CORRMAP Plug-in for EEGLAB The CORRMAP structure is constituted by six main fields.
>> CORRMAP =
template: [1x1 struct]
datasets: [1x1 struct]
input: [1x1 struct]
clust: [1x1 struct]
corr: [1x1 struct]
output: [1x1 struct]
Here is a detailed explanation of each field.
- CORRMAP.template
This field contains three sub-fields ('setname', 'index', 'ic') where information about the template IC is stored. In this example the template was the IC 5 from dataset 1, which was named: 01_010906.
>> CORRMAP.template
ans =
setname: '01_010906'
index: 1
ic: 5
- CORRMAP.datasets
This field contains three sub-fields ('setnames', 'index', 'ics') where information about the datasets evaluated is stored. In this example, 16 datasets where evaluated and each dataset contains 68 ICs. The names for each dataset are stored in a cell array.
>>CORRMAP.datasets
ans =
setnames: {1x16 cell}
index: [1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16]
ics: [68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68]
- CORRMAP.input
This field contains six sub-fields ('corr_th', 'ics_sel', 'plots', 'title', 'cl_name') corresponding to all input parameters of the corrmap() function, excepting the 'dataset index' and 'IC index' that are stored in CORRMAP.template. In this example, the input parameters show the default values. Note that as CORRMAP.input.corr_th=’auto’ (correlation threshold), only the second output plot is displayed (CORRMAP.input.plots=’2nd’).
>> CORRMAP.input
ans =
chanlocs: ''
corr_th: 'auto'
ics_sel: 2
plots: '2nd'
title: ''
clname: ''
badcomps: 'no'
Disclaimer: When CORRMAP is run from the EEGLAB GUI the plot parameter is not displayed because it is dependent on the correlation threshold option. When running CORRMAP from the command line it is also possible to change the plot option to ‘none’, so the output plots are not displayed.
>> CORRMAP=corrmap(STUDY,ALLEEG,1,5,’pl’,’none’);
- CORRMAP.clust
This field contains five sub-fields ('best_th', 'ics', 'sets', 'mean_corr', 'similarity') corresponding to output information about the cluster created.
>> CORRMAP.clust
ans =
best_th: 9.1000e-001
ics: [6 7]
sets: [1x1 struct]
mean_corr: [9.4916e-001 9.7135e-001]
similarity: 9.9956e-001
In this example, as the correlation threshold was set to 'auto', a best threshold is displayed (CORRMAP.clust.best_th). The sub-field CORRMAP.clust.ics shows the number of ICs clustered in each iteration. The sub-field CORRMAP.clust.sets is divided into three other fields ('number', 'more_oneIC', 'absent') as shown below.
>> CORRMAP.clust.sets
ans =
number: [6 7]
more_oneIC: [0 0]
absent: {[1 3 5 6 10 11 12 13 14 15] [3 5 6 10 11 12 13 14 15]}
The field CORRMAP.clust.sets.more_oneIC shows the total number of datasets that contribute with two ICs to the cluster in each iteration. In this example, in both iterations, there are no datasets contributing with two ICs. The indices of the datasets not contributing to the cluster are listed in the field CORRMAP.clust.sets.absent.
The last two sub-fields in CORRMAP.clust store information about the mean correlation values for each iteration (CORRMAP.clust.mean_corr) and the similarity index (CORRMAP.clust.similarity).
- CORRMAP.corr
This field contains three sub-fields ('abs_values', 'sets', 'ics') corresponding to the correlation calculation. The sub-field CORRMAP.corr.abs_values contains the absolute correlation values between the template and each selected IC. The values are sorted in descending order. Note that in this example, as there are 16 datasets being evaluated and as the maximum number of ICs selected was set to 2, the output vectors will have 32 (16×2) correlation values. The other two sub-fields show the dataset indices and IC indices sorted according to the absolute correlation values.
>>CORRMAP.corr
ans =
abs_values: {[32x1 double] [32x1 double]}
sets: {[32x1 double] [32x1 double]}
ics: {[32x1 double] [32x1 double]}
- CORRMAP.output
This field contains four sub-fields ('chanlocs', 'average_plot', 'sets', 'ICs'). The sub-field CORRMAP.output.chanlocs contains channel location information from the file used as input. The sub-field CORRMAP.output.average_plot contains the average plots calculated in both iterations. By using the first two sub-fields one can re-plot the average plots without having to run CORRMAP again. The other two sub-fields show the dataset indices and IC indices for the components clustered in each iteration.
>>CORRMAP.output
ans =
chanlocs: [1x68 struct]
average_plot: {[68x1 double] [68x1 double]}
sets: {[6x1 double] [7x1 double]}
ics: {[6x1 double] [7x1 double]}
Disclaimer: Please remember that when you run CORRMAP using the EEGLAB GUI the CORRMAP structure IS NOT saved in the MATLAB working space. Please also remember that when using CORRMAP from the command line the parameter for the correlation threshold (‘th’) must be a string. See example below.
>> CORRMAP=corrmap(STUDY,ALLEEG,1,5,’th’,’0.85’,’ics’,1);
This page was last updated May 20, 2011.