172 lines
12 KiB
Plaintext
172 lines
12 KiB
Plaintext
/**
|
|
\page Doc_Mensia_AdvViz_Concepts Concepts
|
|
|
|
\section Doc_Mensia_AdvViz_Concepts_Intro Introduction
|
|
The <b>Mensia Advanced Visualization Toolset</b> is a collection of boxes dedicated to
|
|
the visualization of the result of electrophysiological signal analysis, and are especially suitable for the <b> real-time
|
|
analysis of EEG signals </b>, from raw signal display to 3D source reconstruction.
|
|
|
|
It addresses many different use-cases among users.
|
|
Neurophysiologists can observe accurately in real-time <b> spatial and
|
|
temporal patterns </b> in the brain activity (motor activity, cognitive processes). EEG signal
|
|
processing specialists can <b>evaluate and compare</b> instantly algorithms effects (source
|
|
separation, denoising techniques). BCI researchers can study how their ERP-based system may
|
|
be tuned to elicit and detect the best brain response.
|
|
|
|
|
|
\image html designer-box-list.png "Simple integration in the graphical user interface"
|
|
\image latex designer-box-list.png "Simple integration in the graphical user interface" width=\textwidth
|
|
|
|
\section Doc_Mensia_AdvViz_Concepts_VisualizationParadigms Visualization paradigms
|
|
This Toolset has been designed to be very versatile. The main design concept revolves around
|
|
the data presentation. You basically want to display matrices of numbers which may have temporal,
|
|
and/or spatial meanings. The most adapted data presentation may vary from one case to another,
|
|
according to the type of events or patterns on which you need to get a good contrast.
|
|
|
|
Before choosing the right visualization box, ask yourself:
|
|
- How do I want my data to be displayed ? curves ? levels ?
|
|
- What will be the best way to <b> enhance the contrast </b> between the information I want to extract and the rest of the data ?
|
|
- Is my data stream <b>continuous</b> in time ? or am I dealing with discontinuous epochs (e.g. ERPs) ?
|
|
|
|
To be adapted in most situation, the Mensia Advanced Visualization Toolset has been designed to cover
|
|
different visualization paradigms. Take a look at all the possibilities and choose what will best fit your needs.
|
|
- \ref Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Oscilloscope
|
|
- \ref Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Bars
|
|
- \ref Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Bitmap
|
|
- \ref Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Topo
|
|
- \ref Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Reco
|
|
|
|
You can also have a look at the \ref Doc_Mensia_AdvViz_UseCases "list of use-cases", showing how each box can be used on concrete, real-life examples.
|
|
|
|
\subsection Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Oscilloscope The Oscilloscope view
|
|
It is the most basic paradigm, used to display temporal numerical data in the form of <b> curves </b> (dots linked by lines).
|
|
The Oscilloscope views are all expecting <b>centered</b> values (i.e. distributed around 0).
|
|
Hence it is advised to use at least one temporal filter (e.g. band passing between 2 and 40 Hz using a \ref Doc_BoxAlgorithm_TemporalFilter box) before displaying an EEG signal.
|
|
|
|
Four boxes use this paradigm:
|
|
- <b> \ref Doc_BoxAlgorithm_ContinuousOscilloscope </b> box: displays continuous data from left to right on a defined horizontal scale (goes back to origin upon reaching the end of the scale),
|
|
channels are displayed vertically one after another, but spikes may overlap.
|
|
- <b> \ref Doc_BoxAlgorithm_InstantOscilloscope </b> box: displays each block of data received as it comes, filling all the horizontal space available.
|
|
- <b> \ref Doc_BoxAlgorithm_ContinuousMultiOscilloscope </b> box: same as the Continuous Oscilloscope, but every input channels are displayed along the same horizontal axis with a different color, additively.
|
|
- <b> \ref Doc_BoxAlgorithm_InstantMultiOscilloscope </b> box: same as the Instant Oscilloscope, but every input channels are displayed along the same horizontal axis with a different color, additively.
|
|
|
|
<b>Example</b>: raw EEG signal display.
|
|
\image html ContinuousOscilloscope_Display.png "Continuous Oscilloscope displaying 2 EEG channels"
|
|
\image latex ContinuousOscilloscope_Display.png "Continuous Oscilloscope displaying 2 EEG channels" width=10cm
|
|
|
|
\subsection Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Bars The Bar view
|
|
Like histograms, this paradigm can be used to display and compare <b> series of levels </b> . Levels are displayed one after another from left to right, within a <b> color gradient </b> .
|
|
Channels are displayed vertically, one after another with a fixed interval (thus some "high" levels may overlap).
|
|
With a high definition (i.e. a rather high frequency display), the result can be viewed as a curve colored below the line.
|
|
|
|
Two boxes uses this paradigm:
|
|
- <b> \ref Doc_BoxAlgorithm_ContinuousBars </b> box: displays continuous data from left to right on a defined horizontal scale (goes back to origin upon reaching the end of the scale).
|
|
- <b> \ref Doc_BoxAlgorithm_InstantBars </b> box: displays each block of data received as it comes, filling all the horizontal space.
|
|
|
|
<b>Example</b>: spectrum display.
|
|
\image html InstantBars_Display.png "Instant Bars displaying the signal spectrum"
|
|
\image latex InstantBars_Display.png "Instant Bars displaying the signal spectrum" width=10cm
|
|
|
|
\subsection Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Bitmap The Bitmap view
|
|
The bitmap paradigm displays matrices of data using a color gradient. The result is a <b> 2D map where each cell is given a color "bit" </b> .
|
|
This view using colors can enhance easily the constrast between 2 temporal or spatial patterns, as the difference
|
|
between "cold" and "hot" colors is quickly caught by the analyst's eye.
|
|
You can even add an additional dimension by using <b> stacked bitmaps </b> : every time a new bitmap is received, it is placed on top or left to the previous one.
|
|
|
|
Four boxes uses this paradigm:
|
|
- <b> \ref Doc_BoxAlgorithm_ContinuousBitmap </b> box: displays continuous data from left to right on a defined horizontal scale (goes back to origin upon reaching the end of the scale).
|
|
- <b> \ref Doc_BoxAlgorithm_InstantBitmap </b> box: displays each block of data received as it comes, filling all the horizontal space.
|
|
- <b> \ref Doc_BoxAlgorithm_StackedBitmapVertical </b> box: each bitmap is placed on <b> top </b> of the previous one.
|
|
- <b> \ref Doc_BoxAlgorithm_StackedBitmapHorizontal </b> box: each bitmap is placed <b> left </b> to the previous one.
|
|
|
|
<b>Example</b>: Time-frequency map.
|
|
\image html StackedBitmapHorz_Display.png "Stacked Bitmap (Horizontal) displaying the result of a Time-Frequency analysis"
|
|
\image latex StackedBitmapHorz_Display.png "Stacked Bitmap (Horizontal) displaying the result of a Time-Frequency analysis" width=10cm
|
|
|
|
\subsection Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Topo The Topographic view
|
|
This paradigm adds a strong spatial constraint on the input data: each channel must be <b> labelled
|
|
with an electrode name </b> in a defined nomenclature, such as the standard 10-20 system.
|
|
Please see \ref Doc_Mensia_AdvViz_Concepts_ChannelLocalization for further details.
|
|
|
|
Here again the data itself is displayed using a color gradient, mapped to a 2D or 3D model using <b> spherical spline interpolation</b>.
|
|
For more details about the spherical spline interpolation, please check <i>F. Perrin, J. Pernier, O. Bertrand, J.F. Echallier,
|
|
Spherical splines for scalp potential and current density mapping, Electroencephalography and Clinical Neurophysiology, Volume 72, Issue 2, February 1989, Pages 184-187</i>.
|
|
|
|
The 2D model is a planar projection of the scalp, covering the scalp roughly from the frontal area to the occipital area (i.e. from Fp1-Fp2 to O9-O10 sites).
|
|
The projection result takes the shape of a disk with a crescent growth at the back for the occipital region.
|
|
|
|
Three boxes uses this paradigm:
|
|
- <b> \ref Doc_BoxAlgorithm_2DTopography </b> box: maps the input (which channels are labelled in the 10-20 system standard) to a planar projection of the scalp.
|
|
- <b> \ref Doc_BoxAlgorithm_3DTopography </b> box: maps the input (which channels are labelled in the 10-20 system standard) to a projection on a 3D model of the scalp.
|
|
- <b> \ref Doc_BoxAlgorithm_3DCubes </b> box: an alternative view where each channel is represented by a 3D cube, positionned in space as the electrode would be on the 3D model.
|
|
The activity is rendered by changing the size and color of the cubes.
|
|
|
|
<b>Example</b>: Displaying the power of a specific frequency band on a 3D head model.
|
|
\image html 3DTopography_Display.png "Alpha power mapped on a head model using the 3D topography"
|
|
\image latex 3DTopography_Display.png "Alpha power mapped on a head model using the 3D topography" width=10cm
|
|
|
|
\subsection Doc_Mensia_AdvViz_Concepts_VisualizationParadigms_Reco The Reconstruction view
|
|
Tomographic reconstruction algorithms offer an inside look, into the brain, from only surface measurements.
|
|
Several techniques exist, including the algorithms of the popular LORETA family which slice the brain in a stack of little cubes called voxels,
|
|
and computes the <em>inverse model</em>, a model reconstructing the sources of the potentials acquired at the measurement site.
|
|
|
|
One box implements the source reconstruction view:
|
|
- <b> \ref Doc_BoxAlgorithm_3DSourceVisualization </b> box : displays a 3D source reconstruction using 2394 colored/translucent voxels in a 3D head model.
|
|
This box expects 2394 input channels, produced by an inverse model (i.e. a spatial filter with N sensor inputs for 2394 sources outputs). This model must be
|
|
tailor-made for the precise EEG setup being used (e.g. using sLORETA).
|
|
|
|
\image html 3DTomographicVisualization_Display.png "3D tomographic reconstruction using the 3D Tomographic Visualization box"
|
|
\image latex 3DTomographicVisualization_Display.png "3D tomographic reconstruction using the 3D Tomographic Visualization box" width=10cm
|
|
|
|
\section Doc_Mensia_AdvViz_Concepts_ChannelLocalization Channel localization
|
|
Every visualization box can use the spatial information conveyed by the electrode naming. The channels can be positionned relatively to each
|
|
other as long as you provide in the box settings a file containing the cartesian coordinates of the electrodes.
|
|
Most of the time, EEG manufacturers use the 10-20 system as an electrode naming standard. For convenience, we provide within the Toolset a file compiling all the
|
|
coordinates of the electrodes in the 10-20 system.
|
|
|
|
The cartesian coordinates of all the electrodes are computed in the 3D space, where the origin is at the center of [Fpz,Oz] and [T7,T8].
|
|
- the X axis goes from the occipital lobe to the frontal lobe
|
|
- the Y axis goes from the right temporal lobe to the left temporal lobe
|
|
- the Z axis goes from the center of the head to the top
|
|
|
|
And as for the unit, here are some key points at the maximum of the axis:
|
|
- Fpz (1,0,0)
|
|
- Oz (-1,0,0)
|
|
- T7 (0,1,0)
|
|
- T8 (0,-1,0)
|
|
- Cz (0,0,1)
|
|
|
|
The following figures illustrates the cartesian coordinates of the extended 10-20 system used in the Mensia Advanced Visualization Toolset.
|
|
\image html CartesianCoordinates1.png "Cartesian coordinates of the 10-20 system, side view."
|
|
\image latex CartesianCoordinates1.png "Cartesian coordinates of the 10-20 system, side view." width=8cm
|
|
\image html CartesianCoordinates2.png "Cartesian coordinates of the 10-20 system, front view."
|
|
\image latex CartesianCoordinates2.png "Cartesian coordinates of the 10-20 system, front view." width=8cm
|
|
|
|
For more information, please see <i>Oostenveld, R. & Praamstra, P. (2001). The five percent electrode system
|
|
for high-resolution EEG and ERP measurements. Clinical Neurophysiology, 112:713-719</i>
|
|
|
|
Please note that using the 10-20 system is not mandatory. To use all the Toolset features related to the spatial disposition of the electrodes, you
|
|
just need to provide a file that maps electrode name with their coordinates in the space described above.
|
|
|
|
The format of this file is simple text. You must provide:
|
|
- the electrode names as a list of quoted labels
|
|
- the coordinate system labels
|
|
- the electrode coordinates of the electrodes, in the same order as in the electrode names
|
|
|
|
For example:
|
|
\code
|
|
[
|
|
["O1" "O2" ... ]
|
|
["x" "y" "z" ]
|
|
]
|
|
[
|
|
[-0.309017 -0.951057 4.48966e-011 ]
|
|
]
|
|
[
|
|
[0.309017 -0.951057 4.48966e-011 ]
|
|
]
|
|
...
|
|
\endcode
|
|
|
|
For a complete example, please look at the file provided with the Toolset (<i>../share/mensia /openvibe-plugins/cartesian.txt</i>)
|
|
*/ |