-
Notifications
You must be signed in to change notification settings - Fork 21
/
matlab_tfce_correlation.m
122 lines (106 loc) · 3.16 KB
/
matlab_tfce_correlation.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
function [varargout] = matlab_tfce_correlation(imgs,covariate,tails,nperm,H,E,C,dh)
% MATLAB_TFCE_CORRELATION computes TFCE corrected p-values for
% individual difference correlation between a covariate and brain activity.
% Note that actual inference is performed on Fisher r-to-z transforms of
% the Pearson correlation coefficents for linearity.
%
% Arguments:
% imgs -- a 4D (x,y,z,subject) matrix of images.
% covariate -- a vector of length = number of subjects containing values
% to be correlated with brain activity
% tails -- 1 or 2 tailed test
% nperm -- number of permutations to perform. More permutations yield
% more precise correct p-values.
% -- img the 3D image to be transformed
% -- H height exponent
% -- E extent exponent
% -- C connectivity
% -- ndh step number for cluster formation
%
% Output:
% If tails == 1:
% pcorr -- wholebrain map of corrected p-values
% If tails == 2:
% pcorr_pos -- corrected p-values for positive effects
% pcorr_neg -- corrected p-values for negative effects
% calculate matrix size
bsize = size(imgs);
nsub = bsize(4);
covariate = covariate(:);
bsize = bsize(1:3);
% set tranform function
if tails == 1
transform = @matlab_tfce_transform;
else
transform = @matlab_tfce_transform_twotailed;
end
% calculate implicit mask
sumimg = sum(imgs,4);
implicitmask = ~isnan(sumimg) & sumimg~=0;
nvox = sum(implicitmask(:));
% extract occupied voxels for permutation test
occimgs = NaN(nsub,nvox);
for s = 1:nsub
curimg = imgs(:,:,:,s);
occimgs(s,:) = curimg(implicitmask);
end
% calculate true correlation image
truestat = corr(occimgs,covariate);
truestat =.5*log((1+truestat)./(1-truestat));
trueimg=NaN(bsize);
trueimg(implicitmask) = truestat;
trueimg = transform(trueimg,H,E,C,dh);
tfcestat = trueimg(implicitmask);
cvals = tfcestat;
if tails == 2
cvals = abs(tfcestat);
end
% initialize progress indicator
parfor_progress(nperm);
global parworkers
% cycle through permutations
exceedances = zeros(nvox,1);
parfor(p = 1:nperm,parworkers)
% permute covariates
rsel = randperm(nsub);
rcov = covariate(rsel);
% calculate permutation correlations
rstats = corr(occimgs,rcov);
rstats =.5*log((1+rstats)./(1-rstats));
rbrain = zeros(bsize);
rbrain(implicitmask) = rstats;
rbrain = transform(rbrain,H,E,C,dh);
rstats = rbrain(implicitmask);
if tails == 2
rstats = abs(rstats);
end
% compare maxima to t-values and increment as appropriate
curexceeds = max(rstats) >= cvals;
exceedances = exceedances + curexceeds;
% update progress indicator (only does so 1 in 5 to minimize overhead)
if ~randi([0 4]);
parfor_progress;
end
end
% create corrected p-value image
corrected = exceedances./nperm;
pcorr = ones(bsize);
pcorr(implicitmask) = corrected;
% split into positive and negative effects (if needed)
if tails == 2
btruestat = NaN(bsize);
btruestat(implicitmask) = truestat;
pos = btruestat>0;
pcorr_pos = pcorr;
pcorr_pos(~pos) = 1;
pcorr_neg = pcorr;
pcorr_neg(pos) = 1;
end
% assign output to varargout
if tails == 1
varargout{1} = pcorr;
else
varargout{1} = pcorr_pos;
varargout{2} = pcorr_neg;
end
end