Hi everyone,

I am interested to know how many LSFM user do we have here, and what was your experience using automatic tracing in Vaa3D.

I am new to neuron tracing, and am wondering how many of you have successfully applied automatic tracing plug-ins, and how much pre-processing did you have to do, in order to achieve it?

Cheers,
Vytas

Hi Vytas

I have just started working with LSFM data myself. I have been using a number of plugins but have had some trouble with getting APP2 to run correctly. When it has run it doesn't work well but I suspect some adjustment of the threshold parameters may be required. Probably the plugin I've had the most success with is neutube but that still isn't very impressive yet. What about you?

Cheers
James

I think the "doesn't work well" case should be described in detail so others could help when the specific information is provided.

-Hanchuan

Originally posted by James Wilsenach:

Hi Hanchuan

Thanks for your interest, I am running the CentOS v3.458 version of Vaa3D on a Fedora machine. My problems with APP2 has been that the method under default or auto thresholding parameter settings seems to produce either one huge node in the former case or an overly thick arbor with some incorrect branching and a number of thin, spurious branches that all seem to line-up in parallel (screenshot of swc and max Z projection attached). Perhaps this could be improved by increasing the threshold further? Adding markers has not proven successful yet and has lead to arbors that ignore branches entirely and only go through marker points. I am trying to avoid manual curation however so this is not too concerning. I am not expecting perfect results from APP2 with my test image however as the soma of the single cell I am imaging is just out of field.

For other plugins  I often received plugin failed to run or other messages. For instance, when attempting to run TreMap I get:

open file [/..._tmp/group_one.raw_group_1.swc] failed!

Or when using one of the distance field methods I get for instance:

The plugin fails to run [Other method 2: dfs]. Check your plugin code please.

Or running Rivulet 2:

Making Gradients of Time Crossing Map...The plugin fails to run [tracing]. Check your plugin code please.

The plugins appear to be intact and available (github, pre-compiled version). Per another post on the site I had to re-name the libstdc++.so.6 file to default to the system version. Could other incompatibilities like this be the source?

http://nitrc1.rssing.com/browser.php?ind...

Here is a link to the input tif file:

https://drive.google.com/drive/folders/1...

Thanks in advance for any help!

Kind regards
James

I have since found out that there was a memory limiting issue in that 32GB of RAM was insufficient to perform distance transforms on a 1500*1500*600 pixel image. After re-running results have much improved and APP2 seems to be the best choice (most detailed reconstruction) for my LSFM data.