Back to the main page.
Bug 1983 - ft_prepare_sourcemodel should support cfg.grid.unit
| Status | CLOSED WONTFIX |
| Reported | 2013-02-13 14:53:00 +0100 |
| Modified | 2016-11-29 09:04:39 +0100 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P3 normal |
| Assigned to: | Jan-Mathijs Schoffelen |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: | http://bugzilla.fcdonders.nl/show_bug.cgi?id=2755 |
Jörn M. Horschig - 2013-02-13 14:53:04 +0100
In http://fieldtrip.fcdonders.nl/example/create_single-subject_grids_in_individual_head_space_that_are_all_aligned_in_mni_space it is implicitly expected that ft_prepare_sourcemodel assumes 'cm' for cfg.grid, independently of the input cfg.hdm. In the example, the units for cfg.hdm are 'mm', but the example script states cfg.grid.xgrid/ygrid/zgrid in 'cm' (as can be deduced from the comment in the example script. There used to be a workaround before, which multiplied these numbers by 10 after ft_prepare_sourcemodel, which is a pretty nasty hack. Nowadays, given that we became old and wise and knowledgable, we know that the correct way to go about with this is either (1) Specify cfg.grid in the same units as cfg.hdm or (2) Set the unit-field of cfg.grid correctly, i.e. set cfg.grid.unit = 'cm' However, (2) does not work, but I see no reason why it should not work. Work to be done: 1) Change example script 2) Add support for cfg.unit to ft_prepare_sourcemodel (if no unit-field is present, assume same units as cfg.hdm).
Johanna - 2013-02-13 15:07:50 +0100
What do you mean by 1) 'specify' cfg.grid in the same units? Do you mean call ft_convertunits(cfg.grid,cfg.hdm.unit)? 2) why does this option not work? what aspect of it does not work? code crash? wrong results? did you try cfg.grid = ft_convertunits(cfg.grid) and did it either fail or give wrong units? Either way, I agree example script should be changed and add support to ft_prepare_sourcemodel for units. however, instead of: if no unit field, assume same as cfg.hdm, is there a way to determine it with ft_convertunits? See also related bugs to inconsistent units: bug 686 bug 1794
Jörn M. Horschig - 2013-02-13 15:29:19 +0100
also check why cfg.grid.tight does not do what it should
Jörn M. Horschig - 2013-02-13 15:34:07 +0100
(In reply to comment #1) 1) no, I meant that the actual numbers should be converted, e.g. cfg.grid.xgrid = -200:10:200 (so -200mm to +200mm) rather than -20:1:20 (which is what is currently done in the example script) 2) sorry, not working is indeed misleading here. Here it stands for: FieldTrip is currently disregaring the cfg.grid.unit field. Imho, determining the units is obviously impossible (the plain number 20 can be mm, cm, m, T, kg, $, etc., but I think deducing if they need to be converted from hdm is also not a very good idea. Better would be to output a warning, that the function assumes that units are in 'xx'. I am not quite sure if this is what you meant
Johanna - 2013-02-13 15:59:51 +0100
Ok, to clarify, if you load 'standard_grid3d8mm', then this grid has no units, but it is clearly 'cm'. Calling grid_out=ft_convert_units(grid) will correctly give grid_out the .unit field with 'cm'. However, grid_mm=ft_convert_units(grid,'mm') will give grid_mm a .unit field with 'mm' and a .pos in 'mm', but as you said, the .xgrid/.ygrid/.zgrid are not modified. So one action point is to make ft_convert_units change the values in .xgrid/.ygrid/.zgrid as well as .pos.
Jörn M. Horschig - 2013-02-13 16:07:56 +0100
hm, that's a bit tricky right? You can say it's clearly 'cm' because you know that we are working on a human brain. If it was a mouse brain, 'mm' might be more appropriate with those numbers - how should FieldTrip know? I wouls suggest to add the .unit field to the template grids, and let ft_convert_units only convert if there is a conversion between known units.
Johanna - 2013-02-13 16:31:32 +0100
(In reply to comment #5) It's tricky indeed, but I think Robert intends for it to be used to guess, due to all the calls to ft_estimate_units within ft_convert_units. It's not always right, and yes the mouse brain example would fail. But still I think forcing a call to ft_convert_units somewhere in the pipeline would be the way to go (and would match ways of doing it in other pieces of code). I agree .unit should be added to templates.
Marc Lalancette - 2014-11-06 00:54:50 +0100
To fix the "FIXME" at line 312, which is an issue with units (thus part of this bug), and which creates a 2-meter-wide grid when units are 'm', replace lines 315,318,321: grid.xgrid = floor(min(sens.chanpos(:,1))):cfg.grid.resolution:ceil(max(sens.chanpos(:,1))); %(similarly for y and zgrid) with: grid.xgrid = cfg.grid.resolution*(floor(min(sens.chanpos(:,1))/cfg.grid.resolution):ceil(max(sens.chanpos(:,1))/cfg.grid.resolution)); grid.ygrid = cfg.grid.resolution*(floor(min(sens.chanpos(:,2))/cfg.grid.resolution):ceil(max(sens.chanpos(:,2))/cfg.grid.resolution)); grid.zgrid = cfg.grid.resolution*(floor(min(sens.chanpos(:,3))/cfg.grid.resolution):ceil(max(sens.chanpos(:,3))/cfg.grid.resolution));