Difference between revisions of "AnyWave:H2"

From WikiMEG
Jump to: navigation, search
(How to run)
(The result file)
 
(31 intermediate revisions by one user not shown)
Line 1: Line 1:
 
=Purpose=
 
=Purpose=
 
The plugin computes the connectivity between channels using h2/r2 algorithms.<br/>
 
The plugin computes the connectivity between channels using h2/r2 algorithms.<br/>
It will produce a .mat file containing the results. This file can be then processed either in MATMAB or with the Correlation Graphs plugin of AnyWave.<br/>
+
It will produce a .mat file containing the results. This file can be then processed either in MATLAB or by the Correlation Graphs plugin of AnyWave.<br/>
==How to run==
+
=Data input=
 +
==channels==
 +
This plugin requires at least two channels as input.<br/>
 +
The user can select the channels of interest in the signal view before running the plugin.<br/>
 +
'''NOTE:''' if no selection exists when running the plugin then ALL the channels will be set as input: A warning prompt will ask you to confirm that.<br/>
 +
==markers==
 +
The user may also used markers to select the different part of data on which to compute.<br/>
 +
It is possible to run the plugin on selected markers or to use the Main GUI once the plugin has started to specify the data input.<br/>
 +
'''NOTE:''' if there is no suitable marker (either no markers at all or markers with no duration) then the plugin will compute on the whole data.<br/>
 +
=How to run=
 
''Step 1 - Selected channels of interest'':<br/>
 
''Step 1 - Selected channels of interest'':<br/>
 
Before running the plugin, make sure you have selected the channels of interest.<br/>
 
Before running the plugin, make sure you have selected the channels of interest.<br/>
 
If no channels are selected before running the plugin, you will get a warning message.
 
If no channels are selected before running the plugin, you will get a warning message.
Computing on many channels can lead to a long calculation as many pairs of channels must be considered.<br/>
+
Computing on many channels can lead to long calculation as many pairs of channels must be considered.<br/>
 
''Step 2 - Select data of interest'':<br/>
 
''Step 2 - Select data of interest'':<br/>
 
Mark the data of interest with one or more markers.<br/>
 
Mark the data of interest with one or more markers.<br/>
Line 13: Line 22:
 
There are two ways to run the plugin:<br/>
 
There are two ways to run the plugin:<br/>
 
* Running using the Processes menu of AnyWave.
 
* Running using the Processes menu of AnyWave.
* Running user the markers GUI.
+
* Running from the markers GUI.
 +
==Run from the menu==
 +
[[File:h2Menu.png]]<br />
 +
==Run from markers==
 +
Select the markers of interest in the widget, then right click and pick the plugin to run.<br/>
 +
[[File:LaunchUsingMarkers.png]]<br />
  
If you have marked data, you can specify which artefacted parts of the data to avoid in computation ''using the skip data using markers'' options:<br/>
+
==Main GUI==
* select the marker corresponding to the part to skip (ie 'artefact').
+
[[File:H2GUI.png]]<br />
* click Skip button to add it to the list of markers to skip.
+
 
<br/>
 
<br/>
In the same way, you can also specify on which particular marked data compute the connectivity:<br/>
+
''Computation parameters'':<br/>
* select the marker corresponding to a period of interest.
+
Change here the computation method (h² or r²) and the timing parameters.<br/>
* click the Use button to add it to the list of markers to use for computation.
+
<br/>
===Frequency bands===
+
''Data input'':<br/>
 +
- Step 1 (OPTIONAL) : reduce computation time by downsampling the data.<br/>
 +
By default the ratio is 1 meaning that no downsampling will be applied.<br/>
 +
Change the ratio if you want to reduce computation time.<br/>
 +
<br/>
 +
- Step 2 : skip AND/OR use markers (this section will be hidden if you have launched the the plugin using the markers).<br/>
 +
<br/>
 +
''Avoiding artefacted data'' (OPTIONAL):<br/>
 +
Pick a marker name that matches an artefacted region in the data and click the skip button to add it to the main list of markers to avoid.<br>
 +
All the markers with that name will be skipped.<br/>
 +
[[File:h2skipmarker.png]]<br />
 +
<br/>
 +
''Using markers'':<br/>
 +
Pick a marker name that matches data region you want to use and click the Use button to add the name to the main list of markers to use.<br/>
 +
[[File:h2usemarker.png]]<br />
 +
<br/>
 +
In our example we will compute the correlations on all markers named h2 avoiding markers named artefact that may overlapped the h2 markers.<br/>
 +
<br/>
 +
'''IMPORTANT''': If you dont specify markers to use, the plugin will use ALL the markers available (with a duration). <br/>
 +
 
 +
==Frequency bands==
 
You can compute the connectivity on different frequency bands.<br/>
 
You can compute the connectivity on different frequency bands.<br/>
 
By default, the ''AnyWave band'' is selected which means the filter options set in AnyWave will be used.<br/>
 
By default, the ''AnyWave band'' is selected which means the filter options set in AnyWave will be used.<br/>
 
You can enable the different predefined bands and even change the frequency boundaries.<br/>
 
You can enable the different predefined bands and even change the frequency boundaries.<br/>
  
==Review resulting graphs==
+
=Run in batch mode=
When the computation is finished, the resuslts are saved to a MATLAB file located in the data file's folder.<br/>
+
Another plugin is automatically launched allowing you to browse the different connectivy graphs:<br/>
+
[[File:h2graphs_ui.png]]<br />
+
The UI propose all the graphs resulting from the computation. A graph for each frequency band.<br/>
+
You can see what algorihtm was used on what frequency band and the number of sections which is the number of selected part of data used for computation.<br/>
+
<br/>
+
Click on rows to select them and click on View to see the graphs.<br/>
+
 
+
=Batch mode=
+
 
The plugin is able to run in batch mode.<br/>
 
The plugin is able to run in batch mode.<br/>
 +
You will need to write a shell script and create a json file.<br/>
 +
<br/>
 
See [[AnyWave::CLI|this section for more details]]<br/>
 
See [[AnyWave::CLI|this section for more details]]<br/>
 
<br/>
 
<br/>
Line 59: Line 85:
 
| marker_file || path to the .mrk file to use || if not specified, AnyWave will try to locate the .mrk associated with the data file and use it.
 
| marker_file || path to the .mrk file to use || if not specified, AnyWave will try to locate the .mrk associated with the data file and use it.
 
|-  
 
|-  
| use_markers || array of marker labels to use || optional but may required marker_file option.
+
| use_markers || array of marker labels to use || optional but may require marker_file option.
 
|-
 
|-
|skip_markers ||array of marker labels to SKIP (artefacts) ||  optional but may required marker_file option.
+
|skip_markers ||array of marker labels to SKIP (artefacts) ||  optional but may require marker_file option.
 
|-
 
|-
 
| hp || high pass filter in Hz ||  optional
 
| hp || high pass filter in Hz ||  optional
Line 103: Line 129:
 
anywave --run h2.json --input_file d:\data\data_file.eeg --marker_file d:\data\data_file.mrk --hp 1 --lp 45
 
anywave --run h2.json --input_file d:\data\data_file.eeg --marker_file d:\data\data_file.mrk --hp 1 --lp 45
 
</syntaxhighlight>
 
</syntaxhighlight>
This will compute compute r2 values for the file data_file.eeg using 1-45Hz frequency band only on data marked by a EI marker.
+
This will compute compute r2 values for the file data_file.eeg using 1-45Hz frequency band only on data marked by EI markers.
 +
 
 +
=The result file=
 +
The plugin will generate a MATLAB file named using the data file name as a prefix.<br/>
 +
The name will also contain the frequency band defined and the algorithm used to compute.<br/>
 +
<br/>
 +
'''NOTE''': If several frequency bands where defined in the main GUI, the plugin will generate a result file for each of the frequency band.<br/>
 +
==variables==
 +
The MATLAB file contains the following variables:<br/>
 +
* aw_h2                : a 3D matrix containing the correlation values. The dimensions are ''n x n x it'' (where ''n'' is the number of channels and ''it'' the number of time windows computed)
 +
* aw_lag              : a 3D matrix containing the lag values. Same dimensions than aw_h2.
 +
* aw_maxLag            : reminder of the max lag parameter used.
 +
* aw_step              : reminder of the step parameter used.
 +
* aw_windowSize        : reminder of the time_window paremeter used.
 +
* band                : a string containing the name of the frequency band used. This variable is empty if the plugin ran in batch mode.
 +
* electrode_names      : ordered labels of all channels.
 +
* filters              : values of the frequency band used.
 +
* nb_section          : The total number of data parts used by the plugin.
 +
* section              : an index vector containing the section number of each slices in the 3D matrices. ('''WARNING''' indexes start at zero).
 +
* section_iterations  : a vector containing the total number of iterations (or slices) for each sections.
 +
* time                : a vector containing the sample index of each slices in the 3D matrices.
 +
* time_s              : a vector containing the position in seconds in the file of each slices of the 3D matrices.
 +
* samplingRate        : sampling rate of the data.
 +
* method              : the algorithm used.
 +
* year, month, day    : recorded date (may contain weird values if the data file does not contain a valid recording date.)
 +
* hour, minute, second : recorded time (may contain weird values if the data file does not contain a valid recording time.)
 +
==Notice==
 +
No matters the number of markers set as input, all the results are stacked in the matrices.<br/>
 +
So for two markers which don't overlap each other, there will be two sections in the .mat file.<br/>
 +
However, all the time windows computed are stacked in the matrix. Each slice is an iteration of a sections.<br/>
 +
<br/>
 +
Using the section vector will give you the section number (starting at zero) of a slice in the matrices.<br/>
 +
That way, it is possible to get the correlation values through time.<br/>

Latest revision as of 15:36, 9 August 2019

Purpose

The plugin computes the connectivity between channels using h2/r2 algorithms.
It will produce a .mat file containing the results. This file can be then processed either in MATLAB or by the Correlation Graphs plugin of AnyWave.

Data input

channels

This plugin requires at least two channels as input.
The user can select the channels of interest in the signal view before running the plugin.
NOTE: if no selection exists when running the plugin then ALL the channels will be set as input: A warning prompt will ask you to confirm that.

markers

The user may also used markers to select the different part of data on which to compute.
It is possible to run the plugin on selected markers or to use the Main GUI once the plugin has started to specify the data input.
NOTE: if there is no suitable marker (either no markers at all or markers with no duration) then the plugin will compute on the whole data.

How to run

Step 1 - Selected channels of interest:
Before running the plugin, make sure you have selected the channels of interest.
If no channels are selected before running the plugin, you will get a warning message. Computing on many channels can lead to long calculation as many pairs of channels must be considered.
Step 2 - Select data of interest:
Mark the data of interest with one or more markers.
It's a good practice to give the markers an explicit name.
Step 3 - Run the plugin:
There are two ways to run the plugin:

  • Running using the Processes menu of AnyWave.
  • Running from the markers GUI.

Run from the menu

H2Menu.png

Run from markers

Select the markers of interest in the widget, then right click and pick the plugin to run.
LaunchUsingMarkers.png

Main GUI

H2GUI.png

Computation parameters:
Change here the computation method (h² or r²) and the timing parameters.

Data input:
- Step 1 (OPTIONAL) : reduce computation time by downsampling the data.
By default the ratio is 1 meaning that no downsampling will be applied.
Change the ratio if you want to reduce computation time.

- Step 2 : skip AND/OR use markers (this section will be hidden if you have launched the the plugin using the markers).

Avoiding artefacted data (OPTIONAL):
Pick a marker name that matches an artefacted region in the data and click the skip button to add it to the main list of markers to avoid.
All the markers with that name will be skipped.
H2skipmarker.png

Using markers:
Pick a marker name that matches data region you want to use and click the Use button to add the name to the main list of markers to use.
H2usemarker.png

In our example we will compute the correlations on all markers named h2 avoiding markers named artefact that may overlapped the h2 markers.

IMPORTANT: If you dont specify markers to use, the plugin will use ALL the markers available (with a duration).

Frequency bands

You can compute the connectivity on different frequency bands.
By default, the AnyWave band is selected which means the filter options set in AnyWave will be used.
You can enable the different predefined bands and even change the frequency boundaries.

Run in batch mode

The plugin is able to run in batch mode.
You will need to write a shell script and create a json file.

See this section for more details

json settings for h2/r2:

{ 
"plugin" : "h2",
"algorithm" : "r2",
"time_window" : 4,
"max_lag" : 0.1,
"step" : 1,
"downsampling_factor" : 2,
"output_prefix" : "my_result"
}

Common options that can also be specified in the command line if you want to apply them to a specific file and not ALL the files:

key description default behavior
marker_file path to the .mrk file to use if not specified, AnyWave will try to locate the .mrk associated with the data file and use it.
use_markers array of marker labels to use optional but may require marker_file option.
skip_markers array of marker labels to SKIP (artefacts) optional but may require marker_file option.
hp high pass filter in Hz optional
lp low pass filter in Hz optional

H2 specific options:

key description requirement default behavior
plugin Name of plugin MANDATORY n/a
time_window the time window in s. MANDATORY n/a
max_lag maximum lag in s. MANDATORY n/a
step step in s. MANDATORY n/a
downsampling_factor downsampling data OPTIONAL default is 1. 2 means the sampling rate of the data will be divided by 2.
output_prefix prefix of resulting file name OPTIONAL default is no prefix.

Example

using this json file:

{ 
"plugin" : "h2",
"algorithm" : "r2",
"time_window" : 4,
"max_lag" : 0.1,
"step" : 1,
"downsampling_factor" : 2,
"output_prefix" : "my_result",
"use_markers" : [ "EI" ]
}

We will compute on a file stored in D:\Data\

anywave --run h2.json --input_file d:\data\data_file.eeg --marker_file d:\data\data_file.mrk --hp 1 --lp 45

This will compute compute r2 values for the file data_file.eeg using 1-45Hz frequency band only on data marked by EI markers.

The result file

The plugin will generate a MATLAB file named using the data file name as a prefix.
The name will also contain the frequency band defined and the algorithm used to compute.

NOTE: If several frequency bands where defined in the main GUI, the plugin will generate a result file for each of the frequency band.

variables

The MATLAB file contains the following variables:

  • aw_h2  : a 3D matrix containing the correlation values. The dimensions are n x n x it (where n is the number of channels and it the number of time windows computed)
  • aw_lag  : a 3D matrix containing the lag values. Same dimensions than aw_h2.
  • aw_maxLag  : reminder of the max lag parameter used.
  • aw_step  : reminder of the step parameter used.
  • aw_windowSize  : reminder of the time_window paremeter used.
  • band  : a string containing the name of the frequency band used. This variable is empty if the plugin ran in batch mode.
  • electrode_names  : ordered labels of all channels.
  • filters  : values of the frequency band used.
  • nb_section  : The total number of data parts used by the plugin.
  • section  : an index vector containing the section number of each slices in the 3D matrices. (WARNING indexes start at zero).
  • section_iterations  : a vector containing the total number of iterations (or slices) for each sections.
  • time  : a vector containing the sample index of each slices in the 3D matrices.
  • time_s  : a vector containing the position in seconds in the file of each slices of the 3D matrices.
  • samplingRate  : sampling rate of the data.
  • method  : the algorithm used.
  • year, month, day  : recorded date (may contain weird values if the data file does not contain a valid recording date.)
  • hour, minute, second : recorded time (may contain weird values if the data file does not contain a valid recording time.)

Notice

No matters the number of markers set as input, all the results are stacked in the matrices.
So for two markers which don't overlap each other, there will be two sections in the .mat file.
However, all the time windows computed are stacked in the matrix. Each slice is an iteration of a sections.

Using the section vector will give you the section number (starting at zero) of a slice in the matrices.
That way, it is possible to get the correlation values through time.