Hello,

Using V3D (2.485), V3D-Neuron (2.0), I can trace neurons, save and load swc file.
But if I load the swc file (obtained with V3D)in Neuronstudio (http://research.mssm.edu/cnic/tools.html), the structure appears out of scale.
If I load a swc file (obtained with Neuronstudio) in V3D, it appears out of scale.

I tried to set the x, y, z in V3D with :
Image/Data > Geometry > Image resampling
there for each axis (x, y, z) I choose "current pixel 1.0" and "target pixel n" ; n being the real size in µm of my voxels.
This however does not help

How can I scale my images
-so the infos I would extract from the swc file (legnth and so on) are properly scaled in µm
-so I load them in Neuronstudio to detect spines

I thank you very much, V3D is very powerful and user interface very well done
Nicolas Heck
Paris, France

Hi Nico,

V3D traces a neuron (or any 3d curve for quantitative measurement) based on the scale of the actual voxel (pixel) in an image, and thus the result was presented (saved or displayed) in the pixel/voxel space. I did not use the physical metric unit, say, micron, to save the traced structure.

Of course any one can easily multiple a factor (e.g. resolution of an image, or pixel size) to the coordinates of the reconstruction nodes' in an SWC file (as SWC is a simple text file). That is a postprocessing, and can be done very easily, say in Matlab. V3D comes with a Matlab file IO toolbox to read/write the SWC file (and other files used by V3D). Thus quantitative measurement is not a concern.

Neurolucida, on the other hand, saves the file in the "physical" space scale. That explains the scale problem. Also it seems it also have some unknown displacement which I don't understand. Of course, sometimes one has to first convert the neurolucida format to SWC.

There are several ways in V3D for this problem. A simple method is to go to the 3D viewer, right click the SWC object you want to manipulate, and select "Lock scene and adjust this object" from the pop-up menu. Then you can shift, zoom, rescale, and rescale the radius of a neuron (or a 3d curve). In this way, you can manually adjust two neuron to make them similar and compare. Of course, you can save the neuron structures after the transformation, and next time you can directly see them. You can right-click a structure and find the "save to file" menu item from the pop-up menu.

We also developed an automatic method to find the correspondence of any two neurons directly and thus transform them right away. We did that through a V3D plugin.

Hope this helps.

Best,
-hanchuan








Originally posted by nicolas heck: