Source Code for Module mvpa.misc.io.meg

  1  # emacs: -*- mode: python; py-indent-offset: 4; indent-tabs-mode: nil -*- 
  2  # vi: set ft=python sts=4 ts=4 sw=4 et: 
  3  ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ## 
  4  # 
  5  #   See COPYING file distributed along with the PyMVPA package for the 
  6  #   copyright and license terms. 
  7  # 
  8  ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ## 
  9  """IO helper for MEG datasets.""" 
 10   
 11  __docformat__ = 'restructuredtext' 
 12   
 13  import numpy as N 
 14   
 15  from mvpa.base import externals 
 16   
 17 -class TuebingenMEG(object): 
 18      """Reader for MEG data from line-based textfile format. 
 19   
 20      This class reads segmented MEG data from a textfile, which is created by 
 21      converting the proprietary binary output files of a MEG device in 
 22      Tuebingen (Germany) with an unkown tool. 
 23   
 24      The file format is line-based, i.e. all timepoints for all samples/trials 
 25      are written in a single line. Each line is prefixed with an identifier 
 26      (using a colon as the delimiter between identifier and data). Two lines 
 27      have a special purpose. The first 'Sample Number' is a list of timepoint 
 28      ids, similar to `range(ntimepoints)` for each sample/trial (all 
 29      concatenated into one line. The second 'Time' contains the timing 
 30      information for each timepoint (relative to stimulus onset), again for all 
 31      trials concatenated into a single line. 
 32   
 33      All other lines contain various information (channels) recorded during 
 34      the experiment. The meaning of some channels is unknown. Known ones are: 
 35   
 36        M*: MEG channels 
 37        EEG*: EEG channels 
 38        ADC*: Analog to digital converter output 
 39   
 40      Dataset properties are available from various class attributes. The `data` 
 41      member provides all data from all channels (except for 'Sample Number' and 
 42      'Time') in a NumPy array (nsamples x nchannels x ntimepoints). 
 43   
 44      The reader supports uncompressed as well as gzipped input files (or other 
 45      file-like objects). 
 46      """ 
 47   
 48 -    def __init__(self, source): 
 49          """Reader MEG data from texfiles or file-like objects. 
 50   
 51          :Parameters: 
 52            source: str | file-like 
 53              Strings are assumed to be filenames (with `.gz` suffix 
 54              compressed), while all other object types are treated as file-like 
 55              objects. 
 56          """ 
 57          self.ntimepoints = None 
 58          self.timepoints = None 
 59          self.nsamples = None 
 60          self.channelids = [] 
 61          self.data = [] 
 62          self.samplingrate = None 
 63   
 64          # open textfiles 
 65          if isinstance(source, str): 
 66              if source.endswith('.gz'): 
 67                  externals.exists('gzip', raiseException=True) 
 68                  import gzip 
 69                  source = gzip.open(source, 'r') 
 70              else: 
 71                  source = open(source, 'r') 
 72   
 73          # read file 
 74          for line in source: 
 75              # split ID 
 76              colon = line.find(':') 
 77   
 78              # ignore lines without id 
 79              if colon == -1: 
 80                  continue 
 81   
 82              id = line[:colon] 
 83              data = line[colon+1:].strip() 
 84              if id == 'Sample Number': 
 85                  timepoints = N.fromstring(data, dtype=int, sep='\t') 
 86                  # one more as it starts with zero 
 87                  self.ntimepoints = int(timepoints.max()) + 1 
 88                  self.nsamples = int(len(timepoints) / self.ntimepoints) 
 89              elif id == 'Time': 
 90                  self.timepoints = N.fromstring(data, 
 91                                                 dtype=float, 
 92                                                 count=self.ntimepoints, 
 93                                                 sep='\t') 
 94                  self.samplingrate = self.ntimepoints \ 
 95                      / (self.timepoints[-1] - self.timepoints[0]) 
 96              else: 
 97                  # load data 
 98                  self.data.append( 
 99                      N.fromstring(data, dtype=float, sep='\t').reshape( 
100                          self.nsamples, self.ntimepoints)) 
101                  # store id 
102                  self.channelids.append(id) 
103   
104          # reshape data from (channels x samples x timepoints) to 
105          # (samples x chanels x timepoints) 
106          self.data = N.swapaxes(N.array(self.data), 0, 1) 
107   
108   
109 -    def __str__(self): 
110          """Give a short summary. 
111          """ 
112          return '<TuebingenMEG: %i samples, %i timepoints, %i channels>' \ 
113                    % (self.nsamples, self.ntimepoints, len(self.channelids)) 
114