% Concept:
% --------------------------------------------------------------------
%   This walkthrough explains how to do a "pattern of interest" (POI) analysis
%   using an a priori multivariate pattern. It generalizes the "region of
%   interest" (ROI) approach. ROI analyses usually involve averaging signal
%   over the voxels within an ROI. The POI approach applies a weighted average,
%   with the weights specified by the pattern. The "pattern response" is a
%   single number that reflects the magnitude of activity in the pattern.
%   If the pattern is a mask of 1's and 0's, the POI response is equivalent
%   to averaging, up to the scaling of the values (which are usually
%   assumed to be arbitrary, but the same across participants/test
%   datasets).
%
% Required toolboxes:
% --------------------------------------------------------------------
%   Matlab signal processing toolbox
%   Matlab statistics/machine learning toolboxes
%   SPM12
% Two CANlab toolboxes from https://github.com/canlab
%   https://github.com/canlab/CanlabCore
%   https://github.com/canlab/Neuroimaging_Pattern_Masks
%
% Add SPM12 and all CANlab toolboxes with subfolders to the Matlab path
%

% Load some data images. These are N = 30 subjects from Wager et al. 2008,
% Neuron. Each image is a contrast between regulating and looking at
% aversive images.
test_data_obj = load_image_set('emotionreg');

% Load a set of patterns that we want to apply
% These are Partial Least Squares (PLS)-based signatures from Kragel et al.
% 2018 Nature Neurosci.
% Three patterns relate to Pain, Cognitive Control, and Negative Emotion,
% respectively.

[obj, names] = load_image_set('pain_cog_emo');
bpls_wholebrain = get_wh_image(obj, [8 16 24]);
names_wholebrain = names([8 16 24]);

% Incidentally, we could also extract patterns for local subregions,
% e.g., only within the dorsal anterior cingulate (dACC) or other regions
% of interest. The code below would extract relevant images, but we will
% not do this here.
%   bpls_subregions = get_wh_image(obj, [1:6 9:14 17:22]);
%   names_subregions = names([1:6 9:14 17:22]);

%  Make plots
% Yellow: positive associations. Blue: Negative associations.  Plot shows mean +- std. error for each pattern of interest

create_figure('Kragel Pain-Cog-Emo maps', 1, 3);

% Notes:
% An important thing to do is to make sure that the images you're comparing
% have been resampled to the same space and voxel size, with voxels that line up
% across both image sets.  The function below handles this automatically,
% as do other related functions in the toolboxes, e.g., apply_mask.m.
% They use the object method resample_space.m
%
% Another consideration is how to handle missing data values and values of
% 0, which are often considered missing data in fMRI images. This is handled here by
% canlab_pattern_similarity, which treats 0s in the data images as missing, and excludes them,
% but does not exclude 0s in pattern masks, which may be valid values. The
% impact of this on the calculated similarity metrics varies across
% similarity metrics.
%
% In the plot below, yellow areas are positive correlations, blue are
% negative:
stats = image_similarity_plot(test_data_obj, 'average', 'mapset', bpls_wholebrain, 'networknames', names_wholebrain, 'nofigure');
axis image

% Now let's take the values and make a bar plot instead:
subplot(1, 3, 2)

barplot_columns(stats.r', 'nofigure', 'colors', {[1 .9 0] [.2 .2 1] [1 .2 .2]}, 'names', names_wholebrain)
set(gca, 'FontSize', 14)
ylabel('Pattern similarity (r)');
title('Similarity (r) with patterns')

% Now let's recalculate similarity using a different metrric, cosine
% similarity.  We'll have to resample the data image first:

test_data_obj = resample_space(test_data_obj, bpls_wholebrain);

% Then calculate the pattern similarity:

clear csim
for i = 1:3

    csim(:, i) = canlab_pattern_similarity(test_data_obj.dat, bpls_wholebrain.dat(:, i), 'cosine_similarity');

end

% Now we can plot this

subplot(1, 3, 3)

barplot_columns(csim, 'nofigure', 'colors', {[1 .9 0] [.2 .2 1] [1 .2 .2]}, 'names', names_wholebrain)
set(gca, 'FontSize', 14)
ylabel('Pattern similarity (cosine sim)');
title('Pattern response (cosine similarity)')

Loaded images:
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
/home/lukie/CanlabCore/CanlabCore/Sample_datasets/Wager_et_al_2008_Neuron_EmotionReg/Wager_2008_emo_reg_vs_look_neg_contrast_images.nii
Loaded images:
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pMCC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_aMCC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pACC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_sgACC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_vmPFC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_dMFC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_MFC_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_Wholebrain_Pain.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pMCC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_aMCC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pACC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_sgACC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_vmPFC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_dMFC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_MFC_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_Wholebrain_Cognitive_Control.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pMCC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_aMCC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_pACC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_sgACC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_vmPFC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_dMFC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_MFC_Negative_Emotion.nii
/home/lukie/Neuroimaging_Pattern_Masks/Multivariate_signature_patterns/2018_Kragel_MFC_Generalizability/bPLS_Wholebrain_Negative_Emotion.nii
Table of correlations Group:1
--------------------------------------
T-test on Fisher's r to Z transformed point-biserial correlations
 	R_avg	T	P	sig	
Pain Wholebrain	-0.0313	-3.3659	0.0022	1.0000	
Cog Wholebrain	-0.0034	-0.4795	0.6352	0.0000	
Emo Wholebrain	0.0326	3.6973	0.0009	1.0000	
 
Col   1: Pain Wholebrain	Col   2: Cog Wholebrain	Col   3: Emo Wholebrain	
---------------------------------------------
Tests of column means against zero
---------------------------------------------
          Name           Mean_Value    Std_Error       T            P         Cohens_d 
    _________________    __________    _________    ________    __________    _________

    'Pain Wholebrain'     -0.031344    0.0093007     -3.3701     0.0021409     -0.61529
    'Cog Wholebrain'     -0.0034143    0.0071259    -0.47913       0.63544    -0.087478
    'Emo Wholebrain'       0.032647    0.0088234         3.7    0.00089741      0.67553


ans = 

  struct with fields:

         fig_han: [1×1 struct]
        axis_han: [1×1 Axes]
        bar_han1: [1×1 Bar]
         bar_han: {[1×1 Bar]  [1×1 Bar]  [1×1 Bar]}
    errorbar_han: {[1×1 ErrorBar]  [1×1 ErrorBar]  [1×1 ErrorBar]}
      point_han1: {30×3 cell}
       point_han: {30×3 cell}
    star_handles: [3.0005 4.0005 5.0005]

Col   1: Pain Wholebrain	Col   2: Cog Wholebrain	Col   3: Emo Wholebrain	
---------------------------------------------
Tests of column means against zero
---------------------------------------------
          Name           Mean_Value    Std_Error       T            P        Cohens_d 
    _________________    __________    _________    ________    _________    _________

    'Pain Wholebrain'       -0.0277    0.0088243      -3.139    0.0038755     -0.57311
    'Cog Wholebrain'     -0.0024414    0.0064192    -0.38032      0.70648    -0.069437
    'Emo Wholebrain'       0.028817    0.0081249      3.5468    0.0013476      0.64756


ans = 

  struct with fields:

         fig_han: [1×1 struct]
        axis_han: [1×1 Axes]
        bar_han1: [1×1 Bar]
         bar_han: {[1×1 Bar]  [1×1 Bar]  [1×1 Bar]}
    errorbar_han: {[1×1 ErrorBar]  [1×1 ErrorBar]  [1×1 ErrorBar]}
      point_han1: {30×3 cell}
       point_han: {30×3 cell}
    star_handles: [6.0005 7.0005 8.0005]

Published with MATLAB® R2017b