Modelling soma as 3D vs a cylindrical section

Managing anatomically complex model cells with the CellBuilder. Importing morphometric data with NEURON's Import3D tool or Robert Cannon's CVAPP. Where to find detailed morphometric data.
Post Reply
Meena
Posts: 31
Joined: Mon Jan 09, 2006 3:50 pm
Contact:

Modelling soma as 3D vs a cylindrical section

Post by Meena »

Hi,

I have a question on modelling the cell as a 3D structure in Neuron
vs modelling it as cylindrical section.

I received a Neurolucida image of cell soma (in .asc format)
and converted this into Neuron.

I would like to assume the soma as a prolate spheroid as I think this
is a reasonable assumption.

However when I plug the 3D coordinates into Neuron - I find that
Neuron computes L and diam parameters. ( The L in Nueron is the length of the cell and diam
is the average diam across the length of the cell).

The value of L and diam here are much smaller and are misleading as they
produce a surface area that is much smaller.

To correct this I would need to use the major/minor axis values
from the prolate spheroid to compute the surface area and volume and use these
values to compute the accurate L and diam of a cylindrical section and then specify this in Neuron.

My question is :

1. If I need to give Neuron the values of the L and diam - then whats the objective of having the 3D points ?
I would like to model the soma as a 3D structure and not just a big cylindrical section?

2. So do I need to specify the 3D coordinates and the proper L and diam regardless of the 3D points to ensure
the right surface area and volume of cell soma are represented?

Can you please advice? I am a little confused about this and feel that my approach to modelling this soma
is not right.

Again, thanks so much for all your time!!!
Meena
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Modelling soma as 3D vs a cylindrical section

Post by ted »

Meena wrote:converted this into Neuron
How? Using the Import3D tool?
when I plug the 3D coordinates into Neuron
Meaning you do exactly what?
Neuron computes L and diam parameters. ( The L in Nueron is the length of the cell and diam
is the average diam across the length of the cell).

The value of L and diam here are much smaller and are misleading as they
produce a surface area that is much smaller.
They don't count. When pt3d data define the shape of a section, NEURON doesn't
compute section area from L and diam. What counts is the area that NEURON
computes straight from the pt3d data. To discover this value, execute the following:
soma nseg = 1
soma print area(0.5)

Please let me know if this resolves the problem for you.
Meena
Posts: 31
Joined: Mon Jan 09, 2006 3:50 pm
Contact:

Modelling soma as 3D vs a cylindrical section

Post by Meena »

Hi Ted,

<html>How? Using the Import3D tool? </html>

Yes, I used the Import 3D tool

<html>Meaning you do exactly what?</html>

The first thing I did was to look at the cell in Neurolucida Explorer and find the major and minor axis ( I am assuming the soma can be estimated using a prolate shperoid).
The major axis was : 14.7um
The minor axis was: 10.7um

So estimated as prolate spheroid:
The Surface area of the cell was: 1808 um2
The volume of the cell was : 7050 um3

However when I imported the cell from .asc file in Neurolucida to NEURON
and followed your instructions:

soma print area(0.5) was 440.38 ( I am assuming the units are um2)
and soma nseg was 1.

440.38 um2 is way smaller than my estimation of the cell as prolate spheroid.

Does this mean that the pt3d data from the Neurolucida tracing was incorrect ? Could I do something about it ?

I know you mentioned that the L and diam don't count.
But if I find the proper values of the L and diam of a cylinder that has
the same area and volume as the cell estimated as a prolate shperoid,
I get L and diam equals to 36.9 and 15.6 respectively.

And if the end of the pt3d data I specify L =36.9 and diam=15.6, then
do print soma area(0.5) equals to 1808um2...(which is the right area) ?
I assuming this is so bcs NEURON is treating the L and diam as the parameters of a cylindrical section.

What could the issue with my pt3d data be in your opinion?

Thanks so much for your time,
Meena






Please advice.
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Post by ted »

To answer your questions, it will be necessary for me to see the data. Please send the
Neurolucida file to me
ted .dot. carnevale .at. yale .dot. edu
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Modelling soma as 3D vs a cylindrical section

Post by ted »

Meena wrote:The first thing I did was to look at the cell in Neurolucida Explorer and find the major and minor axis ( I am assuming the soma can be estimated using a prolate shperoid).
The major axis was : 14.7um
The minor axis was: 10.7um

So estimated as prolate spheroid:
The Surface area of the cell was: 1808 um2
The volume of the cell was : 7050 um3

However when I imported the cell from .asc file in Neurolucida to NEURON
and followed your instructions:

soma print area(0.5) was 440.38 ( I am assuming the units are um2)
and soma nseg was 1.
The area of a prolate spheroid can't be larger than the area of a sphere whose diameter
is the same as the major axis. So it must be smaller than PI * d2 = PI * 14.72 ~ 680 um2.

Here is how the Import3D tool deals with tracings of the perimeter of the soma:
1. It finds the principal axis of the shape defined by the perimeter.
2. It divides the shape into 20 slices of equal thickness by cutting it with lines that are
perpendicular to the principal axis. These slices define a set of (x,y,z,diam) data that
are treated as pt3d data in the hoc file or CellBuilder that the Import3D tool exports.

The surface area of the soma is calculated by adding up the surface area of these 20
frustae. If nseg is 1,
soma print "soma area(0.5) is ", area(0.5)
will print the total area of the soma as calculated internally by NEURON. For the cell that
you are studying, this will print the following result:
soma area(0.5) is 440.38546

This result is easy to check--

Code: Select all

// distance from (xyz)1 to (xyz)2
// args are (xyz) of 1st point and (xyz) of 2nd point
func diag() {
  return sqrt(($1-$4)^2 + ($2-$5)^2 + ($3-$6)^2)
}
// distance from 3d point $1 to 3d point $1+1
// argument is pt3d datum index
func len() {
  return diag(x3d(i), y3d(i), z3d(i), x3d(i+1), y3d(i+1), z3d(i+1))
}
// area of curved surface of a right cone
// args are diam of base, diam of top, and height
func frustumarea() { local rbase, rtop, height
  rbase = $1/2
  rtop = $2/2
  height = $3
  return PI * (rbase + rtop) * sqrt(height^2 + (rbase - rtop)^2)
}
func my3darea() { local i, temp
  temp = 0
  for i = 0,n3d()-2 {
    temp += frustumarea(diam3d(i), diam3d(i+1), len(i))
  }
  return temp
}
soma print "my3darea function says ", secname(), " area is ", my3darea()
The output produced by this code is
my3darea function says soma area is 440.38544
FRB121
Posts: 2
Joined: Thu Dec 06, 2007 10:55 pm

Post by FRB121 »

We are having some difficulty getting Import3D to recognize the soma and approximate its shape as described above. When we import a cell with a single oval for the soma, it appears in NEURON as a single open oval. We tried to close the loop in Neurolucida by selecting "close contour" when ending the segment. It appears to be a closed loop in Neurolucida, but in Import3D the loop is open, with the connection between the last point and the first point missing. Is this what is preventing Import3D from dealing with the soma?
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Post by ted »

FRB121 wrote:It appears to be a closed loop in Neurolucida, but in Import3D the loop is open, with the connection between the last point and the first point missing. Is this what is preventing Import3D from dealing with the soma?
In order to answer your question correctly and develop a strategy for
dealing with this problem, we will have to be able to reproduce it. If
you zip up both versions of your Neurolucida files (please indicate
which is which) and send them to me
ted dot carnevale at yale dot edu
I'll see what we can do.
bhalterm
Posts: 52
Joined: Wed Mar 08, 2006 10:43 am
Location: University of Pennsylvania

Post by bhalterm »

Hi Ted, I'm working with FRB121. Our problem was that we thought that in Neurolucida the proximal dendrites needed to branch off of the ring defining the soma, i.e. originate at a node on the ring, for Import3D to recognize that the connection was there and not open multiple CellBuilders. Neurolucida would only let us do this if we didn't label the ring as a "Cell Body", instead just calling it "Dendrite" (apparently they are working to remedy this now). I misunderstood another post here, thinking that Import3D would look for a closed curve to make into the soma. After a stroke of good luck, we figured out that the "Cell Body" label in Neurolucida is used by Import3D. I'm under the impression now that for dendrites in the Neurolucida tracing to be connected to the soma in Import3D, they just have to stab into the 3D region that the soma would occupy if it were not just a 2D ring.

After a lot of fiddling around, we did get the cell to open in a single CellBuilder. However, there is a subset of Dendrites that are connected to the rest of the cell not at the soma, but way out at a distal dendrite. It is possible to see a spike start at the soma, propagate out through most of the dendrites, then travel along a single dendrite back toward the soma where it connects to the base of the orphaned subtree and propagates outward along that subtree again.

We've tried to fix this by extending the dendrites further into the soma in Neurolucida. It doesn't appear in Neurolucida that there are any wacky distal dendrite to dendrite connections. Here is the output when we load the file into Import3D:

Code: Select all

40893 lines read

./Desktop/FRB_model/soma5.asc problems

Ignored 1 spines

Main branch starting at line 2720 is outside the soma bounding boxes
  Making a logical connection to center of nearest soma

Main branch starting at line 2727 is outside the soma bounding boxes
  Making a logical connection to center of nearest soma

Main branch starting at line 16766 is outside the soma bounding boxes
  Making a logical connection to center of nearest soma

There are a few sections which seem to be added by Import3D that extend from the soma straight out to various distal dendrites. In the "Possible Root Branch Errors" window, checking the boxes to connect to the closest parent doesn't seem to make much of a difference. I am sending you a copy of our cell model so you can have a look.

Thanks a lot!
Post Reply