Reads physiological data measured by the BioPac system, epochs data
ri = read_biopac(fname,params)
The BioPac system has been used by the Janata Lab at the 3T
scanner at the UC Davis Imaging Research Center, to collect physiological
measures while participants were being scanned. This system, though,
could conceivably be used to collect data outside of the scanner.
Therefore, this script does not assume that physiological data it
analyzes was collected in an NMR environment, and it does not assume that
a TR is available for epoching purposes, but it does allow for a TR to be
used for epoching, and it also allows for a more general presentation
logfile-based epoching.
REQUIRES
- reading data -
fname - path to single file to be processed
params.channels(:) - struct array containing information about the
channels to be extracted and returned
.name - channel name, that will be converted to a fieldname in ri
i.e. if a channel name is "cardiac", then that data will be found
in ri.signal.cardiac in the output data
.comment_text - name given to the data stream in AcqKnowledge/BioPac,
as reflected in .hdr.per_chan_data.comment_text in the .acq file
NOTE: special channel name 'trigger' treated as scanner pulses,
assumed to be TTL pulses time-locked to volume acquisition onset
params.signal_bounds - 2-length vector containing start and end times
in seconds to initially bound the raw signal. this can be used in
the case that one physio file contains data from multiple runs.
Since data here are handled on a run-by-run basis, this can be used
to limit the initial data that is extracted to the samples
pertinent for the given run. a value of 0 in the first position
will be replaced with a 1, while a value of 0 in the second
position will be replaced with the final sample in the data set
- epoching -
params.bound_by_trigger - if this is true, and if there is a channel in
params.channels(:) with name 'trigger', and if this data is not empty
or a flat line, triggers will be used to define the epoch for this run
... the epoch will be first trigger through last trigger+TR*srate ...
NOTE: if params.bound_by_trigger is set, and if a trigger channel
exists, params.link2pres will be ignored
NOTE: this requires params.TR ... if no params.TR is specified, it
will be estimated as median(diff(trigger_onsets_in_ms))/1000
params.baseline - the amount of time, in seconds, to allow at the
beginning of the signal for a baseline calculation.
params.TR - length (in seconds) between repetitions, for fMRI analyses.
see also: params.bound_by_trigger
params.link2pres - structure with settings to link up to a presentation
response file ... MAKE SURE that you specify filt criteria that will
return a set of items that then have a 1 to 1 correlation with event
onsets in a channel within the acq file. If the return from your
filtering criteria exactly matches the channel you've specified to
link to, then all signals from the acq file will be aligned with the
presentation data, and data falling before run_rel_time = 0 will be
thrown out
NOTE: if params.bound_by_trigger is set, and if a trigger channel
exists, link2pres will be ignored
NOTE: link2pres ASSUMES that the run begins AFTER physio data
collection begins. Even if presentation is started before physio data
collection, it will adjust all times in run_rel_time to what it
thinks is the beginning of the run, and all signals processed by
link2pres rely on this.
.presfname - path and filename of the presentation file to link to
.filt - filtering criteria applied to presfname data via
ensemble_filter ... you may want to use this struct to filter by
run, if you are calling read_biopac (or proc_physio_data) from
within a loop over runs, for instance
.channelname - ri.signal.(.channelname) will be compared to
filtered results from presfname. If a one-to-one match is found
between presfname filtered data and .channelname data, then epochs
will be defined by the beginning of the presentation run_rel_time,
and the last event returned by the presfname filtered data
.diff_tol - a diff will be done for all onsets given for presentation
data as well as link signal data, and then a diff will be done
between those diffs. If the diffs don't line up in length, or if
there is any diff between diffs greater than .diff_tol, link2pres
will no be applied
.bound_by_last_pulse - if set to 1, and if EVENT_TYPE={'Pulse'},
EVENT_CODE = {'255'} filtering criteria returns any rows, all data
after sample = ((lastPulseTime+TR*1000)/stime) will be dropped. If
no params.TR is specified, TR=median(diff(pulse_onsets_in_ms))/1000
.estimate_trigger - if a 'trigger' channel can not be found, but
presentation data is provided and contains events in the form of
EVENT_TYPE = {'Pulse'} && EVENT_CODE = 255, and if link2pres is
specified, after temporal linking is achieved, the pulse events
will be extracted and converted to triggers. Missing triggers will
be interpolated, either using the specified TR, or using a TR
estimated from the median pulse period.
OUTPUT
ri - 'runinfo' struct, in a form similar to that of read_keithley.m and
read_mate.m, containing fields for data streams, events, and onsets,
also containing a .meta field, with .meta.srate (sampling rate),
.meta.stime (sampling time, or # ms per sample), and
.meta.baseline_samps (# of baseline samples). This differs from that
of read_keithley and read_mate in that the actual signal fields are
found in ri.signal, rather than within ri
NOTE: ri.trigger values are sample indices, NOT times in ms or sec
FB 2008.10.12
FB 2009.06.24 - added baseline calculation0001 function ri = read_biopac(fname,params) 0002 0003 % Reads physiological data measured by the BioPac system, epochs data 0004 % 0005 % ri = read_biopac(fname,params) 0006 % 0007 % The BioPac system has been used by the Janata Lab at the 3T 0008 % scanner at the UC Davis Imaging Research Center, to collect physiological 0009 % measures while participants were being scanned. This system, though, 0010 % could conceivably be used to collect data outside of the scanner. 0011 % Therefore, this script does not assume that physiological data it 0012 % analyzes was collected in an NMR environment, and it does not assume that 0013 % a TR is available for epoching purposes, but it does allow for a TR to be 0014 % used for epoching, and it also allows for a more general presentation 0015 % logfile-based epoching. 0016 % 0017 % REQUIRES 0018 % - reading data - 0019 % fname - path to single file to be processed 0020 % params.channels(:) - struct array containing information about the 0021 % channels to be extracted and returned 0022 % .name - channel name, that will be converted to a fieldname in ri 0023 % i.e. if a channel name is "cardiac", then that data will be found 0024 % in ri.signal.cardiac in the output data 0025 % .comment_text - name given to the data stream in AcqKnowledge/BioPac, 0026 % as reflected in .hdr.per_chan_data.comment_text in the .acq file 0027 % NOTE: special channel name 'trigger' treated as scanner pulses, 0028 % assumed to be TTL pulses time-locked to volume acquisition onset 0029 % params.signal_bounds - 2-length vector containing start and end times 0030 % in seconds to initially bound the raw signal. this can be used in 0031 % the case that one physio file contains data from multiple runs. 0032 % Since data here are handled on a run-by-run basis, this can be used 0033 % to limit the initial data that is extracted to the samples 0034 % pertinent for the given run. a value of 0 in the first position 0035 % will be replaced with a 1, while a value of 0 in the second 0036 % position will be replaced with the final sample in the data set 0037 % 0038 % - epoching - 0039 % params.bound_by_trigger - if this is true, and if there is a channel in 0040 % params.channels(:) with name 'trigger', and if this data is not empty 0041 % or a flat line, triggers will be used to define the epoch for this run 0042 % ... the epoch will be first trigger through last trigger+TR*srate ... 0043 % NOTE: if params.bound_by_trigger is set, and if a trigger channel 0044 % exists, params.link2pres will be ignored 0045 % NOTE: this requires params.TR ... if no params.TR is specified, it 0046 % will be estimated as median(diff(trigger_onsets_in_ms))/1000 0047 % params.baseline - the amount of time, in seconds, to allow at the 0048 % beginning of the signal for a baseline calculation. 0049 % params.TR - length (in seconds) between repetitions, for fMRI analyses. 0050 % see also: params.bound_by_trigger 0051 % params.link2pres - structure with settings to link up to a presentation 0052 % response file ... MAKE SURE that you specify filt criteria that will 0053 % return a set of items that then have a 1 to 1 correlation with event 0054 % onsets in a channel within the acq file. If the return from your 0055 % filtering criteria exactly matches the channel you've specified to 0056 % link to, then all signals from the acq file will be aligned with the 0057 % presentation data, and data falling before run_rel_time = 0 will be 0058 % thrown out 0059 % NOTE: if params.bound_by_trigger is set, and if a trigger channel 0060 % exists, link2pres will be ignored 0061 % NOTE: link2pres ASSUMES that the run begins AFTER physio data 0062 % collection begins. Even if presentation is started before physio data 0063 % collection, it will adjust all times in run_rel_time to what it 0064 % thinks is the beginning of the run, and all signals processed by 0065 % link2pres rely on this. 0066 % .presfname - path and filename of the presentation file to link to 0067 % .filt - filtering criteria applied to presfname data via 0068 % ensemble_filter ... you may want to use this struct to filter by 0069 % run, if you are calling read_biopac (or proc_physio_data) from 0070 % within a loop over runs, for instance 0071 % .channelname - ri.signal.(.channelname) will be compared to 0072 % filtered results from presfname. If a one-to-one match is found 0073 % between presfname filtered data and .channelname data, then epochs 0074 % will be defined by the beginning of the presentation run_rel_time, 0075 % and the last event returned by the presfname filtered data 0076 % .diff_tol - a diff will be done for all onsets given for presentation 0077 % data as well as link signal data, and then a diff will be done 0078 % between those diffs. If the diffs don't line up in length, or if 0079 % there is any diff between diffs greater than .diff_tol, link2pres 0080 % will no be applied 0081 % .bound_by_last_pulse - if set to 1, and if EVENT_TYPE={'Pulse'}, 0082 % EVENT_CODE = {'255'} filtering criteria returns any rows, all data 0083 % after sample = ((lastPulseTime+TR*1000)/stime) will be dropped. If 0084 % no params.TR is specified, TR=median(diff(pulse_onsets_in_ms))/1000 0085 % .estimate_trigger - if a 'trigger' channel can not be found, but 0086 % presentation data is provided and contains events in the form of 0087 % EVENT_TYPE = {'Pulse'} && EVENT_CODE = 255, and if link2pres is 0088 % specified, after temporal linking is achieved, the pulse events 0089 % will be extracted and converted to triggers. Missing triggers will 0090 % be interpolated, either using the specified TR, or using a TR 0091 % estimated from the median pulse period. 0092 % 0093 % OUTPUT 0094 % ri - 'runinfo' struct, in a form similar to that of read_keithley.m and 0095 % read_mate.m, containing fields for data streams, events, and onsets, 0096 % also containing a .meta field, with .meta.srate (sampling rate), 0097 % .meta.stime (sampling time, or # ms per sample), and 0098 % .meta.baseline_samps (# of baseline samples). This differs from that 0099 % of read_keithley and read_mate in that the actual signal fields are 0100 % found in ri.signal, rather than within ri 0101 % NOTE: ri.trigger values are sample indices, NOT times in ms or sec 0102 % 0103 % FB 2008.10.12 0104 % FB 2009.06.24 - added baseline calculation 0105 0106 ri = struct('signal',struct(),'meta',struct()); 0107 global r 0108 0109 r = init_results_struct; 0110 r.type = 'physio_biopac'; % Identify the type of this reporting instance 0111 r.report_on_fly = 1; 0112 0113 if ~isempty(fname) && exist(fname,'file') 0114 % read data 0115 pdata = load_acq(fname); 0116 chans = {pdata.hdr.per_chan_data(:).comment_text}; 0117 else 0118 % data can't be found, skip this file 0119 msg = sprintf('couldn''t find biopac file %s, SKIPPING\n',fname); 0120 r = update_report(r,msg); 0121 return 0122 end 0123 0124 % get sampling rate 0125 stime = pdata.hdr.graph.sample_time; % msec per sample 0126 srate = 1000/stime; % 1000 msec / stime = sampling rate 0127 ri.meta.srate = srate; 0128 ri.meta.stime = stime; 0129 0130 % get TR 0131 if isfield(params,'TR') 0132 TR = params.TR; 0133 else 0134 TR = 0; 0135 warning('no TR found in the params struct\n'); 0136 end 0137 0138 % signal bounds? 0139 if isfield(params,'signal_bounds') && length(params.signal_bounds) == 2 0140 % convert from params.signal_bounds in secs to sig_start/end in samples 0141 sig_start = params.signal_bounds(1)*srate; 0142 sig_end = params.signal_bounds(2)*srate; 0143 if ~isnumeric(sig_start) || sig_start < 1 0144 sig_start = 1; 0145 end 0146 if ~isnumeric(sig_end) || sig_end < 1 || sig_end > size(pdata.data,1) 0147 sig_end = size(pdata.data,1); 0148 end 0149 else 0150 sig_start = 1; 0151 sig_end = size(pdata.data,1); 0152 end 0153 0154 % find channels, get data 0155 ctxts = {pdata.hdr.per_chan_data.comment_text}; 0156 if isfield(params,'channels') 0157 nc = length(params.channels); 0158 for ic=1:nc 0159 ctxt = params.channels(ic).comment_text; 0160 name = params.channels(ic).name; 0161 cidx = strmatch(ctxt,ctxts); 0162 if ~isempty(cidx) 0163 ri.signal.(name) = pdata.data(sig_start:sig_end,cidx); 0164 end 0165 end 0166 end 0167 0168 % find trigger onsets, if they exist 0169 if isfield(ri.signal,'trigger') 0170 % onsets and offsets are in samples, NOT time in msec or sec 0171 [ri.trigger.onsets,ri.trigger.offsets] = ... 0172 find_thresh_cross(ri.signal.trigger,params); 0173 end 0174 0175 % extract run-length epochs 0176 % epoch_start and epoch_end are expressed in samples, NOT in time 0177 if isfield(ri,'trigger') && isfield(ri.trigger,'onsets') ... 0178 && ~isempty(ri.trigger.onsets) && isfield(params,'bound_by_trigger') ... 0179 && params.bound_by_trigger 0180 % check for TR 0181 if ~exist('TR','var') || ~TR 0182 warning('estimating TR from trigger onsets'); 0183 TR = median(diff(ri.trigger.onsets*stime))/1000; 0184 warning('TR estimated to be %ds',TR); 0185 end 0186 0187 % bound by triggers 0188 epoch_start = ri.trigger.onsets(1); 0189 epoch_end = ri.trigger.onsets(end)+TR*srate; % add one TR at the end 0190 0191 elseif isfield(params,'link2pres') ... 0192 && isfield(params.link2pres,'channelname')... 0193 && ischar(params.link2pres.channelname) ... 0194 && isfield(ri.signal,params.link2pres.channelname) 0195 0196 % extract signal 0197 linksig = ri.signal.(params.link2pres.channelname); 0198 [lsons,lsoffs] = find_thresh_cross(linksig,params); 0199 lsdiff_ms = diff(lsons)*stime; 0200 nlsdiff = length(lsdiff_ms); 0201 0202 % link to presentation file, bound by presentation data 0203 pfname = params.link2pres.presfname; 0204 presdata = load(pfname); 0205 pcol = set_var_col_const(presdata.vars); 0206 0207 if isfield(params.link2pres,'filt') 0208 pfilt = params.link2pres.filt; 0209 linkdata = ensemble_filter(presdata,pfilt); 0210 else 0211 linkdata = presdata; 0212 end 0213 pdiff_ms = diff(linkdata.data{pcol.RUN_REL_TIME}); 0214 npdiff = length(pdiff_ms); 0215 0216 if isfield(params.link2pres,'diff_tol') 0217 diff_tol = params.link2pres.diff_tol; 0218 else 0219 diff_tol = 100; % 100ms tolerance 0220 end 0221 nlsd = length(lsdiff_ms); 0222 npd = length(pdiff_ms); 0223 0224 % the first two conditionals within this if/else train are a hack to 0225 % account for an additional leading or trailing signal in either linksig 0226 % or presdata, that doesn't appear in the other ... essentially, if 0227 % linksig has items 1 2 3 and pdata has 2 3 or 1 2 (i.e. pdata is missing 0228 % a leading or trailing element, but otherwise is intact), or vice versa 0229 % (linksig is missing an element), we can still link the two measurements 0230 % together 0231 0232 % FIXME: right now, d3 = d1>d2, and the sum of # of indices where one is 0233 % greater than the other is used to figure out where 'less difference' 0234 % exists, but this is not really a metric of 'less difference', rather 0235 % 'indices with less difference' ... might another formulation work 0236 % better? how about d4 = d1-d2, then compare abs(sum(d4(d3))) to 0237 % abs(sum(d4(~de))) ... ? if problems arise, I will try this strategy 0238 if (nlsd - npd) == 1 0239 % one more event in the link signal than in the presentation file 0240 0241 % assume the first pres signal is missing 0242 d1 = abs(diff([lsdiff_ms'; [lsdiff_ms(1); pdiff_ms]'])); 0243 0244 % assume the last pres signal is missing 0245 d2 = abs(diff([lsdiff_ms'; [pdiff_ms; lsdiff_ms(end)]'])); 0246 0247 % greater diff for first signal compared to last signal assumption? 0248 d3 = d1 > d2; 0249 0250 if (sum(d3) < nlsd/2) 0251 % smaller difference between link sig and pres times when assuming 0252 % the first pres time is missing 0253 % in other words, linksig(2) = prestime(1) 0254 if ~any(d1 > diff_tol) 0255 % no differences greater than diff_tol, safe to assume the first 0256 % pres time is missing 0257 pbegin = linkdata.data{pcol.RUN_REL_TIME}(1); % 1st pres time 0258 % check to see that the first epoch time is not before the 0259 % beginning of the run as presentation has calculated it 0260 if lsons(2)*stime - lsons(1)*stime > pbegin 0261 msg = sprintf(['possible error calculating the beginning of '... 0262 'the epoch. There are more link signal events %d than '... 0263 'presentation linking events %d, and it appears that the '... 0264 'first presentation event is missing, but if this is the '... 0265 'case, the first link signal event is before the start '... 0266 'of the run, when comparing event times to presentation '... 0267 'run-relative times'],nlsd,npd); 0268 r = update_report(r,msg); 0269 ri = struct(); 0270 return 0271 else 0272 % start of the epoch is the onset of the second link signal 0273 % event, minus the time of the first (but really, the second) 0274 % presentation linking event, converted to link signal samples 0275 epoch_start = round((lsons(2)*stime - pbegin)/stime); 0276 end 0277 else 0278 % differences greater than diff_tol, likely a middle time that is 0279 % missing 0280 msg = sprintf('missing presentation link, can not compute'); 0281 r = update_report(r,msg); 0282 ri = struct(); 0283 return 0284 end 0285 elseif (sum(d3) > nlsd/2) 0286 % smaller difference between link sig and pres times when assuming 0287 % the last pres time is missing 0288 % in other words, linksig(1) = prestime(1) 0289 if ~any(d2 > diff_tol) 0290 % no differences greater than diff_tol, safe to assume the last 0291 % pres time is missing 0292 pbegin = linkdata.data{pcol.RUN_REL_TIME}(1); % 1st pres time 0293 % start of the epoch is the onset of the first link signal event, 0294 % minus the time of the first presentation linking event, converted 0295 % to link signal samples 0296 epoch_start = round((lsons(1)*stime - pbegin)/stime); 0297 else 0298 % differences greater than diff_tol, likely a middle time that is 0299 % missing 0300 msg = sprintf('missing presentation link, can not compute'); 0301 r = update_report(r,msg); 0302 ri = struct(); 0303 return 0304 end 0305 else 0306 msg = sprintf('missing presentation link, can not compute'); 0307 r = update_report(r,msg); 0308 ri = struct(); 0309 return 0310 end % if (sum(d3) < nlsd/2 0311 elseif (npd - nlsd) == 1 0312 % one more event in the presentation file than in the link signal 0313 0314 % assume the first link signal is missing 0315 d1 = abs(diff([pdiff_ms'; [pdiff_ms(1); lsdiff_ms]'])); 0316 0317 % assume the last link signal is missing 0318 d2 = abs(diff([pdiff_ms'; [lsdiff_ms; pdiff_ms(end)]'])); 0319 0320 % greater diff for first signal compared to last signal assumption? 0321 d3 = d1 > d2; 0322 0323 if (sum(d3) < npd/2) 0324 % smaller difference between link sig and pres times when assuming 0325 % the first link signal is missing 0326 % in other words, prestime(2) == linksig(1) 0327 if ~any(d1 > diff_tol) 0328 % no differences greater than diff_tol, safe to assume the first 0329 % link signal is missing 0330 pbegin = linkdata.data{pcol.RUN_REL_TIME}(2); % 2nd pres time 0331 % check to see that the beginning of physio data collection is 0332 % before the beginning of the run as presentation has logged it 0333 if lsons(1)*stime < pbegin 0334 msg = sprintf(['possible error calculating the beginning of '... 0335 'the epoch. There are more presentation linking events %d'... 0336 ' than link signal events %d, and it appears that the '... 0337 'first link signal event is missing, but if this is the '... 0338 'case, the first presentation linkin event is before the '... 0339 'start of physio data collection, when comparing physio '... 0340 'data to presentation run-relative times'],npd,nlsd); 0341 r = update_report(r,msg); 0342 ri = struct(); 0343 return 0344 else 0345 % start of the epoch is the onset of the first link signal 0346 % event, minus the time of the second presentation linking event, 0347 % converted to link signal samples 0348 epoch_start = round((lsons(1)*stime - pbegin)/stime); 0349 end 0350 else 0351 % differences greater than diff_tol, likely a middle time that is 0352 % missing 0353 msg = sprintf('missing link signal, can not compute'); 0354 r = update_report(r,msg); 0355 ri = struct(); 0356 return 0357 end 0358 elseif (sum(d3) > npd/2) 0359 % smaller difference between link sig and pres times when assuming 0360 % the last link signal is missing 0361 % in other words, linksig(1) = prestime(1) 0362 if ~any(d2 > diff_tol) 0363 % no differences greater than diff_tol, safe to assume the last 0364 % link signal is missing 0365 pbegin = linkdata.data{pcol.RUN_REL_TIME}(1); % 1st pres time 0366 % start of the epoch is the onset of the first link signal event, 0367 % minus the time of the first presentation linking event, converted 0368 % to link signal samples 0369 epoch_start = round((lsons(1)*stime - pbegin)/stime); 0370 else 0371 % differences greater than diff_tol, likely a middle time that is 0372 % missing 0373 msg = sprintf('missing link signal, can not compute'); 0374 r = update_report(r,msg); 0375 ri = struct(); 0376 return 0377 end 0378 else 0379 msg = sprintf('missing link signal, can not compute'); 0380 r = update_report(r,msg); 0381 ri = struct(); 0382 return 0383 end 0384 elseif npd ~= nlsd 0385 msg = sprintf(['link signal size (%d) is not equal in length to '... 0386 'presentation data size (%d), therefore link2pres skipped\n'],... 0387 nlsdiff,npdiff); 0388 r = update_report(r,msg); 0389 ri = struct(); 0390 return 0391 elseif any(diff([lsdiff_ms'; pdiff_ms']) > diff_tol) 0392 msg = sprintf(['more than %d ms diff btwn link signal onset diffs '... 0393 'and presentation data onset diffs, therefore link2pres skipped\n'],... 0394 diff_tol); 0395 r = update_report(r,msg); 0396 ri = struct(); 0397 return 0398 else 0399 pbegin = linkdata.data{pcol.RUN_REL_TIME}(1); 0400 epoch_start = round((lsons(1)*stime - pbegin)/stime); 0401 end 0402 0403 epoch_end = length(linksig); 0404 0405 if isfield(params.link2pres,'bound_by_last_pulse') || ... 0406 isfield(params.link2pres,'estimate_trigger') 0407 lpfilt.include.all.EVENT_TYPE={'Pulse'}; 0408 lpfilt.include.all.EVENT_CODE={'255'}; 0409 lpfilt.include.all.RUN=params.run; 0410 lpdata = ensemble_filter(presdata,lpfilt); 0411 rrtimes = lpdata.data{pcol.RUN_REL_TIME}; 0412 0413 if ~isempty(lpdata.data{1}) 0414 if ~exist('TR','var') || ~TR 0415 warning('estimating TR from pulse data'); 0416 TR = median(diff(rrtimes))/1000; 0417 warning('TR estimated to be %ds',TR); 0418 end 0419 0420 if isfield(params.link2pres,'estimate_trigger') 0421 cpp = check_pulse_periods(rrtimes); 0422 pptl = find(cpp.data.pulse_period_too_long_mask); 0423 for ipp=1:length(pptl) 0424 intpp = rrtimes(pptl(ipp)):(TR*1000):rrtimes(pptl(ipp)+1); 0425 rrtimes = union(rrtimes,intpp); 0426 end 0427 samples = rrtimes/stime; 0428 ri.trigger.onsets = samples; 0429 ri.trigger.offsets = samples+1; 0430 end 0431 0432 if isfield(params.link2pres,'bound_by_last_pulse') 0433 % the end of the epoch = run_rel_time/stime + epoch_start 0434 pend = ceil((lpdata.data{pcol.RUN_REL_TIME}(end)+TR*1000)/stime); 0435 epoch_end = pend + epoch_start; 0436 end 0437 else 0438 warning('no pulses found, no trigger estimation or bounding'); 0439 end 0440 end 0441 0442 else 0443 warning('could not generate run-length epochs'); 0444 return 0445 end 0446 0447 % add baseline period? 0448 if isfield(params,'baseline') 0449 blstart = epoch_start - params.baseline*srate; 0450 if blstart < 1, blstart = 1; end 0451 ri.meta.baseline_samps = epoch_start - blstart; 0452 epoch_start = blstart; 0453 else 0454 ri.meta.baseline_samps = 0; 0455 end 0456 0457 % apply bounds 0458 flds = fieldnames(ri.signal); 0459 nflds = length(flds); 0460 for ifld = 1:nflds 0461 fld = flds{ifld}; 0462 nsamp = length(ri.signal.(fld)); 0463 if epoch_start > nsamp 0464 continue 0465 end 0466 if epoch_end > nsamp 0467 epoch_end = nsamp; 0468 end 0469 ri.signal.(fld) = ri.signal.(fld)(epoch_start:epoch_end); 0470 end 0471 0472 % adjust triggers? 0473 if isfield(ri,'trigger') && isfield(ri.trigger,'onsets') ... 0474 && ~isempty(ri.trigger.onsets) && isfield(params,'bound_by_trigger') ... 0475 && params.bound_by_trigger 0476 ri.trigger.onsets = ri.trigger.onsets + ri.meta.baseline_samps; 0477 ri.trigger.offsets = ri.trigger.offsets + ri.meta.baseline_samps; 0478 end