method to return FIR design matrix columns for session
FORMAT Xn = event_x_fir(D, e_spec, bin_length, bin_no, opts)
D - design object
e_spec - event specification for single event
[session no; event no]
bin_length - bin length in seconds
bin_no - number of bins for FIR
opts - structure, containing fields with options
'single' - if field present, gives single FIR
This option removes any duration information, and
returns a simple per onset FIR model, where ones in the
design matrix corresponds to 1 event at the given
offset. See event_fitted_fir.m for more details.
Returns
Xn - columns in design matrix for FIR model
Note that we have a problem, in that the assumed start bin is not saved
in the SPM99 design format, so we have to hope it has not changed from
the current defaults.
$Id$
0001 function Xn = event_x_fir(D, e_spec, bin_length, bin_no, opts) 0002 % method to return FIR design matrix columns for session 0003 % FORMAT Xn = event_x_fir(D, e_spec, bin_length, bin_no, opts) 0004 % 0005 % D - design object 0006 % e_spec - event specification for single event 0007 % [session no; event no] 0008 % bin_length - bin length in seconds 0009 % bin_no - number of bins for FIR 0010 % opts - structure, containing fields with options 0011 % 'single' - if field present, gives single FIR 0012 % This option removes any duration information, and 0013 % returns a simple per onset FIR model, where ones in the 0014 % design matrix corresponds to 1 event at the given 0015 % offset. See event_fitted_fir.m for more details. 0016 % 0017 % Returns 0018 % Xn - columns in design matrix for FIR model 0019 % 0020 % Note that we have a problem, in that the assumed start bin is not saved 0021 % in the SPM99 design format, so we have to hope it has not changed from 0022 % the current defaults. 0023 % 0024 % $Id$ 0025 0026 % global parameters 0027 global fMRI_T; 0028 global fMRI_T0; 0029 if isempty(fMRI_T), fMRI_T = 16; end; 0030 if isempty(fMRI_T0), fMRI_T0 = 1; end; 0031 0032 if nargin < 2 0033 error('Need event specfication'); 0034 end 0035 if nargin < 3 0036 bin_length = []; 0037 end 0038 if nargin < 4 0039 bin_no = []; 0040 end 0041 if nargin < 5 0042 opts = []; 0043 end 0044 0045 s = e_spec(1); 0046 e = e_spec(2); 0047 if isempty(bin_length) 0048 bin_length = tr(D); 0049 end 0050 if isempty(bin_no) 0051 bin_no = round(25/bin_length); 0052 end 0053 0054 SPM = des_struct(D); 0055 Sess = SPM.Sess{s}; 0056 dt = SPM.xX.dt; 0057 0058 % Check dt against fMRI_T, warn if it differs 0059 recorded_fMRI_T = round(SPM.xX.RT / dt); 0060 if recorded_fMRI_T ~= fMRI_T & verbose(D) 0061 warning(sprintf([... 0062 'fMRI_T (%d) does not match recorded dt, using recorded dt (%d).\n' ... 0063 'The original fMRI_T0 has not been recorded, assuming %d.'],... 0064 fMRI_T, recorded_fMRI_T, fMRI_T0)); 0065 end 0066 T = recorded_fMRI_T; 0067 bf = kron(eye(bin_no),ones(round(bin_length/dt),1)); 0068 bf = pr_spm_orth(bf); 0069 0070 % Reset columns to 1 after orthogonalization 0071 BF{1} = bf / bf(1); 0072 0073 k = length(Sess.row); 0074 0075 if isfield(opts, 'single') 0076 [onsets durations] = event_onsets(D, e_spec); 0077 ons = sparse(k*T,1); 0078 for p = 1:length(onsets) 0079 q = round(onsets(p)*T + 1); 0080 ons(q) = 1; 0081 end 0082 SF{1} = ons(1:(k*T)); 0083 if verbose(D) & any(diff(durations)) 0084 warning(['Apparently there were different event durations; ' ... 0085 'single FIR model likely to be invalid']); 0086 end 0087 else 0088 SF{1} = Sess.sf{e}(:,1); 0089 end 0090 0091 Xn = pr_spm_volterra(SF,BF,{'FIR'},1); 0092 0093 % Resample design matrix {X} at acquisition times 0094 %----------------------------------------------- 0095 Xn = Xn([0:k-1]*T + fMRI_T0,:);