Abcam Ad

Author Topic: manual counting of rat hippocampal cells  (Read 24405 times)

0 Members and 1 Guest are viewing this topic.

Offline formalanne

  • Trainee
  • **
  • Posts: 5
manual counting of rat hippocampal cells
« on: February 08, 2007, 04:32:06 PM »
I am currently working on a project which requires rat hippocampal cells to be counted after the section is stained with cresyl violet. Since we do not have any analysis software, the only option is to do this manually. I know it is a silly question, but does anyone have any suggestions for how to go about this? currently, we project the image to a television and count the densest area. Has anyone had any experience with this??? ???

manual counting of rat hippocampal cells
« on: February 08, 2007, 04:32:06 PM »

Offline ImmunoNYC

  • Global Moderator
  • GoldMember
  • *****
  • Posts: 684
Re: manual counting of rat hippocampal cells
« Reply #1 on: February 10, 2007, 12:34:04 AM »

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #2 on: February 21, 2007, 08:41:10 AM »
You have a problem here that can be accomplished, but is going to be a little more complicated than you had probably hoped. Counting cells is not as simple as you might have hoped. It turns out that a good solution to this problem was not worked out until 1984. Many people had spent lots of time on this issue and it took years to solve.

I think that a reason that it was difficult to solve is that counting seems so simple to us. We are taught from early on how to count and we do it in such a repetitive manner that all of us feel that counting is second nature. That might be a good assumption with a pile of coins, but is not true under the microscope.

I called the DC Sterio solution of the disector a good solution. By that I mean a solution that gets the right answer.

If you would like some references I can post them for you.

Do more, less well. ~Ewald Weibel

Offline formalanne

  • Trainee
  • **
  • Posts: 5
Re: manual counting of rat hippocampal cells
« Reply #3 on: February 24, 2007, 09:25:22 PM »
I would really appreciate references! thank you!

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #4 on: February 25, 2007, 10:17:02 AM »
A simple place to start is an online resource:

http://nervenet.org/papers/3DCounting.html

This is a good place that discusses an important paper in the history of counting. Here Abercrombie shows how to use either 1 section or a pair of sections to get a count of cells, not profiles. The problem is that what is needed to correctly implement the protocol cannot be done in practice.

A good review of the history can be found in Hedreen.

Hedreen, Centenary of Gaule and Lewin, J of Micr, 1997

Stepping forward in time from Abercrombie there are a number of proposals. Then comes DC Sterio.

Sterio, Unbiased estimation of number and sizes of ..., J of Micr, 1984

This is the solution to the counting profiles vs counting cells problem.

A text available in many libraries is:

Howard and Reed, Unbiased Stereology

The book is reasonable to read.

Through all of this it is important to understand that what is seen through the microscope is not the same as
what was in the original object. There are many cases of misinterpretation of biological objects in a qualitative sense let alone a quantitative sense. The most famous of these seems to be the 99 year history of the structure of the human liver. There are other cases as well. Structure of invertebrate intestines, chicken gut, mitochondria, ER, and so forth. These are examples of trying to figure out just the shape of an object by looking at tissue slices. Getting the numbers right is sometimes trickier.

It seems reasonable to look at tissue in a slice and count the profiles and connect them to the cells from whch the profiles came. It turns out not to be that way.

Take a look for instance at:

Andersen, Reduction of Purkinje cells ..., Brain Research, 2004

Here is a case where counting on single sections was tested against proper stereological methods. The conclusion once again is that counting profiles provides the wrong answer. The article also provides references to other places where mistakes due to profile counting are made.

Good luck. If for any reason you want to take anything offline just use the private message feature.
Do more, less well. ~Ewald Weibel

Offline ImmunoNYC

  • Global Moderator
  • GoldMember
  • *****
  • Posts: 684
Re: manual counting of rat hippocampal cells
« Reply #5 on: February 28, 2007, 12:37:56 AM »
Because this is such an important thread I merged the two similar threads to keep it active. Here are the posts from the other thread:

http://www.ihcworld.com/smf/index.php/topic,1892.0.html

formalanne
Analysis? (Help!)
on: January 23, 2007, 02:44:49 PM

Does anyone have any information/ advice about morphometric analysis of neurons stained with cresyl violet?

richard03
Re: Analysis? (Help!)
Reply #1 on: January 24, 2007, 08:11:55 PM

After staining with cresyl violet, you can use an image analysis software to do morphometric analysis (neuron size, density, total number, etc).

Use this page to find a suitable software.

http://www.ihcworld.com/imageanalysis.htm

formalanne
Re: Analysis? (Help!)
Reply #2 on: February 08, 2007, 03:33:26 PM

thank you! but unfortunately the software is not available to us, so we have to do it manually, do you know anything about how to count effectively without the software?

ImmunoNYC
Re: Analysis? (Help!)
Reply #3 on: February 08, 2007, 09:39:18 PM

You can take photos and print them and just count by marking with "x" the ones you have already counted. But that, in my opinion, is archaic and painfully time consuming. So why not download free software:

http://rsb.info.nih.gov/ij/
http://ddsdx.uthscsa.edu/dig/itdesc.html

You can also use programs such as Adobe Photoshop to quantitate purple pixels and use this as an "index" of positive cells.

formalanne
Re: Analysis? (Help!)
Reply #4 on: February 08, 2007, 10:35:08 PM

well, I certainly feel like I have crawled out of the dark ages...thank you so much

hokie
Re: Analysis? (Help!)
Reply #5 on: February 21, 2007, 05:35:56 AM

There is a serious problem here --

This does NOT obtain the correct answer.

The problem is that the things you see in a section are profiles, not cells. What you want to do is to count cells and counting profiles from a single image cannot work. That is a well known problem. Solutions have been posted since at least the 1890s. A solution that works and can be implemented in the lab was published in 1984 under the pseudonym DC Sterio.

Look at any of a number of books before the 1984 solution is published to see good discussions about the problem. Books by DeHoff, Elias, and Weibel all go into lengthy discussions about the problems of counting cells. Post 1984 books all push proper counting methods.

The use of image analysis packages to count pixels or to count segmented profiles speeds up the process of getting the wrong answer faster. The number of pixels counted in a section is an indicator of the total volume of the cells, not the number of cells. The mean area of a profile is not related to the mean volume of a cell.

ImmunoNYC
Re: Analysis? (Help!)
Reply #6 on: February 21, 2007, 09:26:25 PM

Interesting points, however if she counts absolute number of events purely without counting pixels or area ina  2D image then doesn't the disector principle not apply here? Also, if she is counting cultured cells in a monolayer then the disector principle also does not apply.

hokie
Re: Analysis? (Help!)
Reply #7 on: February 22, 2007, 04:12:11 AM

This is not a question of whether or not the disector principle applies. It is a question of whether or not the disector principle should be applied. Sorry for my silly semantics here.

The question really is about "counts absolute number of events purely without counting pixels or area in a  2D image". If the events can only occur in a single section, then sure a section is sufficient. Counting profiles is counting an event. Unfortunately cells may not appear in a single section. The event in this case is a particular cell appearing in a section. Counting nucleoli is counting an event. The chance of a nucleoli appearing in multiple sections is lower since this object is smaller than th cell. This works as long as each cell has a single nucleoli. The bias is reduced, but not eliminated. The event being counted has to be 0-dimensional.

Remember that cameras are sampling devices. The smallest object that is visible is a pixel in size. The best that can be done with an image is to count pixels. Objects that are larger than pixels run into the problem that they need to be counted in an unbiased manner. That requires a counting frame.

Image analysis packages often use a technique in which objects on the edge of the screen are counted if they touch 2 of the edges and not counted if they touch the other 2 edges. This sounds reasonable, but leads to a bias.

If "counting cultured cells in a monolayer" then sure the disector principle canbe avoided. Why? Because nothing is sectioned and objects being couonted are not being split. But, a counting frame is required. The only way around this is to count everything. Counting everything is the same as placing a very large counting frame around the entire culture.

I happened to read an interesting article last night in Brain Research about alcoholics and neuron loss. According to the article the claim that neurons are lost is a mistake due to counting in single sections. Proper stereological counting showed that there is no loss of neurons. The changes seen included changes in compartment volume and the volume of neurons. These factors do not affect correct counting. These factors do affect counting profiles.


Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #6 on: February 28, 2007, 04:01:08 PM »
I am glad that the imortance of this thread was realized by the moderator.

Last year I was at the Neuroscience meeting and was surprised at the change in posters. It used to be that most of the posters were profile counting. After a number of journals began to require unbiased methods there was a dramatic swing to using unbiased stereological methods. This last year it was easier to locate profile counting posters than it was to find posters doing counting in a correct manner.

I think the problem is that the difficulties in counting are not apparent, nor are the solutions. It seems so reasonable to count profiles that few people do as was done here - ask. I find that most people are skeptical when I talk to them about profile counting. I hope that an interesting dialog can materialize so that all of us can learn more about counting, its perils, and what is it that makes profile counting seem so reasonable.

My hat off to the moderator
Do more, less well. ~Ewald Weibel

Offline rmcgandara

  • HighTech
  • ***
  • Posts: 12
Re: manual counting of rat hippocampal cells
« Reply #7 on: March 06, 2007, 09:30:37 AM »
Thanks for the discussion. this raised some questions in my work... I am counting cell profiles in the crypts of the gut :D

gave more to think about...

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #8 on: March 06, 2007, 09:22:55 PM »
If you have any specific questions please ask. I know that the gut of birds was an interesting area of discussion years ago. There are structures in the gut of the bird that were identified as tubular glands. It turns out that these were not tubular at all but slit-like glands. I believe the proper term for this is a sulciform structure.

Here is another example of how the qualitative information on a histological section was misinterpreted. It should not be a surprise at all that the quantitative information on histological sections can also be misinterpreted.

Do more, less well. ~Ewald Weibel

Offline Mr. Gunn

  • Trainee
  • **
  • Posts: 9
Re: manual counting of rat hippocampal cells
« Reply #9 on: May 16, 2007, 01:38:22 PM »
If you have any specific questions please ask. on histological sections can also be misinterpreted.

What about counting osteoclasts in bone sections?  We have a stereology setup, but it should be sufficient to count absolute events per some fixed area, right?

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #10 on: May 16, 2007, 05:21:11 PM »
The short answer is no, but that it all depends on what you want to know.

The number of osteoclasts per unit area is not a measure of the number of osteoclasts in a bone. What is seen in the section is a profile and this is a measure of the number of osteoclast profiles per unit area. Depending on the shape and orientation of osteoclasts it is likely that the number of osteoclasts per unit area will change depending on the angle at which the bone is sectioned.

Depending on what you need to know it may turn out that osteoclasts per unit area is a useful indicator of some property of the bone. It won't be an indicator of the number of osteoclasts, but it may be useful for some other property such as the strength of the bone. It all depends on how the population of osteoclasts is organized.

Bone I realize poses its own problems. The matrix is hard and it is difficult to cut. Much of the material I have read has dealt with trabecular bone. Research on resorption surface area was done by a group back in the late 1980s. It turns out that it is easier to determine the surface area where resorption occurs than it is to determine the number of osteoclasts involved.

The ability to determine one aspect of the problem and not another is due to the limitations imposed by the math  and the lab work. The math requirements to obtain a solution are balanced with the ability to manifet those math requirements in the lab.
Do more, less well. ~Ewald Weibel

Offline Mr. Gunn

  • Trainee
  • **
  • Posts: 9
Re: manual counting of rat hippocampal cells
« Reply #11 on: May 16, 2007, 08:13:33 PM »
Research on resorption surface area was done by a group back in the late 1980s. It turns out that it is easier to determine the surface area where resorption occurs than it is to determine the number of osteoclasts involved.

Interesting.  Would you happen to have a reference for this method?

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #12 on: May 17, 2007, 04:55:52 AM »
Look in the Journal Bone under the name Vesterby. You should find a number of interesting articles including one that has vertical in the title. The word surface might also be in the title.
Do more, less well. ~Ewald Weibel

Offline Mr. Gunn

  • Trainee
  • **
  • Posts: 9
Re: manual counting of rat hippocampal cells
« Reply #13 on: May 17, 2007, 12:08:13 PM »
I see those.  Thanks!

Offline hokie

  • SuperTech
  • ****
  • Posts: 43
    • Stereothena Inc
Re: manual counting of rat hippocampal cells
« Reply #14 on: May 17, 2007, 01:39:35 PM »
The reason that surface area is easier to do is that surface area can be determined from slices through the bone. Getting paired slices for cell counts in bone would be hard to do.

Vesterby did a lot of important work in bone. She was one of the first researchers doing modern stereological procedures in bone. And she developed things in bone such as star volume and promoted the usefulness of the concept of star volume.

Another name to check out is Boyce. Boyce's work on a method called the conneulor is another innovative method that saw its early use in the analysis of bone.

If there are any questions about the paper or you need some references for background material please ask.
Do more, less well. ~Ewald Weibel

Re: manual counting of rat hippocampal cells
« Reply #14 on: May 17, 2007, 01:39:35 PM »