MeasureNeurons

Hi,

I’m trying to use the MeasureNeurons module. I’ve attached the images I have been working with. I’m not sure what I’m doing wrong. When I only have images as outputs (pipeline_that_works.cp), I get back images that I think are correct. Then when I check the “Do you want the neuron graph relationship?”, I keep getting this error message:

Traceback (most recent call last):
File “cellprofiler\pipeline.pyc”, line 309, in run
File “cellprofiler\modules\measureneurons.pyc”, line 350, in run
File “cellprofiler\modules\measureneurons.pyc”, line 675, in make_neuron_graph
File “cellprofiler\cpmath\cpmorphology.pyc”, line 3648, in pairwise_permutations
File “scipy\sparse\csc.pyc”, line 147, in getitem
File “scipy\sparse\csr.pyc”, line 252, in getitem
File “scipy\sparse\csr.pyc”, line 190, in check_bounds
IndexError: index (19) out of range

I’ve selected every possible image as the “intensity image” and I still get back the same error message. It made sense to me to use the InvertGreen (pipeline_that_crashes.cp). I’ve used nuclei (OrigBlue) and soma (OrigGreen) as the seed objects and I still get the same error. I’m not sure what to do. Did I make a crazy mistake somewhere else in the pipeline?

Thanks,

Yalls




pipeline_that_crashes.cp (10.8 KB)
pipeline_that_works.cp (10.8 KB)

Hi Yalls,

This looks like a bug. I haven’t tracked it down yet, and I can’t work around it either. DO some images run OK? Frankly, we really haven’t stressed the graph relationship code much yet, since we don’t have a tool yet to use the output.
I will report back here when a fix is in place, though you would need to install a new, special version, which I should be able to supply. What OS do you use? 32 or 64 bit?

As an aside, even the “pipeline_that_works” errors for me.
Also, it looks to me like you might benefit from “distinguish clumped objects” in IdentifyPrimaryObjects with Intensity, but that depends on if you really want the nuclei all declumped.

David

Hi David,

Thanks for getting back to me so quickly. I am currently running Windows 32-bit.

I just threw together the pipeline quickly to check out the MeasureNeurons module. I hadn’t moved on to optimizing the segmentation, because I got stuck with the error. I am actually in the process of testing out CP on a number of analyses that have already been outsourced. I’m trying to assess if CP can replace most of the analyses that have been purchased in the past… My goal is to convince the people I work with the move over to CP for future analyses.

Sorry, I think I posted the same pipeline twice. I’ve reattached another pipeline_that_works_28072011.cp. I just ran it again and know for sure that it works and I’ve also attached the images produced from it using the the input images from my previous post.

Yalls









pipeline_that_works_28072011_.cp (10.8 KB)

Hi Yalls,

Thanks for trying us out! Feel free to post further questions, comments, and requests on the forum. We do have plans to add features to the MeasureNeurons module, like Sholl analysis, a graph structure viewer/interpreter, better crossing point vs. branchpoint identification, etc.

For now, to be safe, you will need to uncheck the graph relationship button. It may work for some images, but under some conditions, it may fail (we’ll look into it soon). Again, this is a somewhat experimental feature, and the rest of the measurements (branchpoints, etc) should be just fine.

David

Well, our main programmer Lee has fixed this one quickly! You can download the latest, bleeding-edge version of CP here:
cellprofiler.org/cgi-bin/trunk_build.cgi
(version 11387 has the fix).

It has passed our tests, would you try it on your images?
Note that this version has many other updates, and that any pipelines you create with this version will not be back-compatible with the current website version.

Best,
David

Hi David,

Thanks again for being so quick about this. I will test it out in the next few days and let you know how it goes.

Yalls