Back to the main page.
Bug 800 - Create and optimize templates for neighbourselection
|2011-07-07 11:33:00 +0200
|2014-03-12 12:21:36 +0100
|Jörn M. Horschig
Jörn M. Horschig - 2011-07-07 11:33:28 +0200+ write a FAQ and update the mailinglist
Jörn M. Horschig - 2011-07-07 17:03:25 +0200last piece o' work for today http://fieldtrip.fcdonders.nl/development/neighbourtemplates
Jörn M. Horschig - 2011-07-11 13:40:55 +0200added to http://fieldtrip.fcdonders.nl/faq added to neighbourselection in template mode a remark to contact me in case of error updated neighbourselection help text
Jörn M. Horschig - 2011-10-20 15:44:58 +0200created templates for all data types from test data and for all layouts that don't suffer from the sens.tra bug. all templates are not optimized yet, but are reasonable neighbourstructures
Jörn M. Horschig - 2012-04-20 09:37:13 +0200note for myself to easen the work on this task: - add a cfg.edit (or enableedit) option to ft_neighbourplot if in editmode - clicking on two channels right after each other makes or removes (toggle) these channels as neighbours - add a button for undoing the last step
Jörn M. Horschig - 2012-08-22 16:30:49 +0200ft_neighbourplot now supports cfg.enableedit, which allows the user to add or remove edges between nodes, i.e. mark or unmark sensors as being neighbours. Primarily, this one is for me to make my life working on this bug easier, but it will also help people to tinker their own neighbour structure svn ci ft_neighbourplot.m -m "enhancement-#800-neighbourplot supports an edit mode which makes it possible to flexibly add and remove edges between neighbouring sensors" Sending ft_neighbourplot.m Transmitting file data . Committed revision 6397.
Jörn M. Horschig - 2012-08-23 13:49:56 +0200tedious work, half of the templates are adjusted now, the other half will follow after (or while at^^) biomag. gonna update svn all at once when the work is finished
Jörn M. Horschig - 2012-09-05 15:01:18 +0200- beautified + enhanced all templates based on the 2D layouts, for anyone to pick up on this in a few years from now: rationale is that squares+diagonals should be preferred over triangles. Occasionally, when it makes sense i.e. sensors are closeby and would have few connections, a pentagram like connection is allowed. - next step: use 3D MEG testdata to verify whether the templates make sense in 3D space as well svn ci *_neighb.mat -m "enhancement-#800- neighbour templates improved" Sending biosemi16_neighb.mat Sending biosemi32_neighb.mat Sending biosemi64_neighb.mat Sending bti148_neighb.mat Sending bti248_neighb.mat Sending ctf151_neighb.mat Sending ctf275_neighb.mat Sending easycap128ch-avg_neighb.mat Sending easycap32ch-avg_neighb.mat Sending easycap64ch-avg_neighb.mat Sending easycapM11_neighb.mat Sending easycapM14_neighb.mat Sending easycapM15_neighb.mat Sending easycapM1_neighb.mat Sending ecog256_neighb.mat Sending ecog256bipolar_neighb.mat Sending eeg1010_neighb.mat Sending elec1005_neighb.mat Sending elec1010_neighb.mat Sending elec1020_neighb.mat Sending language29ch-avg_neighb.mat Sending neuromag306cmb_neighb.mat Sending neuromag306mag_neighb.mat Sending neuromag306planar_neighb.mat Transmitting file data ........................ Committed revision 6432.
Jörn M. Horschig - 2013-03-20 16:28:41 +0100we still lack ctf275 data with 275 channels, one way around this would be template grad structure, see bug 800