Searchlight using a data measure

Using cosmo_searchlight, run cross-validation with nearest neighbor classifier

For CoSMoMVPA's copyright information and license terms, #
see the COPYING file distributed with CoSMoMVPA. #

Define data
Set measure
Define neighborhood
Run the searchlight
Make a histogram of classification accuracies
Plot a map

Define data

config = cosmo_config();
data_path = fullfile(config.tutorial_data_path, 'ak6', 's01');

targets = repmat(1:6, 1, 10);
chunks = floor(((1:60) - 1) / 6) + 1;

ds = cosmo_fmri_dataset(fullfile(data_path, 'glm_T_stats_perrun.nii'), ...
                        'mask', fullfile(data_path, 'brain_mask.nii'), ...
                        'targets', targets, 'chunks', chunks);

Set measure

Use the cosmo_cross_validation_measure and set its parameters (classifier and partitions) in a measure_args struct. >@@>

measure = @cosmo_crossvalidation_measure;
measure_args = struct();
measure_args.classifier = @cosmo_classify_lda;
measure_args.partitions = cosmo_oddeven_partitioner(ds);
% <@@<

Define neighborhood

radius = 3; % 3 voxels
% define a neighborhood using cosmo_spherical_neighborhood
% >@@>
nbrhood = cosmo_spherical_neighborhood(ds, 'radius', radius);
% <@@<

% show a histogram of the number of voxels in each searchlight
% >@@>
count = cellfun(@numel, nbrhood.neighbors);
hist(count, 100);
% <@@<

+00:00:00 [####################] -00:00:00  mean size 111.5

Run the searchlight

hint: use cosmo_searchlight with the measure, args and nbrhood

% >@@>
results = cosmo_searchlight(ds, nbrhood, measure, measure_args);
% <@@<

% the following command would store the results to disk:
% >> cosmo_map2fmri(results, [data_path 'measure_searchlight.nii']);

+00:00:23 [####################] -00:00:00

Make a histogram of classification accuracies

figure();
hist(results.samples, 47);

Plot a map

figure();
cosmo_plot_slices(results);