift+F1.

Snapshots

Using Brainstorm, you will quickly feel the necessity to save the beautiful images you produce. For that you can:

• Press the PrintScreen key and paste the copied screen in your favorite text editor.

• Even more subtile: Alt+PrintScreen will only copy the figure that is currently selected.

• But there is also a Snapshots menu present in the popup of all the figures, to help you do this in a more efficient way. Here are two examples of Snapshots menus, respectively on Time series figures and 2D/3D figures:
  
  ---

• Save as image: save the figure (without the title bar and borders) in a file. Many formats available.

• Save as Matlab figure: save the figure as a Matlab .fig file. Honestly, I didn't find any use for this yet...

• Export to database: get the time series in the figure, and create a new entry in database with these values.

• Save time series: extract the time series displayed in this figure, and save them in a file. Available formats:

  • ASCII (.txt)
  • Matlab (.mat): saves much more information (titles, time values, etc.)
  • Cartool (.eph)
  • EGI (.raw)

• Export to Matlab: do the same as Save time series in Matlab .mat format, but exports the structure in a variable in Matlab workspace instead of creating a new file.

• Movie (...): Create .avi movies, to show time evolution, or to rotate spatially the 3D scene.

  • Only available on MS Windows, because Matlab or Java do not provide access to any video codecs on Linux or MacOS...
  • Even on MS Windows, you only have access to the worst codec still distributed nowadays (Indeo5).
  • So you can set the compression to 99% and get an ugly video of less than 1Mb

  • Or if you need quality, you can leave the movie uncompressed (quality=100%), get an outrageously big file, and then compress it with a XviD codec using a nice and free software, such as VirtualDub.

  • The size of the movie depends on the actual size of the figure on the screen.
  • Don't do anything else while creating the movies: the figure which is captured must be visible all the time.

• Contact sheet: Produce a big image with all the time frames in it.

  • Same recommendations than for movies: if you don't want the final image to be too huge, reduce the size of your figure, zoom a bit, and maybe eventually the colorbar; and don't hide the figure during the capture.

  • At the end, the image is displayed in a viewer with which you can zoom (menu or wheel), move (left-click+move), and save the image (File > Save as).
    
    

• Figure menu: Before you save you figure, you can also use the Figure menu in popup, with which you can edit the figures with the Matlab tools.

Keyboard shortcuts

Here is a memo with all the keyboard shortcuts of time series and topography figures. If you don't remember, you can find most of them in the figure popup menus.

• Arrows: left, right, PageUp, PageDown: Move in time

• Delete: Mark selected sensors as bad

• Shift + Delete: Mark non-selected sensors as bad (=keeps ony the selected sensors)

• Enter: View time series for selected sensors

• Shift + Enter: Set all the bad sensors as good (=brings back all the channels in the display)

• Escape: Unselect all the selected sensors

• Ctrl + A: Show axis on 3D figures (X,Y,Z)

• Ctrl + A: Add event in raw file viewer (see raw file viewer tutorial)

• Ctrl + B: Set trial as bad

• Ctrl + D: Dock/undock figure in Matlab's figures list

• Ctrl + E: Show sensors and labels (E stands initially for Electrode)

• Ctrl + I: Save figure as image

• Ctrl + R: Open Time series view (R stands for Recordings)

• Ctrl + S: Open Sources view (see next tutorial)

• Ctrl + T: Open 2D sensor cap view (T stands for Topography)

• Shift + letter: Change sensors display when in "column" display mode for the time series.

• F1, F2, F3: with or without Shift, calls the database navigator (F1=subject, F2=condition, F3=file)

Next

Now you are able to use all the tools that are available for importing and displaying your MEG or EEG recordings, and should start to feel comfortable with the Brainstorm database.

The next two steps are the resolution of the forward problem and the inverse problem. ?Let's go.