Back to the main page.

Bug 1307 - producing a layout from a bti248 grad structure results in a faulty layout

Reported 2012-02-01 15:14:00 +0100
Modified 2012-02-06 16:44:57 +0100
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P3 normal
Assigned to: Jan-Mathijs Schoffelen
Depends on:
See also:

Jan-Mathijs Schoffelen - 2012-02-01 15:14:50 +0100

because the Height and Width are set to 0. This is because the mininum distance between coils is used to set these values. The reference channels occur in doubles for the bti248 system, i.e. either highpass filtered or not. Solution: exclude the points with 0 distance from this evaluation

Jan-Mathijs Schoffelen - 2012-02-01 15:24:49 +0100

changed the mindist criterion to >100*eps in revision 5211. Still does not work, because the distances are defined in meters, so still yielding too small values

Jan-Mathijs Schoffelen - 2012-02-01 15:33:38 +0100

just use a layout from disk in this case

Boris Reuderink - 2012-02-06 15:53:52 +0100

Is the fix better than the original if it still does not work? And, is the factor 100 necessary in the comparison?

Jan-Mathijs Schoffelen - 2012-02-06 15:58:55 +0100

No, the fix does not help in the case of bti248 data, because I suspect that there is still a discrepancy between the metric units (meters) and the expected scale/range of the numbers). The factor 100 is just included to be safe. This said, it could be that this fix works for (as of yet unknown to me) systems that may have similar issues in the gradiometer definition (e.g. yokogawa or neuromag systems). I think in general that creating a layout based on the grad/elec structure should be a last resort; people should use ready-made (or custom made) layouts for this. Therefore I would not spend any more time on this bug, but leave the fix in, because it shouldn't hurt and in the best case solves problems for other systems.

Boris Reuderink - 2012-02-06 16:44:57 +0100

Ok, I am closing this issue. Thank you for the elaboration.