Re: Counting osteoclasts in culture

[Artsi]

sorry the image...

> Hi all,
> 1)I'm trying to use imagej to count osteoclasts in culture, picture
> attached (low resolution) each purple stain is a cell. I need to count the
> area% occupied and number. I'm new to this but i thought it was possible (in
> the high resolution image the boundaries are better visible). I tried color
> segmentation but I did not manage to separate the cells, I tried converting
> to gray and use various options suggested to others, I tried the fiji
> trainable segmentation but it simply got stuck.
> 2)A different matter entirely, how can i save the masks from the roi
> manager.
> Please help me...
> Hanna
>
> --
> *Curiosity Killed The Cat Satisfaction Brought It Back*
>
>

--
*Curiosity Killed The Cat Satisfaction Brought It Back*

[Rob van 't Hof]

Hi Hanna,
I've analysed these kind of cultures for the past 20 years, and have
quite a bit of experience in image analysis. In my experience, it is
much faster to manually count. The shape of the cells is irregular, they
are usually found in dense clusters, and they often have parts connected
by thin bits of cytoplasm. Oh, and you' have to distinguish between
mononuclear and multinucleated cells (and authentic osteoclast is
multinuclear). This all makes automated segmentation unreliable, and
you'd have to spend a lot of time correcting the errors. However, if
anyone does come up with a robust algorithm i'd be very happy. Something
that is very easily done is calculate TRAP stained area with a simple
colour threshold, and this may be sufficient for your needs.

bye,
Rob

On 21/09/2011 14:45, Hanna Artsi wrote:

> sorry the image...
>
>> Hi all,
>> 1)I'm trying to use imagej to count osteoclasts in culture, picture
>> attached (low resolution) each purple stain is a cell. I need to count the
>> area% occupied and number. I'm new to this but i thought it was possible (in
>> the high resolution image the boundaries are better visible). I tried color
>> segmentation but I did not manage to separate the cells, I tried converting
>> to gray and use various options suggested to others, I tried the fiji
>> trainable segmentation but it simply got stuck.
>> 2)A different matter entirely, how can i save the masks from the roi
>> manager.
>> Please help me...
>> Hanna
>>
>> --
>> *Curiosity Killed The Cat Satisfaction Brought It Back*
>>
>>
>

--
_____________________________
Dr. Rob van 't Hof
Senior Lecturer
Bone Group
Molecular Medicine Centre
Western General Hospital
Edinburgh EH4 2XU
UK

Phone: [phone]
email: [hidden email]
_____________________________



The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.