method to compute fitted event time course
FORMAT [tc dt] = event_fitted(D, e_spec, dur)
D - design
e_spec - 2 by N array specifying events to combine
with row 1 giving session number
and row 2 giving event number in session
This may in due course become an object type
dur - duration in seconds of event to estimate for
Returns
tc - fitted event time course, averaged over events
dt - time units (seconds per row in X)
$Id$
0001 function [tc, dt] = event_fitted(D, e_spec, dur) 0002 % method to compute fitted event time course 0003 % FORMAT [tc dt] = event_fitted(D, e_spec, dur) 0004 % 0005 % D - design 0006 % e_spec - 2 by N array specifying events to combine 0007 % with row 1 giving session number 0008 % and row 2 giving event number in session 0009 % This may in due course become an object type 0010 % dur - duration in seconds of event to estimate for 0011 % 0012 % Returns 0013 % tc - fitted event time course, averaged over events 0014 % dt - time units (seconds per row in X) 0015 % 0016 % $Id$ 0017 0018 if nargin < 2 0019 error('Need event specification'); 0020 end 0021 if nargin < 3 0022 dur = 0; 0023 end 0024 0025 if ~is_fmri(D) | isempty(e_spec) 0026 tc = []; 0027 return 0028 end 0029 if ~is_mars_estimated(D) 0030 error('Need a MarsBaR estimated design'); 0031 end 0032 if size(e_spec, 1) == 1, e_spec = e_spec'; end 0033 0034 e_s_l = size(e_spec, 2); 0035 SPM = des_struct(D); 0036 betas = SPM.betas; 0037 tc = zeros(1, size(betas, 2)); 0038 for e_i = 1:e_s_l 0039 es = e_spec(:, e_i); 0040 ss = es(1); 0041 [X dt]= event_regressor(D, es, dur); 0042 B = betas(event_cols(D, es), :); 0043 Yh = X*B; 0044 0045 % Sum over events 0046 sz = size(Yh, 1); 0047 szo = size(tc, 1); 0048 if sz > szo 0049 tc(end+1:sz, :) = 0; 0050 end 0051 tc(1:sz,:) = tc(1:sz,:) + Yh; 0052 0053 end 0054 tc = tc / e_s_l;