Difference between revisions of "AnyWave:WriteMatlabScripted"

From WikiMEG
Jump to: navigation, search
(aw_getdata)
(aw_getdata)
Line 146: Line 146:
 
%      if this field is not specified, AnyWave will return ALL the
 
%      if this field is not specified, AnyWave will return ALL the
 
%      available data.
 
%      available data.
 +
%  NOTE:
 +
%  start and duration are depecrated. Consider using chunks starting with the september 18 release of AnyWave.
 +
%
 +
%  Starting with september 18 release, you can use chunks vector to get one or several part of the signals at once.
 +
%  cfg.chunks = [10 20]; % at least one chunk
 +
%  a chunk must have at least two values (start and duration).
 +
%  NOTE: if start and duration are used, chunks will be ignore.
 +
%
 
%  cfg.labels = {'A1', 'A2'};
 
%  cfg.labels = {'A1', 'A2'};
 
%      cell array of strings to identify the required channel labels.
 
%      cell array of strings to identify the required channel labels.

Revision as of 13:36, 17 August 2018

Introduction

This section targets people who have a good knowledge and practice of the MATLAB programming language.
The purpose is to explain how to write a MATLAB function that will be the heart of a plug-in executed in MATLAB by AnyWave.
We will also explain how to create a text file to describe our plug-in to AnyWave

The AnyWave-MATLAB API (Application Programming Interface) which is a set of MATLAB functions, will be described in details, with examples to illustrate their use.

Requirements

MATLAB software must be installed on the computer. AnyWave should be able to detect the installation by itself. However, in some special cases, MATLAB could be installed on a custom path AnyWave may not check. In that case, the user must specify the path to MATLAB in AnyWave by changing options in the Preferences settings.

Linux

On Linux, csh must be installed at /bin/csh, or the MATLAB Engine API used by AnyWave will not function, and AnyWave will report "Failed to connect to MATLAB!" in the corresponding Process log.

sudo apt-get install tsch csh

GCC

MATLAB supports only gcc up to 4.9. Ubuntu 16.04 comes with gcc 5.0 or later.
To make sure the plugins could be executed, you must make sure gcc 4.9 is installed and set as the preferred version to use:

sudo apt-get install gcc-4.9 g++-4.9
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.9 100
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-5.0 50
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-4.9 100
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-5.0 50

Where to start?

The first thing to do is to create the basic structure for a plug-in.
A MATLAB Scripted plug-in is very simple: it is a folder containing at least two files.

Let's begin by creating a folder (called MyPlugin) somewhere on the computer.
This can be done in MATLAB: create a folder and create a new function called main in that folder.
The main function is MANDATORY. It will be the main function AnyWave will call to execute our plugin.

Matlab1.png

As shown in the image above, a MyPlugin folder has been created and a main function was added.

We are now ready to write our first Matlab plug-in!!

Refer to the AnyWave-MATLAB functions section to learn how to program a plug-in that will communicate with AnyWave.

Writing the desc.txt file

We have written the MATLAB code and now all we have to do is to create a descriptive text file to inform AnyWave about our plug-in.
Let's dot it in Maltab:

Matlab2.png

The file must be named desc.txt and may looks like:

name = My Matlab Plugin
description = I am a plug-in written in Matlab
category = Process:MATLAB:My MATLAB Plugin

The syntax is to set keywords and values.

Here we have three keywords (name, description, category).
Two keywords are mandatory : name and description. Other keywords are optional.

keywords

name: The plugin name used by Anywave (here PyExample Plugin).
description: a brief description of what the plugin does.
category (optional): It tells AnyWave where the plug-in will appear in the menus. Here, we decided to make it appear under the Python sub-menu in the Processes main menu.

The category feature is usefull to separate plug-ins that won't really do some calculation but convert data to another format or launch external tools. It could also be useful to classify signal processing algorithms.

Three category keywords are recognized:

  • Process : The plug-in will be set in the Processes menu with a subcategory and a name, for example 'Process:Correlation:Compute correlation'
  • File: The plug-in will be set in the File Menu under the Export sub-menu. Example : 'File:Export to file.'
  • View: The plug-in will be set in the View Menu. Example : 'View:Launch 3D viewer'

If no category is specified, AnyWave will set the plug-in in the Processes menu using the name defined in the file.

Copy the plug-in to the right location

We are ready to add the plug-in to AnyWave.

Copy the folder MyPlugin to your user's AnyWave plugins directory. Remember to place it in the Matlab subfolder. For example on Windows:

Matlab3.png

As you can see MyPlugin is located in the user's AnyWave path for Matlab Scripted plug-ins.

Use the plug-in in AnyWave

Launch AnyWave: the plug-in should be shown as available.

Matlab4.png

AnyWave-MATLAB functions

The table below shows a summary of all MATLAB functions available when writing a MATLAB Scripted plug-in:

Note: All AnyWave MATLAB functions start with 'aw_' to avoid confusions with other MATLAB functions.

Function Short description
aw_getplugininfo Returns useful information about the plugin.
aw_getdata Returns data from AnyWave.
aw_getmarkers Returns markers from AnyWave.
aw_addmarkers Sends markers to AnyWave.

aw_getplugininfo

function [ infos ] = aw_getplugininfo()
%aw_getplugininfo returns information about the input set for the plugin.
%   infos = aw_getplugininfo();
%
%   returns a structure with the following fields:
%   .file
%       Name of the file open in AnyWave.
%   .labels
%       labels of channels set as input for the plugin.
%   .refs
%       references of channels: Empty strings for monopolar channels.
%   .max_sr
%       maximum sampling rate in Hz.
%   .total_duration
%       the total duration in seconds of the data.
%   .temp_dir
%       the path to a temporary directory created by AnyWave for the
%       plugin.
%
end

aw_getdata

function [ channels ] = aw_getdata(cfg )
%aw_getdata request data from AnyWave
%   channels = aw_getdata(cfg) 
%   returns channels' data according to the settings defined in cfg structure.
%
%   cfg may contain the following fields: 
%   
%   cfg.file = 'file.eeg';
%       Specify the data file to use. (optional)
%   cfg.start = 10.;
%       starting position in seconds of requested data.
%       if this field is not specified, AnyWave will return data starting
%       at position 0s.
%   cfg.duration = 20.;
%       duration of requested data in seconds.
%       if this field is not specified, AnyWave will return ALL the
%       available data.
%   NOTE:
%   start and duration are depecrated. Consider using chunks starting with the september 18 release of AnyWave.
%
%   Starting with september 18 release, you can use chunks vector to get one or several part of the signals at once.
%   cfg.chunks = [10 20]; % at least one chunk 
%   a chunk must have at least two values (start and duration).
%   NOTE: if start and duration are used, chunks will be ignore.
%
%   cfg.labels = {'A1', 'A2'};
%       cell array of strings to identify the required channel labels.
%       if no labels are specified, AnyWave will return the current selected
%       channels set as input for the plugin.
%       Optional field.
%   cfg.types = {'EEG'};
%       cell array of strings to specify the types of channels we want.
%       Optional field.*
%
%   NOTE: if labels and types are used, then the channel selection will be made with respect with these two constraints.
%   For example, if the labels cell array contains the A1 label, and the types cell array contains the EEG type, then
%   AnyWave will return the A1 EEG channel if it exists or nothing. If A1 is MEG or SEEG then nothing will be returned.
%
%   cfg.filtering = 'no';
%       specifies that we don't want AnyWave to filter the data.
%   cfg.filtering = 'yes';
%       specifies that we want data to be filtered.
%   Note:
%       if the filtering field is not specified, the data will be filtered by
%       AnyWave using the current filtering options.
%
%   cfg.eeg_lp = 10.;  requires filtering = 'yes'.
%       specifies that we want EEG data with low pass filter of 10Hz.
%   cfg.eeg_hp = 1.;  requires filtering = 'yes'.
%       specifies that we want EEG data with high pass filter of 1Hz.
%   cfg.meg_lp = 10.;  requires filtering = 'yes'.
%       specifies that we want MEG data with low pass filter of 10Hz.
%   cfg.meg_hp = 1.;  requires filtering = 'yes'.
%       specifies that we want MEG data with high pass filter of 1Hz.
%
%   cfg.decimate = 8;
%       specifies that we only take 1 sample of data every 8 samples.
%
%Output:
%   channels is an array of strucrures with the following fields:
%   .name; 
%       a string representing the channel's label.
%   .ref;
%       a string representing the reference channel. Can be empty.
%   .data;
%       a data vector containing the samples.
%   .sr;
%       the sampling rate of data. 
%   .hpf;  
%       the High Pass filter set.
%   .lpf;
%       the Low Pass filter set.
end

Example:

function main
 
% request 10 seconds of data starting at 2.5s
cfg = [];
cfg.start = 2.5;
cfg.duration = 10;
channels = aw_getdata(cfg);
 
% request raw data (with no filtering)
cfg = [];
cfg.start = 2.5;
cfg.duration = 10;
cfg.filtering = 'no';
channels = aw_getdata(cfg);
 
% request data with low pass filter of 25Hz on EEG channels
cfg = [];
cfg.start = 2.5;
cfg.duration = 10;
cfg.filtering = 'yes';
cfg.eeg_lp = 25;
channels = aw_getdata(cfg);
 
%  request data for all EEG channels, no filtering specified
cfg = [];
cfg.start = 0;
cfg.duration = 10; % 10 seconds of data
cfg.types = { 'EEG' };
channels = aw_getdata(cfg);
 
end

aw_addmarkers

function aw_addmarkers(markers)
%aw_addmarkers add new markers to AnyWave
%   aw_addmarkers(markers) 
%
%   markers is an array of structs. Each element is a marker to be added.
%
% See also AwMarker
%   
%
end
 
function [ marker ] = AwMarker( label, position, duration, value, targets)
% defines the structure for a marker
% 
%   marker.label
%       the label for the marker.
%   marker.position
%       the position in seconds from the beginning of data.
%   marker.duration
%       the duration in seconds. Can be 0 if the marker is just a position
%       in time.
%   marker.value
%       the value associated to the marker. -1 indicates that no value is
%       set.
%   marker.targets
%       a cell array containing the targeted channels. Empty if the marker
%       is global.
marker.label = label;                 
marker.position = position;
marker.duration = duration;           
marker.value = value;
marker.targets = targets;             
 
end

Example:

function main
%
% add one marker labeled spike at position 3s.
%
marker.label = 'spike';
marker.position = 3;
 
 
% send the marker to AnyWave
aw_addmarkers(marker);
 
% add two markers with value and targeted channels
markers(1).label = 'spike';
markers(1).position = 3;
markers(1).value = 2;
markers(1).targets = {'A2'};  % this marker is targeting the channel named A2
 
markers(2).label = 'selection';
markers(2).position = 15;
markers(2).duration = 1.5; % this marker is a selection of 1.5s starting at position 15s.
% no targets are specified, the marker is GLOBAL
 
% send the markers to AnyWave
aw_addmarkers(markers);
 
end

aw_getmarkers

function [ markers ] = aw_getmarkers( cfg )
%aw_getmarkers retrieves markers from AnyWave
%   markers = aw_getmarkers(cfg) 
%   markers = aw_getmarkers()
%   returns markers which match the settings defined in cfg structure or
%   all available markers if no cfg structure is passed as paramater.
%
%   cfg may contain the following fields: 
%   
%   cfg.values = [1 10 12];
%       get markers depending on their value. Several values can be
%       specified.
%   cfg.labels = {'Spike1', 'Spike2'};
%       get markers depending on their labels.
%   cfg.channels = {'A1', 'A2'};
%       get markers which target specific channels.
%
% Note: if a marker matches several conditions (values, labels or channels) only one
% instance of the marker is returned.
%
%Output:
%   markers is an array of strucrures with the following fields:
%   .label 
%       a string representing the marker's label.
%   .position
%       the position in seconds from beginning of data.
%   .duration
%       duration in seconds. Can be zero if the marker is just a time position.
%   .value
%       a numerical value associated with the marker. -1 indicates that no value is set. 
%   .channels  
%       a cell array of channel labels. Can be empty if the marker is
%       global to all channels.
%
end

Example:

function main()
% get all the markers currently available in AnyWave
markers = aw_getmarkers();
 
% get markers labeled spike
cfg=[];
cfg.labels = {'spike'};
 
spikes = aw_getmarkers(cfg);
 
% get markers with specified values
cfg=[];
cfg.values = [2 3 4];
 
markers = aw_getmarkers(cfg);
 
% mixing values and labels conditions
cfg=[];
cfg.labels = {'spike', 'selection'};
cfg.values = [2 10];
 
% Note that the returned markers are markers which satisfied one of the conditions (it's a logical OR test).
% That means you can get a marker labeled spike but with a value that is not 2 or 10;
% It's up to you to parse the array of structures afterward.
markers = aw_getmarkers(cfg);
 
% CAUTION
markers can be an empty matrix if no markers are returned by AnyWave
 
end

AwSendMessage(message)

This function sends a text message to AnyWave. The message will appear in the Processes console in AnyWave and also in the plugin's log.

Example:

cfg = [];
cfg.start = 0;
cfg.duration = -1;
channels = aw_getdata(cfg); % get all data
 
for (i=1:numel(channels)
   AwSendMessage(sprintf('Processing channel %d', i));
   % do something
end

AwIsProcessTerminated

This function will return true or false depending on the action of the user about the currently running plug-in.

This function is usefull when processing heavy and long calculations. If the user want to cancel the current processing, it may be suitable to cancel the current calculation running in MATLAB as well.

Example:

cfg = [];
cfg.start = 0;
cfg.duration = -1;
channels = aw_getdata(cfg); % get all data
 
% 
for (i=1:numel(channels)
    if (~AwProcessIsTerminated())
         % do some heavy calculation on a data contained in channel
   end
end