[Erppcatoolkit-support] ERP PCA Toolkit 2.17 release

[Joseph D. <jd...@ma...>]

I just posted version 2.17.  Aside from the usual bug fixes and improved error handling, the major feature addition is jack-knife PCA analysis.  Whenever a PCA is run, the Toolkit will run a series of additional PCAs, one for each subject, where each subject in turn is left out of the sample (thus a twenty subject sample will have twenty PCAs run each with nineteen subjects).  The factor loadings from the exercise are then stored for later use.  This does have the effect of causing the PCAs to take a little longer than before (the command line window will provide a running count of which jack-knife sample it is working on).  Currently, the jack-knife results are only used in the Topos function.  If you right-click on a factor and then choose "jack-knife" it will use the results of the jack-knife to compute the source analyses for all the jack-knife samples and then display their locations.  The resulting spread gives a sense of how stable the PCA solution was to variations in the subject sample.  The jack-knife provides more useful results than doing a separate one for each subject, which would largely just result in a random cloud as the noise level would generally be too high for both PCA and source analysis.  The jack-knife source solutions are also used to compute a t-test for which hemispheric dipole is stronger (using the Miller et al, 1998 jack-knife t-test).

There are also a couple user interface changes of note:  1) the file suffix for EP file format files is now .ept rather than .mat, to help avoid confusion.  Existing files can just be renamed from .mat to .ept.  Note that both OS X and Windows can hide the suffix so you may need to make sure it is being displayed so that you can change it, otherwise you could end up with a file that looks like name.ept but is actually name.ept.mat.  2) The Toolkit now puts an asterisk in front of the name of data in the active set that have been changed to remind you to save them if you want to keep them.

1) Fixed crash when rereferencing and there are no implicit channels.
2) Fixed crash when postprocessing (bug introduced in 2.16).
3) Fixed not keeping FID channels in implicit channel info for fixed channel order file formats (e.g., EGIS, text).
4) Fixed electrode information not matching the data for second file onwards when reading files in single cell file mode.
5) For single cell file mode, no longer need to access ced file for every file.
6) Fixed not successfully merging average files together when reading in data using single cell file mode.
7) Fixed cell labels and sub labels not being applied when reading EP file formats, as when using single cell file mode.
8) Fixed baseline control on Samples subpane failing to change baseline value.
9) When importing data via single cell file mode, only adds trial names for single trial data files.
10) View function no longer crashes when average file erroneously has non-empty trial names field.
11) Fixed crash when reading file formats with fixed-order channels when ced file has wrong number of channels.
12) Fixed files sometimes not being recognized as being selected when names are in uppercase.
13) Fixed not ignoring extra tab at end of line of text files, resulting in "not-a-number" errors.
14) Fixed crash in Topos function when trying to display 3D plot or dipole source using .ced file generated from a .elp file.
15) Fixed average files not being generated with .dataName field, which could result in crashes down the line.
16) Added dipole analysis of jack-knifed PCA results to the Topos function of the View Pane.
17) Added name of first session file to the name of the artifact correction log file if just one file or the number of files if more.
18) In listings of files, puts a star in front of files that have unsaved changes.
19) PCA output no longer adds grand average to subjects dimension if the data type is continuous to avoid resulting crashes.
20) Fixed crash when retaining only one factor for an Infomax rotation.
21) Changed the suffix of EP files to ".ept".
22) Fixed crash when loading EP or .set file with name of original ced file in addition to eloc information.
23) Fixed losing the electrode coordinate information (eloc and and ced) when loading in a .study file.
24) Fixed crash when windowing adds regional channel and there is no electrode coordinate information (eloc).
25) Fixed crash when conducting robust ANOVA and there are no spec columns in the ANOVA data file.

Thanks to Noriko Tanigawa for reporting a series of bugs and Matt Mollison for making suggestions on how to improve the Toolkit.

Cheers!

Joe

--------------------------------------------------------------------------------

Joseph Dien,
Senior Research Scientist
University of Maryland 
7005 52nd Avenue
College Park , MD 20742-0025

E-mail: jd...@ma...
Phone: 301-226-8848
Fax: 301-226-8811
http://homepage.mac.com/jdien07/