Hi SUMA gurus,

From time to time, our lab has trouble loading recently drawn .niml.roi files. Ziad, I think we've talked with you about this and attributed the cause to using different afni/suma versions between drawing and loading. I think I've found another issue which exists irregardless of the afni/suma versions. I have some .niml.roi files which give error messages even if I load the roi under the same afni/suma version that I drew them. Loading these files in SUMA will cause SUMA to crash and the following error messages to appear:

Fatal Signal 10 (SIGBUS) received
SUMA_OpenDrawnROI_NIML
SUMA_OpenDrawnROI
SUMA_FileSelection_file_select_cb
suma
Bottom of Debug Stack
** AFNI version = AFNI_2008_07_18_1710 Compile date = Nov 7 2008
** Program Abort **

When looking at the .niml.roi files that causes errors when loading, my last roi entry looks like this:
# <Node_ROI
# self_idcode = "XYZ_NFhckMJW8i7qlQUe0-Qx1g"
# domain_parent_idcode = "XYZ_j6FN2v0EtwvGyzfiG2ViIQ"
# Label = "33"
# iLabel = "33"
# Type = "0"
# ColPlaneName = "ROI.L.full.pat"
# FillColor = "0.690196 0.427451 0.788235"
# EdgeColor = "0.000000 0.000000 1.000000"
# EdgeThickness = "2"
# />


The last roi entry in a file that loads correctly looks like this:

# <Node_ROI
# ni_type = "SUMA_NIML_ROI_DATUM"
# ni_dimen = "4"
# self_idcode = "XYZ_tejjpoZyFCEylAS3Ug945w"
# domain_parent_idcode = "XYZ_j6FN2v0EtwvGyzfiG2ViIQ"
# Label = "33"
# iLabel = "33"
# Type = "0"
# ColPlaneName = "ROI.L.full.pat"
# FillColor = "0.690196 0.427451 0.788235"
# EdgeColor = "0.000000 0.000000 1.000000"
# EdgeThickness = "2"
# >
1 4 1 54568
1 4 7 54568 54567 54585 54584 54602 54601 54600
1 4 5 54600 54599 54598 53493 53492
1 4 3 53492 53472 52334
# </Node_ROI>


I see two major differences:

1. the correct files have a list of what appears to be node or vertice intersections
2. the correct files end with # </Node_ROI> as opposed to the problematic files which end with # />.

If I edit out the last roi entry for the problematic roi files, they will load fine in SUMA. Further, it appears as if all of my ROIs are accounted for, meaning the last (problematic) roi in the problematic roi file does not appear to correspond to my last drawn roi (though it has the same header values).

I'm not sure why this happens, but it is quite random and I've experienced this in multiple versions of afni / suma. The times I experience these errors (typically) are when I have numerous rois (30-40) within one .niml.roi file. I'm not sure if I'm hitting some size limit, but this seems doubtful since all of my rois are accounted for in the problematic files after deleting the last entry.

It seems possible that on occasion when drawing rois, I start a "last" roi, then decide to not use it, and undo or delete the roi (never having clicked finish). Could there be some partial write out of that roi, even though it was never finished?

If this is the cause, is there any way to fix this in the .niml.roi writeout? Or is there anyway to load the rois (either in SUMA or ROI2dataset), in such a way that it can ignore these problematic entries?

Thanks,
Mike