Thanks all to their comments on this!
reducepatch indeed works nicely.
of course it cannot preserve all the structure, i.e. the gyration
pattern is a bit harder to follow when downsamplin the 8000 vertices
grid by a factor of 5 - but I think it is detailed enough to capture the
relevant spatial structure from MEG source 'reconstructions' and may
help with MCP and reduce computational costs for certain analyses
best regards
Markus
> ---------- Forwarded message ----------
> From:[log in to unmask] <[log in to unmask]>
> Date: Sat, May 29, 2010 at 9:57 PM
> Subject: Re: [SPM] downsampling cortical MNI grid
> To: Vladimir Litvak<[log in to unmask]>
>
>
> Dear Markus,
>
> It could be that it's not what you want, but did you try matlab's
> reducepatch function? As far as I know it aims to reduce the number of
> triangles in a patch while trying to keep the overall shape intact. I
> am not sure whether this gives a closed surface though, but I remember
> there were some issues there when I tried to play with it in the past.
> Yet, as long as you don't aim to do a surface based normalisation
> between subjects (requiring the closed surface) I think it's worth a
> try.
>
> Cheers,
>
> JM
>
> ----- Original Message -----
> From: "Vladimir Litvak"<[log in to unmask]>
> To:[log in to unmask]
> Sent: Saturday, May 29, 2010 6:01:48 PM GMT +01:00 Amsterdam / Berlin
> / Bern / Rome / Stockholm / Vienna
> Subject: Re: [SPM] downsampling cortical MNI grid
>
> Hi Markus,
>
> I'm not familiar with a simple and principled way to downsample the
> cortical mesh, at least not something in SPM. Those meshes were
> prepared originally with freesurfer and there might be a way to
> downsample them there or just prepare lower resolution meshes from
> scratch.
>
> Rik Henson and Daniel Wakeman should be able to advise you.
>
> Best,
>
> Vladimir
>
> On Fri, May 28, 2010 at 7:32 PM, Markus Bauer<[log in to unmask]> wrote:
>
>> Hi Vladimir
>>
>> I wonder whether it would be easily possible to (further) downsample an
>> existing MNI grid
>> the coarsest grid possible in SPM seems to be ~5000 points whcih is still a
>> lot, possibly a higher resolution that one might need (and leading to an
>> enhanced MCP)
>> I realized that when subsampling the grid (taking e.g. only every 4th vertex
>> from the usual 8000 grid) leaves the distance relations approximately
>> preserved (i.e. the grid is still somewhat regular). So this could be a very
>> rough way of doing it - and for display purposes extrapolating again
>>
>> But would it be possible to either make a coarser grid from the beginning -
>> or do you know how I can get new 'faces' - ie the connection surfaces of the
>> subsampled grid?
>> Any idea on this?
>>
>> thanks
>> Markus
>>
>>
>>
>>
>
|