Re: Submitting autoseg jobs on SLASH portal

[Rick Giuly]

Hi Alex,

I'd recommend marking them in IMOD, partial or not. That's what I always
do when testing. You do end up with contours that stray off of the
volume of interest a little bit, but that's not a big issue.

Hope that helps
-Rick

On 07/12/2012 02:41 PM, Alex Perez wrote:
> Hey Rick,
>
> When making training data, what do you do in the case of a mito that
> is cut off at the boundary of the volume? Do you use a closed contour
> and close it together along the boundary of the volume? Or use an open
> contour and don't close it along the boundary?
>
> I guess it's probably best to avoid these, but it's hard to get a
> subvolume that doesn't have any partially cut off mitos.
>
> Thanks,
> Alex

[Rick Giuly]

OK, I see what you're asking now. You'll need a closed contour. I
usually end up making the contour extend beyond the volume of interest a
bit in IMOD - it's definitely closed.