Help: colony counting with image J

[Alan Thames]

Dear imageJ users,
                                   I have an issue with cell colonies
counting
using image J for my drug treated cells in 3 D soft agar. My control( non
treated)  images
have big colonies in which cells are fused and in the drug treated images,
 sizes of the colonies are reduced (with increasing dosage of drug) . As
the size of colonies between no treatment and drug treatment is much
different, i cannot get a proportionate decrease in colony numbers as
i can see with average colony "size". For example; in no treatment picture
- colony number may be less but it's very obvious there r huge colonies,
using the same setting for treatment given pics - colony count may be
increased due to tiny small colonies though big colonies are not present.
It makes me confused that the drug doesn't reduce colony count, but the
size.
Can any one help me how to get a setting that could count numbers
proportionate to treatment efficacy and at the same time it would detect
the decrease in size?
                                      If i increase the size setting  just
to count big colonies, the small ones are all missed. If i decrease the
size to include all colonies including tiny ones, the count is still
increased in cells treated with higher drug concentrations, due to the
presence of
tiny  smaller colonies, even the big ones are not seen.


My steps are imageJ --> Type--> change the image to 8 bit format
                                    -->adjust-->  threshold--> set ; lower
threshold - 0, upper threshold- 155
                                   --> Analyze --> analyze particles
-->size-->25 to infinity, cicularity 0.25 to 1, show outlines, Check
-display results, clear results, summarize, exclude edges

Regards,
Alan

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Gabriel Landini]

On Saturday 21 Sep 2013 00:24:00 Alan Thames wrote:
> Can any one help me how to get a setting that could count numbers
> proportionate to treatment efficacy and at the same time it would detect
> the decrease in size?

Without knowing anything further, I think that you should not change any
settings to fit your a priory experimental groups.

>                                       If i increase the size setting  just
> to count big colonies, the small ones are all missed.

Why? Do not use any size setting, analyze everything there is.
If that leads to false counts, then the segmentation step is incorrect.

> If i decrease the
> size to include all colonies including tiny ones, the count is still
> increased in cells treated with higher drug concentrations, due to the
> presence of tiny  smaller colonies, even the big ones are not seen.

You probably need to look at your experimental design from a different
perspective. There might be a relation to treatment in the number of colonies
and their sizes depending on treatment, given that they all need to share the
same plate size. However modifying the quantification step according to
treatment introduces a bias in your sampling that is not scientifically sound:
*you* are introducing a change in the counts.

Cheers

Gabriel

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Alan Thames]

Hi Gabriel, Thanks for the reply.


"> Can any one help me how to get a setting that could count numbers
> proportionate to treatment efficacy and at the same time it would detect
> the decrease in size?

Without knowing anything further, I think that you should not change any
settings to fit your a priory experimental groups."

--->I keep one setting to quantify all.

"
>                                       If i increase the size setting  just
> to count big colonies, the small ones are all missed.

Why? Do not use any size setting, analyze everything there is.
If that leads to false counts, then the segmentation step is incorrect."

I double checked now, without changing any thing, i.e - 8 bit image -
threshold - auto; analyze0 analyze particles --- count was way increased to
thousands (not possible).


"You probably need to look at your experimental design from a different
perspective. There might be a relation to treatment in the number of
colonies
and their sizes depending on treatment, given that they all need to share
the
same plate size. However modifying the quantification step according to
treatment introduces a bias in your sampling that is not scientifically
sound:
*you* are introducing a change in the counts."

You are absolutely right. The treatment, i believe should decrease both
size and number of the colonies.

I will send a few images to you email, if u hv some time to spare.
Thanks again

Alan



On Sat, Sep 21, 2013 at 6:57 AM, Gabriel Landini <[hidden email]>wrote:

> On Saturday 21 Sep 2013 00:24:00 Alan Thames wrote:
> > Can any one help me how to get a setting that could count numbers
> > proportionate to treatment efficacy and at the same time it would detect
> > the decrease in size?
>
> Without knowing anything further, I think that you should not change any
> settings to fit your a priory experimental groups.
>
> >                                       If i increase the size setting
>  just
> > to count big colonies, the small ones are all missed.
>
> Why? Do not use any size setting, analyze everything there is.
> If that leads to false counts, then the segmentation step is incorrect.
>
> > If i decrease the
> > size to include all colonies including tiny ones, the count is still
> > increased in cells treated with higher drug concentrations, due to the
> > presence of tiny  smaller colonies, even the big ones are not seen.
>
> You probably need to look at your experimental design from a different
> perspective. There might be a relation to treatment in the number of
> colonies
> and their sizes depending on treatment, given that they all need to share
> the
> same plate size. However modifying the quantification step according to
> treatment introduces a bias in your sampling that is not scientifically
> sound:
> *you* are introducing a change in the counts.
>
> Cheers
>
> Gabriel
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html