Back to the main page.

Bug 1918 - clarify how data transformations are affecting the sens structure

Reported 2013-01-07 15:21:00 +0100
Modified 2013-06-02 10:04:26 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P3 normal
Assigned to: Diego Lozano Soldevilla
Depends on:
See also:

Robert Oostenveld - 2013-01-07 15:21:19 +0100

FT has not always been consistent in dealing with how transformations of the data are represented in transformations of the sensor array. Examples of data transformations are - megplanar - combineplanar - componentanalysis - rejectcompunent - denoise_synthetic A test script would be useful that does each of them and then checks that the transformation on the data is matched with a corresponding transformation (*) on the data.grad or data.elec. *) or not, if that transformation is not desired. The thing is: right now it is not clear when a sens-transformation is desired and when not. We could even make a FAQ that lists the transformations and explains why the sens is updated and how (e.g. "after combineplanar, the channel-level data is not a linear mixture of the source level data, thereby invalidating the forward model...")

Johanna - 2013-01-07 15:27:26 +0100

see related bug 1443 which also says to see bug 1249