comp_parad_task.m

%%
% This script is used to generate the task paradigm
% used as ground truth to assess the performances and
% validate the algorithm.

%%
clear all
close all
clc

path = genpath('/Users/iricchi/Documents/MIP_LabImmersion');
addpath(path)

%load HCP_Experimental_Task_paradigms.mat
load HCP_Paradigms_updated.mat

Ns = 50; % Num of subjects
ave = 1;
ave_t = 0;
cor = 1;   % correlation
%% Build the matrix of the paradigm


% Working Memory3 
WM = task_paradigm.WM;
% Move the first 10 zeros at the end
%WM_ = [WM_(end-9:end),WM_(1:end-10)]; 
% repeat it twice because the task repeats per subj
%WM = [WM_'; WM_'];

WM_mat = zeros(length(WM), 9); % 5 because we have 0 = no task
                              % 1= body , 2= faces ...
                               
for i = 1:9
   WM_mat(:,i) = WM == i-1;
end

% Motor task
Mot_ = task_paradigm.Motor;
% Move the first 10 zeros at the end
Mot_ = [Mot_(end-9:end),Mot_(1:end-10)]; 
% repeat it twice because the task repeats per subj
M = [Mot_'; Mot_'];
M_mat = zeros(length(M), 6); % 6 because we have 0 = no task
                               % 1= body , 2= faces ...     
                               
for i = 1:6
   M_mat(:,i) = M == i-1;
end

% Relational Memory
RM = task_paradigm.Relational;
% Move the first 10 zeros at the end
% RM_ = [RM_(end-9:end),RM_(1:end-10)]; 
% repeat it twice because the task repeats per subj
%RM = [RM_'; RM_'];
RM_mat = zeros(length(RM), 3); % 4 because we have 0 = no task
                               % 1= body , 2= faces ...     
                               
for i = 1:3
   RM_mat(:,i) = RM == i-1;
end

% Language

L_ = task_paradigms.Language;
L = [L_';L_'];
L_mat = zeros(length(L),9);

for i = 1:9
    L_mat(:,i) = L == i-1;
end

% Emotion

E_ = task_paradigms.Emotion;
E_ = [E_(end-9:end),E_(1:end-10)];
E = [E_'; E_'];
E_mat = zeros(length(E), 3);

for i =1:3
    E_mat(:,i) = E== i-1;
end

% Social
S_ = task_paradigms.Social;
S_ = [S_(end-9:end),S_(1:end-10)];
S = [S_'; S_'];
S_mat = zeros(length(S), 5);

for i =1:5
    S_mat(:,i) = S== i-1;
end

S_mat = S_mat(:, [1,3,5]);

%% plot 
figure()
imagesc(WM_mat)
names = {'no task'; 'body'; 'faces'; 'places'; 'tools'};
set(gca,'xtick',[1:9],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: working memory')


figure()
imagesc(M_mat)
names = {'no task'; 'left foot'; 'right foot'; 'left hand'; 'right hand'; 'tongue'};
set(gca,'xtick',[1:6],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: Motor')


figure()
imagesc(RM_mat)
names = {'no task'; 'relation'; 'match'};
set(gca,'xtick',[1:3],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: Relational memory')


figure()
imagesc(L_mat)
names = {'no task'; 'math'; 'present math'; 'present story'; 'question math';...
    'question story';'response math'; 'response story'; 'story'};
set(gca,'xtick',[1:9],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: Language')

figure()
imagesc(E_mat)
names = {'no task'; 'fear'; 'neutral'};
set(gca,'xtick',[1:3],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: Emotion')

figure()
imagesc(S_mat)
names = {'no task'; 'mental'; 'random'};
set(gca,'xtick',[1:3],'xticklabel',names)
ylabel('x_m')
title('Task Paradigm: Social')

%% Plot gammas


load DataFromServer/Ls_WM_GL.mat
load DataFromServer/mus_WM_GL.mat   % to visulize brain_vis
load DataFromServer/gamma_hats_WM_GL.mat

k = 4; %num of clusters
figure()
imagesc(gamma_hats)
xticks(linspace(1,k,k))
xlabel('K')
ylabel('x_m')
title('gamma values')


%% BIG matrix with repetition of paradigm Ns times, Ns = num of subjects

prompt = 'Which task to use: ';
task = input(prompt, 's');

switch task
    case 'M'
        T = M_mat;
    case 'R'
        T = RM_mat;
    case 'W'
        T = WM_mat;
    case 'L'
        T = L_mat;
    otherwise
        disp('Error! Choose between M, R, W')
end

BigMat = zeros(size(T,1), size(T,2));
for i = 1:Ns
    BigMat((i-1)*length(T)+1:i*length(T),:) = T;
end

% plot
figure()
imagesc(BigMat)
title('50 subjects paradigm')

%% Avarage out the gammas
if (ave==1)
    k = size(gamma_hats,2);
    val = length(gamma_hats)/Ns;
    gammas = zeros(val,k);

    ind = zeros(Ns,val);
    ind(1,1) = 1;
    for i=2:Ns
        ind(i,1) = ind(i-1,1)+val;
    end

    for i = 2:val
       ind(:,i) = ind(:,i-1)+1 ;
    end

    for i = 1:val

        gammas(i,:) = mean(gamma_hats(ind(:,i),:));

    end

    figure()
    imagesc(gammas)
    xticks(linspace(1,k,k))
    xlabel('K')
    ylabel('x_m')
    title('Average gammas')
end

%% Plot correlation btw mean gammas and paradigm
if(cor ==1)
    [C, p] = corr(gammas,T);
    Cp = C;
    Cp(p>0.05) = 0;
    figure()
    imagesc(Cp)
    title('Correlation mean gammas and paradigm')
end