topics - functions - index - search

filtering and segmentation

Command: PrePro, FilterNetFile, FiltNeuroscanFiles, TransNetGeoHist, NeuroScanToEgis, TransMsiContSess, CalcAutoEditMat, CalcAutoEditMEGMat

Filtering and segmentation can be done manually, using scripts like 'Batch.m' that call 'FilterNetFile', or, or by using 'PrePro' user interface (under construction).

Filtering and segmentation using PrePro (under construction)
PrePro, when started, is disabled, and waits for you to open a data file, or to load a batchfile containing paths of datafiles. Datafiles must have the same format (EGI, Neuroscan or BTI) and the same sampling rate and the same number of channels. After succesfully loading a file, the number of  channels and the sampling rate are displayed in the upper right corner.

Filters: After loading a file, you can activate and configure highpass, lowpass and stopband filters by enabling their radio-buttons. Initially you have to configure every filter, that you enable.

Segmentation: By entering numbers in the PreTrig and PostTrig boxes, you can define how many points will be taken before each event and how many after.
Events: The default behaviour in PrePro segmets your data for every kind of event it encounters. If you do not like this, you can activate the 'use only'-box and enter numbers, corresponding the trigger values that you wish to use. All others will be ignored.

Artifact detection: PrePro automatically chooses the entire extracted intervall, to calculate statistical parameters, that afterwards can be used for artifact detection. If you wish to use only a certain part of the intervall for this purpose, you can define separate PreTrig- and PostTrig-points for artifact detection. Generally a longer intervall is more likely to contain artifacts, thus leading to less good trials.
Amplifier calibration: This options so far is only available for EGI data. Neuroscan data is analyzed presuming all GAINS to be 1 and all ZEROS to be 0. For EGI-data, you can one specific set of GAIN and ZEROs, that will be used for all files, or have PrePro search for a GAIN and ZERO for every data file in the batch separately. This is done simply by using the filename, so be sure, to have consistent filenaming.
MiscellaneousThe options  'change  RAW to TAW' and 'remove additional AR-channels'  again only apply to EGI data, and normally can be left active. The option 'skip step if file exists' lets you avoid repeating one step if it already has been performed and procuded a result file.  'One step for all' vs. 'All steps for one' refers only to batch  processing, and decides wether PrePro will perform one step (for instance the segmentation) for all files in the batch before going one with the next step. Otherwise, all steps will be performed for one file, and then all steps for the next etc. The latter is generelly preferable for long batches, that you do not want to crash half way through. Thus, if a problem occurs on one file, PrePro will still get the other files right.

All settings including the filematrix can be saved and loaded using the menu \File\Default File.

Manual filtering and segmentation
Generally, filtering a raw-data-file saves the filtered data in a new file, adding filter-dependant characters to the filename. A lowpass 40Hz filter, for instance, transform the file '01.run1.taw' to 01.run1.fl40.taw'.