[FieldTrip] Functionality change - neighbourselection will be obliged by the user

["Jörn M. Horschig"]

Dear FieldTrip community,

Up until now, many users and functions relied on a predefined way to 
define proximal or neighbouring sensor, namely using a distance 
measurement. We changed the functionality of ft_neighbourselection and 
related functions, which will be explained in the following.

*1 - The user is obliged to define and verify channel neighbours*
Driven by some thoughts and current discussion going on in the FieldTrip 
mailinglist, we decided to change neighbourselection. Within the next 
days, all functions which require information about proximal or 
neighbouring sensors will be required to have a cfg.neighbour field that 
you as the user have to specify explicitly. This may be obvious for some 
functions, such as cluster statistics on channel level, but not so much 
for other functions such as planar gradient computation or channelrepair.

In order to ease the way of getting into the whole neighbourselection 
issue, we wrote 3 FAQs:
http://fieldtrip.fcdonders.nl/faq/how_can_i_define_neighbouring_sensors
http://fieldtrip.fcdonders.nl/faq/how_does_ft_neighbourselection_work
http://fieldtrip.fcdonders.nl/faq/how_can_i_define_my_own_neighbourhood_template

Please note that you can use ft_neighbourplot to see the current 
neighbourselection and verify whether this seems like a good choice. 
Although we try to come up with extremely smart ways to compute 
neighbours of your data, selection of neighbours can still be highly 
subjective. Therefore, we also decided to do the following:

*2 - Neighbours can be defined using different methods*

Since the near beginning of FieldTrip, neighbours were selected based on 
a 'distance' approach, meaning that nearby sensors were defined as 
neighbours. This has shown to be valid for some systems given a 
predefined distance (neighbdist) of 4cm, whereas other systems needed a 
different neighbdist.
In order to reduce subjectivity and arbitrariness of the maximal 
distance of a sensor to be called 'neighbour', we implemented a 
'triangulation' approach. For more information, see the corresponding 
FAQ above.
However, since systems around the world should be similar in terms of 
sensor position, we decided to implement a 'template'-based approach, 
which loads a predefined neighbour-template from file. It is up to you 
to decide what approach you are most comfortable with. The standard 
behaviour of ft_neighbourselection still is the old 'distance' approach.

*3 - Call for neighbour templates
*
Now, dear community, you may become active and share your 
neighbour-templates. Here at the Donders, we are primarily working with 
the CTF275 system and have thus quite some expertise and experience. For 
this system, you can already find a template in the newer FieldTrip 
versions. We will continue improving the neighbourdefinition of this 
template and try to be as objective as possible. Furthermore, we will 
start creating templates for other systems as well and try to optimize 
these. However, when it comes to other systems, also including EEG 
systems, you might be more experienced. Therefore, we would like to ask 
whether you want to share your expertise and send the neighbourstructure 
that you are using on your system to us (== me). I will then try to 
incorporate your subjective selection with ours, probably after some 
discussion. If you are willing to support us in that respect, please do 
*not* send a mail to the mailing-list but to my personal mail address, 
as I will be organizing the templates: jm.horschig at donders.ru.nl Thanks 
in advance for your cooperation.

Best regards on behalf of the developer team,
Jörn

-- 
Jörn M. Horschig
PhD Student
Donders Institute for Brain, Cognition and Behaviour
Centre for Cognitive Neuroimaging
Radboud University Nijmegen
Neuronal Oscillations Group

P.O. Box 9101
NL-6500 HB Nijmegen
The Netherlands

Contact:
E-Mail: jm.horschig at donders.ru.nl
Tel:    +31-(0)24-36-68493
Web: http://www.ru.nl/donders

Visiting address:
Trigon, room 2.30
Kapittelweg 29
NL-6525 EN Nijmegen
The Netherlands

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110708/6a5dae8b/attachment.html>