After you import morphology data, always examine a Shape plot of the resulting model cell for possible problems.
1. Z axis artifact. This appears as sudden jumps parallel to the Z axis. Morphology files with Z axis artifact are useless for modeling. The Shape plot's default view is a projection of the cell onto the xy plane. To rotate around the y axis, click on the Shape plot's menu box (left upper corner of the shape) and scroll down the menu to select 3D Rotate. Release the mouse button, move cursor to the middle of the Shape plot window, and press the mouse button. Hold the mouse very still and press the X key on the keyboard. Now release the mouse button and you'll see a side view of the model cell (projection onto the zy axis).
Return to the standard view by pressing the Z key.
2. Sectioning artifact. This typically happens when a cell is so close to the surface of a brain slice that part of it has been sliced off. The cut ends of the damaged branches will lie in a plane, so if you look at the model cell from a particular angle, it will look like a tree that has been sliced by a giant wielding a chainsaw. To see this, make sure that the Shape plot is in 3D Rotate mode, then click on the canvas off to the side of the cell and drag the cursor around. Release the mouse button to see the cell from a new angle. If part of the cell moves outside of the Shape plot's frame, use the plot's View = plot to recenter it (click on the menu box, move the cursor up a bit to expose the graph's secondary menu, and drag the cursor right to select View = plot).
Return to the standard projection by pressing the Z key. Exit 3D Rotate by clicking on Section in the Shape plot's primary menu.
3. Bad soma. Different labs use different approaches to defining the shape of the soma. Some users trace out one or more outlines of the soma in different planes of focus; this is often seen in Neurolucida files, and it results in the most realistic appearance of the model's soma. Some users just do a series of x,y,z,diam measurements along the soma's centroid; this approximates the soma as a chain of frusta, and it's what swc users have to do, because unlike the Neurolucida ASCII file format, the swc file format has no notion of a "soma outline." If an swc file tries to define the soma with an outline, the soma will end up being a long, very thin section that curls around--nothing like the soma in the real cell. To check the shape of the model cell's soma, put the Shape plot in Show Diam mode--click on the click on the Shape plot's menu box and scroll down the menu to Shape Style, then over to select Show Diam. Now zoom in on the soma--click on the menu box, hold the mouse button down while moving the cursor up a bit to expose the graph's secondary menu, then drag the cursor right to the secondary menu, then down to Zoom in/out. Release the mouse button and click on the Shape plot just to the right of where the soma should be located. Zoom in by dragging the cursor to the right. Do this as many times as needed to see the soma clearly. When done, restore the Shape plot to Centroid mode (in the menu, go to Shape Style, then drag right to select Centroid).
In some Neurolucida files, the 3-dimensional shape of the soma is defined by a stack of traces around the soma at different z axis levels (imagine a stack of pancakes). This produces a very nice appearance in graphs, and can result in the most accurate values for soma surface area and volume. However, serious errors will occur if the contours are not ordered monotonically according to depth--see Soma stacks and the Import3d tool viewtopic.php?f=28&t=3833 in the Hot tips area of the forum.
4. Orphan sections. An orphan section is a child section that has not been attached to its parent section. NEURON's Import3d tool tries to fix this automatically by attaching the 0 end of the orphan to the nearest 0 or 1 end of a potential parent. When it does this, it also reports what it did in a popup window or by printing a message in NEURON's interpreter terminal. If this happens, you should check the resulting model cell to be sure that it "looks right." If it doesn't, you have to decide whether to try to fix it or to throw it away and get a different morphology.
Lots of other things can go wrong with morphometric data; you might want to read
Scorcioni et al.
Quantitative morphometry of hippocampal pyramidal cells: differences between anatomical classes and reconstructing laboratories.
Journal of Comparative Neurology 473:177-193, 2004.
Kaspirzhny et al.
Neuronal morphology data bases: morphological noise and assesment of data quality.
Network 13:357-380, 2002.
A collection of noteworthy items selected by our moderators from discussions about making and using models with NEURON.
1 post • Page 1 of 1